Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Local strength and regional bone mineral density profiles of the thoracolumbar endplate Bailey, Christopher Stewart 2003

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_2003-0418.pdf [ 6.58MB ]
Metadata
JSON: 831-1.0091010.json
JSON-LD: 831-1.0091010-ld.json
RDF/XML (Pretty): 831-1.0091010-rdf.xml
RDF/JSON: 831-1.0091010-rdf.json
Turtle: 831-1.0091010-turtle.txt
N-Triples: 831-1.0091010-rdf-ntriples.txt
Original Record: 831-1.0091010-source.json
Full Text
831-1.0091010-fulltext.txt
Citation
831-1.0091010.ris

Full Text

L O C A L STRENGTH AND REGIONAL BONE MINERAL DENSITY PROFILES OF THE THORACOLUMBAR ENDPLATE by CHRISTOPHER STEWART BAILEY MD, BPE, McMaster University, 1996 FRCSC University of Western Ontario, 2001 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTERS OF SCIENCE in THE FACULTY OF GRADUATE STUDIES DEPARTMENT OF SURGERY We accept this thesis as conforming to the required standard  THE UNIVERSITY OF BRITISH COLUMBIA July 2003 © Chris Bailey, 2003  U B C Rare Books and Special Collections - Thesis Authorisation Form  Page 1 of 1  In p r e s e n t i n g t h i s t h e s i s i n p a r t i a l f u l f i l m e n t o f the requirements f o r an advanced degree a t the U n i v e r s i t y of B r i t i s h Columbia, I agree t h a t the L i b r a r y s h a l l make i t f r e e l y a v a i l a b l e f o r r e f e r e n c e and study. I f u r t h e r agree t h a t p e r m i s s i o n f o r e x t e n s i v e c o p y i n g o f t h i s t h e s i s f o r s c h o l a r l y purposes may be g r a n t e d by the head of my department o r by h i s o r h e r r e p r e s e n t a t i v e s . I t i s understood t h a t c o p y i n g o r p u b l i c a t i o n of t h i s t h e s i s f o r f i n a n c i a l g a i n s h a l l not be a l l o w e d without my w r i t t e n p e r m i s s i o n .  Department o f The U n i v e r s i t y o f B r i t i s h Columbia Vancouver, Canada  http://\\rvvw.library.ubc.ca/spcoll/thesauth.htrnl  8/1/03  Abstract 1.1  Purpose  1. To determine the strength profile of the thoracolumbar endplates using indentation testing. 2. To determine the regional bone mineral density (rBMD) profile of the thoracolumbar endplates using peripheral quantitative computed tomography (pQCT). 3. To compare the thoracolumbar endplate strength profile with the r B M D profile.  1.2  Method  Indentation tests using a materials testing machine and axial pQCT were performed on the T9, T12, and L2 endplates of seven fresh-frozen human cadaver spines. A minimum of twenty-five indentations was performed in a rectangular grid (7 columns by 5 rows). A 3mm hemispherical indenter was lowered at 0.2rnrn/s to a depth of 3mm producing endplate failure. Regional B M D was determined using a rectangular grid which over lay the indentation tests so that the density could be determined for the region of bone beneath each indentation test. Regional BMDs were manually determined using pQCT software. Repeated measures, multivariate A V O V A was used to analyse the affect the independent variables (level, endplate) and dependent variables (AP and lateral position) had on the endplate strength and r B M D profiles. 1.3  Results  The strongest aspect of the thoracolumbar endplate was in the posterolateral corners. The weakest aspect was found in the centre of the endplate, which was located within the apophyseal ring.  The anterior rim was the second strongest aspect of the endplate,  especially for T9. The density of the sub-endplate zone decreased from the posterior and anterior aspects towards the centre of the endplate.  The density decreased from the  centre towards the lateral periphery except in the posterior row, where the posterolateral corners were the densest of the entire endplate. No significant difference existed in the mean strength or density between level or superior/inferior endplate. The strength profile  II  and r B M D profile was influenced by the sagittal alignment of the spine; L2 was relatively stronger and denser in the posterior of the endplate, while T9 was relatively stronger and denser in the anterior of the endplate.  This study found only a low to moderate  correlation between r B M D and local strength. However, comparison of endplate and r B M D profiles showed similarities. 1.4  Conclusion  In conclusion, the thoracolumbar endplate strength profile revealed that the anterior and posterior aspects of the endplate were significantly stronger than the centre. The anterior and posterior aspects of the sub-endplate zone were denser than was the centre. The posterolateral corners were the strongest and densest part, thought to be due to the pedicle insertion. The apophyseal ring was important in explaining the difference in strength between the centre and periphery of the endplate, apart from the increase in strength between the anterocentral aspect of the endplate and the posterior aspect of the endplate, which occurred within the confines of the apophyseal ring. Spinal sagittal contour was shown to influence the endplate strength and r B M D profile. A low to moderate correlation was found between a thoracolumbar endplate strength profile and thoracolumbar endplate r B M D profile.  1.5  Significance  This knowledge may assist in preventing intervertebral implant subsidence by influencing implant positioning and design.  Ill  Table of Contents Abstract  II  Table of Contents  IV  List of Tables  X  List of Figures  XI  Acknowledgements 1.0 Introduction  XV 1  1.1 Interbody Fusion - Concepts and Devices  1  1.2 Factors Influencing Subsidence at the Bone-Implant Interface  1  1.2.1 Patient Factors  5  1.2.1.1 Osteoporosis  5  1.2.1.2 Dual Energy X-Ray Absorptiometry  6  1.2.1.3 Quantitative Computed Tomography  6  1.2.1.4 Regional Variation in Vertebral Body Density  8  1.2.1.5 The Correlation between Subsidence and B M D  11  1.2.2 Surgical Factors  11  1.2.2.1 Differences in the Bone Interface  12  1.2.2.2 Regional Differences in Vertebral Body Strength  13  1.2.3 Implant Factors  16  1.2.3.1 Area of the Bone-Implant Interface  16  1.2.3.2 Implant Design  17  1.3 Summary  17  1.4 Purpose  19  1.5 Hypothesis  19  2.0 Materials and Methods  21  2.1 Specimen Selection  21  2.2 Specimen Preparation  21  2.3 Dual Energy X-ray Absorptiometry (DXA)  22  2.4 Peripheral Qunantitative Computed Tomography (pQCT)  22  2.5 Analysis of Regional Bone Mineral Density  24  IV  2.6 Indentation Testing  28  2.6.1 Indentor  29  2.6.2 Test Format  29  2.6.3 Indentation Procedure  31  2.7 Analysis  33  2.7.1 Curve Analysis  33  2.7.2 Data Analysis  33  3.0 Results  37  3.1 Endplate Strength Profiles 3.1.1 Maximum Profile Analysis  37 37  3.1.1.1 Effect of AP Position  37  3.1.1.2 Effect of Lateral Position  37  3.1.1.3 Effect of the Interaction between A P and Lateral Position  38  3.1.1.4 Effect of Spinal Level  39  3.1.1.5 Effect of the Interaction between A P Position and Spinal Level  40  3.1.1.6 Effect of the Interaction between Lateral Position and Spinal Level  41  3.1.1.7 Effect of Endplate  42  3.1.1.8 Effect of the Interaction between A P Position and Endplate  42  3.1.1.9 Effect of the Interaction between Lateral Position and Endplate  43  3.1.1.10 Effect of the Interaction between Spinal Level and Endplate... 44 3.1.2 Anterior-Posterior Profile Analysis  45  3.1.2.1 Effect of A P Position  45  3.1.2.2 Effect of the Interaction between A P and Lateral Position  45  3.1.2.3 Effect of Spinal Level  47  3.1.2.4 Effect of the Interaction between A P Position and Spinal Level 3.1.2.5 Effect of Endplate  47 48  3.1.2.6 Effect of the Interaction between A P Position  V  and Endplate  49  3.1.2.7 Effect of the Interaction between Spinal Level and Endplate 3.1.3 Lateral Profde Analysis  49 51  3.1.3.1 Effect of APPosition  51  3.1.3.2 Effect of Spinal Level  51  3.1.3.3 Effect of the Interaction between Lateral Position and Spinal Level 3.1.3.4 Effect of Endplate  52 53  3.1.3.5 Effect of the Interaction between Lateral Position and Endplate  53  3.1.3.6 Effect of the Interaction between Spinal Level and Endplate .... 54 3.1.4 Cancellous Anterior-Posterior Profde Analysis  55  3.1.4.1 Effect of APposition  55  3.1.4.2 Effect of the Interaction between AP and Lateral Position  55  3.1.4.3 Effect of Spinal Level  56  3.1.4.4 Effect of the Interaction between A P Position and Spinal Level  57  3.1.4.5 Effect of Endplate  58  3.1.4.6 Effect of the Interaction between AP Position and Endplate  58  3.1.4.7 Effect of the Interaction between Spinal Level and Endplate .... 59 3.1.5 Ratio Posterior Endplate/Anterior Endplate Analysis 3.1.5.1 Effect of Level 3.2  60  Dexa Scan Results 3.2.1  Vertebral Body Bone Mineral Content (BMC) and Bone Mineral Density (BMD)  3.3  60  61  3.2.2 Effect of Level on Vertebral B M C  61  3.2.3 Effect of Spinal Level on Vertebral Bone Mineral Density (BMD)  62  Regional Bone Mineral Density 3.3.1  63  Maximum Profile Analysis  63  3.3.1.1 Effect of APPosition  63  VI  3.3.1.2 Effect of Lateral Position  63  3.3.1.3 Effect of the Interaction between A P and Lateral Position  64  3.3.1.4 Effect of Spinal Level  65  3.3.1.5 Effect of the Interaction between A P Position and Spinal Level  66  3.3.1.6 Effect of the Interaction between Lateral Position and Spinal Level  67  3.3.1.7 Effect of Endplate  68  3.3.1.8 Effect of the Interaction between AP Position and Endplate  68  3.3.1.9 Effect of the Interaction between Lateral Position and Endplate  69  3.3.1.10 Effect of the Interaction between Spinal Level and Endplate... 70 3.3.2 Anterior-Posterior Profile Analysis  71  3.3.2.1 Effect of AP Position  71  3.3.2.2 Effect of the Interaction between A P and Lateral Position  71  3.3.2.3 Effect of Spinal Level  72  3.3.2.4 Effect of the Interaction between AP Position and Spinal Level  73  3.3.2.5 Effect of Endplate  73  3.3.2.6 Effect of the Interaction between A P Position and Endplate  74  3.3.2.7 Effect of the Interaction between Spinal Level and Endplate .... 75 3.3.3 Lateral Profile Analysis 3.3.3.1 Effect of Lateral Position  76 '.  3.3.3.2 Effect of Spinal Level  76 76  3.3.3.3 Effect of the Interaction between Lateral Position and Spinal Level 3.3.3.4 Effect of Endplate  77 78  3.3.3.5 Effect of the Interaction between Lateral Position and Endplate  79  3.3.3.6 Effect of the Interaction between Spinal Level and Endplate .... 79 3.3.4 Ratio Posterior Endplate / Anterior Endplate Analysis  81  VII  3.3.4.1 Effect of Level 3.4  Correlation between Endplate r B M D and Force  81 81  3.4.1  Maximum Profile Analysis  82  3.4.2  Anterior-Posterior Profile Analysis  82  3.4.3  Lateral Profile Analysis  83  3.4.4  Cancellous Anterior-Posterior Profile Analysis  84  4.0 Discussion  85  4.1 The Thoracolumbar Endplate Strength Profile  85  4.1.1  Effect of Level  85  4.1.2  Effect of Position  86  4.1.2.1 A P Position  86  4.1.2.2 Lateral Position  88  4.1.2.3 Rationale for "Cancellous Endplate Analysis" in the Anterior-Posterior 4.1.3  Effect of Endplate  4.1.4  Summary of the Endplate Strength Profile  4.2 The Thoracolumbar r B M D profile  89 90  91  4.1.5  Effect of Level  91  4.1.6  Effect of Position  92  4.1.6.1 APPosition  92  4.1.6.2 Lateral Position  94  4.1.7  Effect of Endplate  95  4.1.8  Summary of the Endplate r B M D Profile  95  4.3 Correlation between the Endplate Strength Profile and r B M D Profile  95  4.4 Study Strengths and Weaknesses  97  4.5 Future Work and Clinical Implications  99  5.0 Conclusion  100  6.0 References  102  7.0 Appendix A: Indentation Tests  109  7.1 Maximum Profile Analysis  109  7.2 Anterior-Posterior Profile Analysis  112  VIII  7.3 Lateral Profile Analysis  114  7.4 Cancellous Anterior-Posterior Profile Analysis  115  7.5 Ratio Posterior Endplate/Anterior Endplate Analysis  116  8.0 Appendix B : Regional Bone Mineral Density  117  8.1 Maximum Profile Analysis  117  8.2 Anterior-Posterior Profile Analysis  120  8.3 Lateral Profile Analysis  122  8.4 Ratio Posterior/Anterior Profile analysis  123  IX  List of Tables Table 2.1: Specimen attributes  21  Table 3.1: D E X A values for each specimen  61  X  List of Figures Introduction Figure 1.1: Degenerative spondylolisthesis  1  Figure 1.2: Lateral and A P x-ray of TLIF  2  Figure 1.3: A P and lateral x-ray of ALIF  3  Figure 1.4: Cage and plate reconstuction  3  Materials and Methods Figure 2.1: pQCT with custom-made tube for suspending vertebrae  23  Figure 2.2: Mask of five axial cuts  24  Figure 2.3: Grid system of squares for r B M D analysis  25  Figure 2.4: Regional B M D grid of 5 rows  26  Figure 2.5: Regional B M D grid of 7 columns  26  Figure 2.6: r B M D grid with labeled columns and rows  27  Figure 2.7: Areas within r B M D grid for which B M D was determined  27  Figure 2.8: Hemispherical indenter  29  Figure 2.9: Rectangular grid system used to perform indentation tests  30  Figure 2.10: Custom made jig for securing vertebral body  31  Figure 2.11: Indentation test setup  32  Figure 2.12: Indentation test load-displacement curve  33  Figure 2.13: Maximum, Anterior-Posterior, and Lateral profile analyses  35  Figure 2.14: Cancellous Anterior-Posterior profile analysis  36  Results  Endplate Strength Profiles  37  Figure 3.1: Effect of A P position in the Maximum profile analysis  37  Figure 3.2: Effect of lateral position in the Maximum profile analysis  38  Figure 3.3: Effect of interaction between A P and lateral position in the Maximum profile analysis Figure 3.4: Effect of spinal level in the Maximum profile analysis  39 40  Figure 3.5: 2-way Interaction between spinal level and A P position in the Maximum profile analysis  41  XI  Figure 3.6: Effect of interaction between spinal level and lateral position in the Maximum profile analysis Figure 3.7: Effect of endplate in the Maximum profile analysis  41 42  Figure 3.8: 2-way interaction between A P position and endplate in the Maximum profile analysis  43  Figure 3.9: Effect of interaction between lateral position and endplate in the Maximum profile analysis  44  Figure 3.10: Effect of interaction between endplate and spinal level in the Maximum profile analysis Figure 3.11: Effect of AP position in the Anterior-Posterior profile analysis  44 45  Figure 3.12: Two-way interaction between A P and lateral position in the Anterior-Posterior profile analysis Figure 3.13: Effect of spinal level in the Anterior-Posterior profile anaylsis  46 47  Figure 3.14: Effect of AP position and spinal level interaction in the Anterior-Posterior profile analysis Figure 3.15: Effect of endplate in the Anterior-Posterior profile analysis  48 48  Figure 3.16: Effect of interaction between A P position and endplate in the Anterior-Posterior profile analysis  49  Figure 3.17: Two-way interaction between spinal level and endplate in the Anterior-Posterior profile analysis  50  Figure 3.18: Effect of lateral position in the Lateral profile analysis  51  Figure 3.19: Effect of spinal level in the Lateral profile analysis  52  Figure 3.20: Two-way interaction between lateral position and spinal level in Lateral profile analysis Figure 3.21: Effect of endplate in the Lateral profile analysis  52 53  Figure 3.22: Effect of lateral position and endplate in the Lateral profile analysis Figure 3.23: Effect of endplate and spinal level in the Lateral profile analysis  54 54  Figure 3.24: Effect of AP position in the Cancellous Anterior-Posterior profile analysis  55  Figure 3.25: Effect of the interaction between A P and lateral position  XII  in the Cancellous Anterior-Posterior profile analysis  56  Figure 3.26: Effect of spinal level in the Cancellous Anterior-Posterior profile analysis  57  Figure 3.27: Effect of the interaction between A P position and spinal level in the Cancellous Anterior-Posterior profile analysis  57  Figure 3.28: Effect of endplate in the Cancellous Anterior-Posterior profile analysis  58  Figure 3.29: Effect of AP position in the Cancellous Anterior-Posterior profile analysis  59  Figure 3.30: Effect of the interaction between spinal level and endplate in the Cancellous Anterior-Posterior profile analysis  59  Figure 3.31: Ratio of posterior endplate / anterior endplate  60  Figure 3.32: Effect of the interaction between B M C and spinal level  61  Figure 3.33: Effect of the interaction between spinal level and D X A B M D  62  Endplate Regional Bone Mineral Density Profiles  63  Figure 3.34: Effect of AP position on endplate in the Maximum profile  63  Figure 3.35: Effect of lateral position in the Maximum profile analysis  64  Figure 3.36: Effect of interaction between A P and lateral position in the Maximum profile analysis Figure 3.37: Effect of spinal level in the Maximum profile analysis  65 66  Figure 3.38: Effect of interaction between spinal level and A P position in the Maximum profile analysis  67  Figure 3.39: Effect of interaction between spinal level and lateral position in the Maximum profile analysis Figure 3.40: Effect of endplate in the Maximum profile analysis  67 68  Figure 3.41: Effect of the interaction between A P position and endplate in the Maximum profile analysis  69  Figure 3.42: Effect of interaction between lateral position and endplate in the Maximum profile analysis  69  Figure 3.43: Effect of interaction between endplate and spinal level in the  XIII  Maximum profile analysis  70  Figure 3.44: Effect of AP position in the Anterior-Posterior profile analysis  71  Figure 3.45: Effect of interaction between A P and lateral position in the Anterior-Posterior profile analysis  72  Figure 3.46: Effect of spinal level in the Anterior-Posterior profile anaylsis  72  Figure 3.47: Effect of interaction between A P position and spinal level interaction in the Anterior-Posterior profile analysis  73  Figure 3.48: Effect of endplate in the Anterior-Posterior profile analysis  74  Figure 3.49: Effect of interaction between A P position and endplate in the Anterior-Posterior profile analysis  75  Figure 3.50: Effect of the interaction between spinal level and endplate in the Anterior-Posterior profile analysis Figure 3.51: Effect of lateral position in the Lateral profile analysis  75 '.  Figure 3.52: Effect of spinal level in the Lateral profile analysis  76 77  Figure 3.53: Effect of the interaction between lateral position and spinal level in the Lateral profile analysis Figure 3.54: Effect of endplate in the Lateral profile analysis  78 78  Figure 3.55: Effect of interaction between lateral position and endplate in the Lateral profile analysis  79  Figure 3.56: Effect of the interaction between endplate and spinal level in the Lateral profile analysis  80  Figure 3.57: Ratio of posterior endplate / anterior endplate  81  Correlation between Endplate r B M D and Force  82  Figure 3.58: Maximum profile analysis  82  Figure 3.59: Anterior-Posterior profile analysis  83  Figure 3.60: Lateral profile analysis  83  Figure 3.61: Cancellous Anterior-Posterior profile analysis  84  XIV  Acknowledgements Thank you to Dr. Tom Oxland for providing me with an excellent experience in your lab. I appreciate your time, patience, and guidance which ensured the completion of my thesis. Under your supervision I have learned much, right up to the completion of my final thesis draft, regarding basic science research and the unique way to ask and answer questions in a biomechanics lab.  Thank you to Dr. Marcel Dvorak and Dr. Charles Fisher for allowing me to be the spine groups first research fellow.  During my two years in Vancouver your mentorship as  clinicians and researchers has been invaluable. Thank you Dr. Chris Reilly for acting as my external examiner. Thank you to Simon Sjovold, Leslie Bryant, and Dr. Heather McKay, without your important contribution this project would never have been possible.  And Michelle, my best friend and wife to be, I can't image what these two years in Vancouver would have been like without you. Your support, understanding, and love mean the world to me.  XV  1.0 Introduction 1.1  Interbody Fusion - Concepts and Devices  The current gold standard for treating spinal instability is a spinal instrumentation and fusion procedure. Spinal instability is most commonly understood to represent abnormal movement that leads to pain, neurologic injury, or progressive spinal deformity (White and Panjabi, 1990). This relatively generic treatment has as its goal a bony arthrodesis and is applicable to a diverse range of spinal problems: degenerative disc disease, spinal trauma, tumours, infection, and deformity,  as well  as revision spinal surgery.  Degenerative disc disease (DDD) is the most common indication for spinal arthrodesis and may involve the entire spinal column, cervical to lumbosacral. D D D may result in spinal instability presenting as spondylolisthesis or scoliosis of the thoracolumbar or lumbar spine (Figure 1.1). Alternatively, DDD causing spinal stenosis may require a wide decompression resulting in iatrogenic instability. In both scenarios, D D D in association  Figure 1.1: Degenerative spondylolisthesis at L3-4 and L4-5 with secondary spinal stenosis at both levels.  with spinal instability is potentially treated with an instrumentation and fusion procedure. As well, although controversial, mechanical low back pain is frequently attributed to D D D and treated with spinal instrumentation and fusion.  l  A spinal fusion and instrumentation procedure for the treatment of D D D may utilize an interbody device. Such a device will rest in the disc space between the superior and inferior endplates of adjacent vertebral bodies (Figures 1.2 - 1.4). A n interbody fusion procedure provides spinal stabilization through the anterior column. The anterior column is defined by the vertebral body, anterior longitudinal ligament, posterior longitudinal ligament, annulus fibrosis, and nucleus pulposus (Holdsworth, 1970).  A n interbody  fusion device provides a number of beneficial functions: it restores disc height and tensions the ligamentous supporting structures of the anterior column, provides decompression of the exiting nerve root within the neuroforamen, helps to correct and maintain sagittal and coronal balance, provides a large area for potential fusion, and facilitates load sharing through the vertebral body (Burkus, 2002; Lin, 1985; McAfee, 1999; Oxland and Lund, 2000; Steffen et al., 2000b).  A n interbody device may be inserted from a posterior or anterior approach. Posterior approaches include the Posterior Lumbar Interbody Fusion (PLIF) or Transforaminal Lumbar Interbody Fusion (TLIF) and are typically performed in conjunction with some form of concomitant posterior fixation (Figure 1.2).  Figure 1.2: Lateral and A P x-ray of an L2-3 left sided T L I F with concomitant pedicle screw instrumentation.  2  The Anterior Lumbar Interbody Fusion (ALIF) may stand on its own as a solitary procedure (Burkus et al, 1999; Burkus, 2002; Hacker, 2003; Kuslich et al, 1998) or be combined with posterior segmental instrumentation as a 360 degree anterior and posterior procedure (Figure 1.3).  Figure 1.3: A P and lateral x-ray of an L5-S1 A L I F in association with pedicle screw instrumentation.  In the case of some spinal trauma, spinal tumours, particularly primary spinal column tumours, and vertebral infection, such as mechanically unstable vertebral osteomylitis or tuberculosis, anterior column reconstruction and stabilization follows single or multilevel verbrectomy and stabilization. In these situations, the thoracolumbar spine is approached  Figure 1.4: Cage and plate reconstruction L1-L3 for a metastatic tumour of L 2 .  3  directly through a thoracotomy, thoracoabdominal, or retroperitoneal approach. In these cases, interbody fusion devices are much larger then those used for a PLIF, TLIF, or ALIF, but provide the same function: immediate height restoration, realignment, stabilization, and site for fusion (Fig 1.4).  Various options are available for anterior column reconstruction and are dependent upon the surgical indication. These include tricortical strut grafts (autograft, allograft, and xenograft) such as femoral dowels, vascular fibula, or rib, and a variety of metal or synthetic cages. Cage technology and design continues to evolve, but they all function by transferring load from the rostral to caudal vertebral body, and by containing osteogenic, and/or osteoconductive material to facilitate fusion across the region spanned by the cage. Success rates are reportedly high, with some authors reporting fusion rates and successful patient outcomes greater than 90% (Holte et al, 1994; Kuslich et al, 1998; Ray, 1997).  Although interbody fusion devices range in type and size and are utilized in a variety of clinical circumstances, they all suffer similar modes of failure. They are recognised to fracture or collapse, subside into the supporting vertebral body, or dislodge and migrate (Burkus, 2002; Steffen et al, 2000b; Hoshijima et al, 1997; Jost et al, 1998; Hollowell et al, 1996; Lin, 1985). A l l modes of failure can lead to progressive deformity, potential pseudoarthosis, and possible patient dissatisfaction. Height loss following anterior spinal reconstruction was attributed to either graft collapse or subsidence. Cages were designed in hopes of eliminating or at least decreasing the rate of device collapse/failure. Today, one of the most recognised failure modes is interbody device subsidence (Lane and Sandhu, 1997).  1.2  Factors Influencing Subsidence at the Bone-Implant Interface  Subsidence of the supporting device into the vertebral body occurs following failure of the bone at the bone-implant interface. Many factors contribute to the strength of the bone-implant interface. They can be divided into patient factors, surgical factors, and implant factors.  4  1.2.1  Patien t Factors  1.2.1.1 Osteoporosis Osteoporosis is a systemic skeletal disease characterised by low bone mass and microarchitectural deterioration of bone tissue, with consequent bone fragility (Lu et al., 2001). It is a major public health problem that promises to escalate with our ageing population. The Osteoporosis Society of Canada reports that 1.4 million Canadians suffer from osteoporosis (Osteoporosis Society of Canada, 2003). One in four women over the age of fifty have osteoporosis, while one in eight men over the age of fifty has the disease. The spine is significantly affected, with vertebral fractures accounting for approximately 50% of all osteoporotic fractures. A n estimated one third of women over the age of fifty will suffer from an osteoporotic vertebral compression fracture (Osteoporosis Society of Canada, 2003). Interbody stabilization may be required for patients with significant osteopenia secondary to primary or secondary osteoporosis. The major surgical problem encountered with these patients is the diminished mechanical properties of bone reflected by a decrease in bone density (Dvorak and Fisher, 1999; Harkey, 1996; Heggeness, 1997; Lane and Sandhu, 1997; Oxland and Lund, 2000). Osteoporotic bone makes surgical reconstruction and stabilization difficult, and is associated with an increased failure rate (Boden and Sumner, 1995; Dvorak and Fisher, 1999; Kostuik and Heggeness, 1997).  In the osteoporotic population, anterior spinal  column stabilization is commonly associated with subsidence of the implant or strut graft (Kumar and Doherty, 1994; Lane and Sandhu, 1997; Sandhu et al, 1996) and often leads to failure (Dvorak and Fisher, 1999; Harkey, 1996).  The severity of osteopenia is an important patient factor affecting the strength of the bone-implant interface. Osteoporosis is defined as a bone mineral density (BMD) measuring 2.5 standard deviations below the mean bone mass for a patient's age and gender (Lu et al, 2001). Vertebral body B M D has been shown to correlate directly with vertebral strength in compression loading (Brassow et al, 1982; Brinckmann et al, 1989; Closkey et al, 1993; Cody et al, 1991; Eriksson et al, 1989; Galante et al, 1970; Grant et al, 2001; Grant et al, 2002; Hannson et al, 2003; Hansson et al, 1980; Hansson and Roos, 1981; Hasegawa et al, 2001; Hollowell et al, 1996; Hoshijima et al, 1997; Jost et  5  al, 1998; Keller.ef al, 1989; Knop et al, 2001; Kumar and Doherty, 1994; Lang et al, 1988; Lim et al, 2001; McBroom et al, 1985; McCubbrey et al, 1995; Myers et al, 1994; Oxland et al, 1996; Oxland et al, 2003; Singer et al, 1995; Steffen et al, 2000a; Veenland et al, 1997). This suggests that patients with osteoporosis are at higher risk of subsidence of the graft or implant following interbody fusion.  1.2.1.2 Dual Energy X-Ray Absorptiometry Dual energy X-ray absorptiometry (DXA) is an accurate measure of bone mineral density as well as a reproducible predictor for osteoporotic fractures. The precision error for measurement of the lumbar spine is 0.5 - 5.9% coefficient of variation, with an accuracy error of 4 - 8% (Guglielmi et al, 1995). The advantage of D X A scanning is that it is precise and delivers a low radiation dose. Both posteroanterior (PA) and lateral D X A scanning have been utilitzed. Posteroanterior D X A scanning is much more common in clinical practice and is more precise (Guglielmi et al, 1995). However, lateral D X A is more accurate and predictive of B M D than is P A D X A (Guglielmi et al, 1995; Jost et al, 1998; Myers et al, 1994). The enhanced predictive value of supine lateral B M D occurs because of the variable contribution of posterior element mineral to the P A B M D measurement (Guglielmi et al, 1995; Mazess et al, 1991; Myers et al, 1994). Oxland et al. suggested that for advanced disc degeneration, the peripheral endplate osteophyes increase the mean B M D above the true trabecular density (Oxland et al,  1996).  Biomechanical investigation has found a positive correlation between both P A and lateral D X A determined B M D with vertebral body, trabecular and endplate strength (Grant et al, 2001; Grant et al, 2002; Hasegawa et al, 2001; Hollowell et al, 1996; Hoshijima et al, 1997; Jost et al, 1998; Kumar and Doherty, 1994; L i m et al, 2001; Oxland et al, 1996; Oxland et al, 2003; Singer et al, 1995).  1.2.1.3 Quantitative Computed Tomography Quantitative  computed  tomography  (QCT) is  widely accepted  clinically  and  experimentally in the measurement of B M D (Lang et al, 1998; Mirsky and Einhorn, 1998). Because it images in a cross-sectional plane, QCT is unique among methods of measurement in providing a true volumetric mineral density (Guglielmi et al, 1995; Lang  6  et al, 1998; L u et al, 2000; Mirsky and Einhorn, 1998). With other instruments, the density of a region of interest is calculated by dividing the bone-mineral content by the entire area scanned, and expressed as a volumetric density (Mirsky and Einhorn, 1998). The precision and accuracy errors of QCT of the lumbar spine are approximately 1 - 4% coefficient of variation and 4 - 15%, respectively (Guglielmi et al, 1995). QCT B M D has been positively correlated using biomechanical models with vertebral body strength in compression (Brassow et al, 1982; Brinckmann et al, 1989; Cody et al, 1991; Eriksson et al, 1989; Knop et al, 2001; McBroom et al, 1985; McCubbrey et al, 1995; Singer et al, 1995; Veenland et al, 1997). The decrease in QCT B M D has also been confirmed to correlate with an increase in subject age (Brassow et al, 1982; McCubbrey etal, 1995; Singer and Breidahl, 1990).  The greatest advantage of QCT is the ability to localize B M D measurements.  Separate  cortical and cancellous B M D measurements can be determined, and surrounding or superimposed tissue does not affect the density measurement (Brassow et al, 1982; Knop et al, 2001; Lang et al, 1998; Lu et al, 2000). Furthermore, regionalized B M D (rBMD) measurements can determine density differences between different areas within a single vertebral body, i.e. superior versus inferior, or anterior versus posterior (Antonacci et al, 1997; Banse et al, 2002; Cody et al, 1991; Hasegawa et al, 2001; Lang et al, 1988; L u et al, 2000; Matsui, 1991; McCubbrey et al, 1995; Nepper-Rasmussen and Mosekilde, 1989; Singer and Breidahl, 1990). Typically, the mean vertebral B M D is approximated from the measurement of an anterocentral elliptical region within the cancellous body (Cody et al, 1991; Mirsky and Einhorn, 1998).  It has been shown that the use of  multiple regional cancellous densities produced a stronger correlation with vertebral failure than does any single density parameter (Cody et al, 1991; McCubbrey et al, 1995).  A further advantage of QCT is its ability to analyse the structural properties of trabecular bone in conjunction with cancellous B M D (Andresen et al, 1999; Banse et al, 2001; Banse et al, 2002; Brinckmann et al, 1989; Simpson et al, 2001; Singer et al, 1995; Singer and Breidahl, 1990; Thomsen et al, 2000; Thomsen et al, 2002; Veenland et al,  7  1997). Andersen et al., after studying forty women with QCT, demonstrated that B M D measurement along with an assessment of quantitative structural properties improves the ability to discriminate between patients with and without osteoporotic compression fractures (Andresen et al, 1999). Eriksson et al., Veenland et al., and Singer et al., found a better correlation between QCT trabecular B M D in association with vertebral cross section area and the ultimate vertebral failure load, than the predictive ability of a mean vertebral B M D alone, as measured by dual photon energy absorptometry, D X A or QCT (Eriksson et al, 1989; Singer et al, 1995; Veenland et al, 1997). Brinckmann et al. correlated vertebral cross sectional area with whole body vertebral strength (Brinckmann et al, 1989). It has been shown that as vertebral cross sectional area increases with caudal descent so does vertebral strength in axial compression (Brinckmann et al, 1989; Singer et al, 1995).  Peripheral QCT (pQCT) is a newer technique than traditional QCT, designed for the analysis of smaller subjects (animals) or body parts (wrist or ankle). It has been recently utilized by a number of investigators in the determination of r B M D for cadaver spinal elements (Banse et al, 2001; L u et al, 2000; Zheng et al, 2000). Peripheral QCT has the same precision and reproducibility as the QCT, with better accuracy (Augat et al, 1998; Groll et al, 1999; L u et al, 2000; Sievanen et al, 1998; Takada et al, 1996).  1.2.1.4 Regional Variation in Vertebral Body Density QCT density analysis has shown that the vertebral body is very heterogeneous. Antonacci et al. studied the regional variations in thoracic, thoracolumbar, and lumbar cadaver vertebral body density using QCT (Antonacci et al, 1997). Four regions of interest, anterior, posterior, superior, and inferior, were analysed in two sections, midsagittal and para-sagittal.  No differences were documented between levels.  The  anterolateral cancellous bone was significantly more dense than was the anterocentral cancellous bone. In the midline, the inferior endplate was significantly more dense than the superior, while laterally although less of a difference in density existed, the opposite was found, with the superior endplate being more dense than the inferior. In the midsagittal cut the QCT B M D for both superior and inferior endplates was positively  8  correlated with the decreasing mean D X A B M D . Contrary to this described correlation with central QCT B M D , laterally, the superior and inferior endplates QCT B M D was relatively unaffected by a reciprocal decrease in D X A B M D (Antonacci et al, 1997). This illustrates that regional variability of density within a vertebral body is dynamic.  Banse et al. studied the density and structure of standardized cored samples taken from T9, T12, and L4 of nine cadaver vertebrae (Banse et al, 2001). Peripheral QCT was performed for each core at four different axial heights, spanning the superior endplate to the inferior endplate.  Core samples with a diameter of 8.2 mm, were harvested from  anterior, posterior, and midlateral positions. In keeping with the findings of Antonacci et al., no differences between levels were found, and the lateral halves were symmetric (Antonacci et al, 1997). A significant difference in density was found between core samples harvested from the anterior aspect of the vertebral bodies and those retrieved from the posterior aspect. The posterior density was 20% greater. The inferior half of the vertebral body was significantly more dense than the superior half. The most significant vertical inhomogeneity was found in the posterior and lateral cores, while the difference in B M D was relatively small and insignificant in the anterior samples. This difference they attribute to the effect of the pedicle insertion.  They believe that because the  posterior vertebral wall is substantially reinforced in its superior half by the inserting cortical bone of the pedicles, the lower half of posterior cancellous bone is "reinforced" in order to transmit load from the pedicle and thick superior posterior wall to the inferior endplate.  Although a difference in density was found between the superior and inferior  half of the vertebral, no difference was found between cancellous bone centrally and that close to either endplate. In contrast, other investigators have shown a relative increased QCT trabecular B M D of the sub-endplate zone relative to the QCT B M D of the trabecular centre (Simpson et al, 2001; Thomsen et al, 2002; Zheng et al, 2000). Thomsen et al. showed that as a patient ages this discrepancy increased (Thomsen et al, 2002).  An increased density in the posterior, inferior and lateral vertebral cancellous body has been confirmed by others (Edwards et al, 1986; Flynn and Cody, 1993; Matsui, 1991;  9  Nepper-Rasmussen and Mosekilde, 1989; L u et al, 2000). Simpson et al., Edwards et al., Matsui, and Flynn et al., utilized QCT to confirm this regional variation of vertebral body B M D (Edwards et al, 1986; Flynn and Cody, 1993; Matsui, 1991; Simpson et al, 2001). Nepper-Rasmussen et al. evaluated six regions of interest (ROI) in twelve lumbar cadaver vertebrae (Nepper-Rasmussen and Mosekilde, 1989). The analysis of a central QCT section of each vertebral body revealed a similar disparity between anterior and posterior B M D (Nepper-Rasmussen and Mosekilde, 1989). They also found a strong correlation between QCT B M D and the ash weight for all 72 regions of interest (6 ROI X 12 vertebra) (Nepper-Rasmussen and Mosekilde, 1989). L u et al. analysed the r B M D of the lumbar and sacral spine (Lu et al, 2000). Peripheral QCT was utilized to perform five transverse 1mm cuts equally positioned between the superior and inferior endplates. Five vertical regions of interest were established: posterolateral, anterolateral, anterocentral, middle, and posterocentral. The trabecular B M D increased between L I to SI. Lateral ROI were symmetric side to side. The B M D was greatest near the subchondral endplate, and least in the centre. The posterolateral and middle ROI having the highest density and the posteromiddle ROI the lowest density was the general trend for the entire lumbosacral spine. However, this difference did not reach significance in the upper lumbar spine (LI and L2). They believe that bone density reflects loading patterns which explains the increased densities at the base of the pedicles (posterolateral) and underlying the disc (Lu et al, 2000).  As stated above, both Antonacci et al. and Banse et al. found that the mean vertebral B M D did not vary with level (Antonacci et al, 1997; Banse et al, 2001). This has been confirmed by Brinckmann et al., who studied human lumbar vertebral bodies using QCT (Brinckmann et al,  1989). In contrast to these findings, other investigators have  independently found that the B M D and bone mineral content (BMC) increases in the lumbar spine when moving from a cranial to caudal direction (Eriksson et al, 1989; Hansson et al, 1980; Singer and Breidahl, 1990). Hansson et a l , used dual energy photon absorptiometry to show an increase in the mean B M D between L I and L3 (Hansson et al, 1980). While Errickson et al. found an increase in lumbar B M D and B M C moving in a craniocaudal direction following analysis with both QCT and dual  10  energy photon absorptiometry (Eriksson et al, 1989), Brassow et al. found no difference in the mean cancellous B M D between thoracic and lumbar vertebrae, but showed an increasing mean cortical B M D with caudal decent (Brassow et al, 1982). Singer et al. found that the mean D X A B M D and B M C increased when moving from thoracic spine to lumbar spine in a craniocaudal direction, but inversely found QCT trabecular B M D to decrease (Singer and Breidahl, 1990).  1.2.1.5 The Correlation between Implant Subsidence and B M D Clinical experience suggests that osteoporosis is a significant risk factor for interbody device subsidence. Biomechanical studies have demonstrated that a decrease in B M D significantly correlates with a decrease in failure load at the bone-implant interface (Hollowell et al, 1996; Jost et al, 1998; Knop et al, 2001; Oxland et al, 1996; Steffen et al, 2000a; Hoshijima et al, 1997).  Hollowell et al. compared seven different  constructs of anterior interbody devices and looked at subsidence into the vertebral bodies harvested from the thoracic spine (Hollowell et al, 1996).  These construct  designs included: a titanium mesh cage, humeral allograft, iliac strut graft with two different techniques of endplate preparation, triple rib strut graft, and single rib strut graft with two different techniques of endplate preparation. Although significant differences existed in the mean failure load between the interbody implants and constructs, B M D had a significant influence on failure load for all seven constructs (Hollowell et al, 1996). Jost et al., compared the failure load of three different PLIF cage designs in lumbar spine segments (Jost et al, 1998). In all three cases, a significant correlation existed between mean vertebral D X A B M D and compression failure (Jost et al, 1998).  1.2.2  Surgical Factors  The preparation of the bone bed is very important to the strength of the bone-implant interface.  Controversy exists in the literature regarding two points: should the bone-  implant interface occur on the subchondral bony endplate or should it be removed, and does it matter where on the bone interface the implant rests (i.e. posterior, central or anterior)?  11  1.2.2.1 Differences in the Bone Interface. Across the bone-implant interface two separate processes must occur. The implant must remain stable and the graft must consolidate to provide an effective arthrodesis (Burkus, 2002; Evans, 1985; Kozak et al, 1994; Lin, 1985; McAfee, 1999). It is commonly believed that higher union rates occur when internal fixation is used to decrease motion in the fusion segment (Boden and Sumner, 1995; Lin, 1985). In order to enhance incorporation of graft material, it has been suggested that the vertebral endplate be removed and the implants should rest on a cancellous bed (Hollowell et al, 1996; Lin, 1985; Pearcy and Evans, 2003; Steffen et al, 2000b). Cancellous bone has better osteoinductive properties then cortical bone (Boden and Sumner, 1995). However, it remains controversial as to whether removal of the endplate decreases the strength of the implant-bone interface, thereby predisposing the interbody device to subsidence. Studies have shown that preservation of the endplate offers no biomechanical advantage (Hollowell et al, 1996; Jost et al, 1998; Steffen et al, 2000a). Furthermore, maintaining 30-40% surface coverage of the trabecular surface by the implant has been shown to significantly limit subsidence (Closkey et al, 1993; Pearcy and Evans, 2003). Others contend that the subchondral bone is structurally stronger in compression than the trabecular bone beneath (Hasegawa et al, 2001; Kumar and Doherty, 1994; Lim et al, 2001; Oxland et al, 2003; Zaccaria, 2003). Opponents to maintaining endplate integrity (Hollowell et al, 1996; Jost et al, 1998; Steffen et al, 2000a) reported a lower specimen mean age or higher specimen mean B M D than those for the specimens used by Grant et al. and some of the other proponents (Grant et al, 2001; Kumar and Doherty, 1994; Lim et al, 2001) (important to note that the mean age and/or mean B M D was not reported in all studies). This suggests that the endplate is important in preventing subsidence in the osteoporotic spine.  Lim et al., found that failure load did not have a significant association with endplate thickness, but tended to decrease with incremental removal of the endplate (Lim et al, 2001). Load to failure of the specimens with an intact endplate was significantly greater than that of the specimens with no endplate. As strength did not correlate with endplate thickness, they suggest that the underlying trabecular bone is important in resisting the  12  deformation of the endplate to compressive loads (Lim et al, 2001). Kurowski et al. found the highest endplate stress concentration to be located close to the trabecularendplate border (Kurowski and Kubo, 1986). Kumar et al., following indentation tests on the endplates of the lumbar spine, found the zone of maximum strength to extend as much as 4mm into the vertebrae (Kumar and Doherty, 1994).  In considering the  correlation between strength and density, the relative increased QCT trabecular B M D of the sub-endplate bone relative to the QCT B M D of the trabecular centre lends indirect support to maintaining the endplate and sub-endplate regions (Simpson et al, 2001; Thomsen et al, 2002; Zheng et al, 2000). Clinically, maintaining the integrity of the endplate is believed to be important in resisting implant subsidence (Kozak et al, 1994; Lin, 1985) especially in the face of osteoporosis (Dvorak and Fisher, 1999).  1.2.2.2 Regional Differences in Vertebral Body Strength - Endplate Strength Profile The other important surgical factor to consider regarding the effect of osteoporosis on implant subsidence, is where on the bone interface the implant should be best positioned. Various studies have tested both the endplate and trabecular bone beneath the endplate, to determine its strength profile (Edwards et al, 1986; Grant et al, 2001; Grant et al, 2002; Keller et al, 1989; Kumar and Doherty, 1994; Lang et al, 1998; L i m et al, 2001; Oxland et al, 2003). Keller et al. found superior strength in the centre of the vertebra compared to the periphery (Keller et al, 1989). Most other studies have described the opposite (Grant et al, 2001; Grant et al, 2002; Kumar and Doherty, 1994; Oxland et al, 2003) or found no regional variation (Lang et al, 1988). Keller et al. harvested 12 one centimetre trabecular cubes from both the cranial and caudal aspects of twelve lumbar vertebra retrieved from three cadavers (Keller et al, 1989). Each vertebral centrum was therefore divided into 24 cubes, represented by three cubes in the sagittal plane, by four cubes in the coronal plane, and by two layers of twelve cubes (4 X 3) in the axial plane (caudal and cranial). Compression tests of each cube showed that the trabecular bone in the posterocentral area was significantly stronger and stiffer than the surrounding peripheral trabecular bone. They also found that with increasing disc degeneration, the strength in the posterocentral region decreased, while the strength at the endplate's periphery increased but did not surpass that of the posterocentral area.  This was  13  attributed to the change in load transmission associated with progressive disc degeneration. They hypothesised that as the nucleus pulpous degenerates, the annulus fibrosis plays a more dominant role in stress transfer. According to Wolffs law, the change in trabecular strength may reflect this increased load transfer to the periphery of the endplate (Keller et al, 1989).  This same phenomena has been further described by others (Kurowski and Kubo, 1986; Grant et al, 2002; Shirazi-Adl, 1992; Shirazi-Adl et al, 1984). Kurowski et al. analysed endplate loading and the effect of degenerative disc disease using finite element analysis (Kurowski and Kubo, 1986).  They found that as the disc degenerate the stress  concentration migrates from a centre to more lateral position.  With advanced  degeneration the stress is located at the endplate rim and vertebral body wall.  They  hypothesised that this change in stress concentration may result in the formation of osteophytes (Kurowski and Kubo, 1986). Grant et al. mapped the endplate strength profile of L3-L5 with indentation testing and analysed changes resulting from progressive levels of disc degeneration (Grant et al, 2002). Their results were similar to Keller et al., in that for the superior endplate the centre became weaker relative to the periphery with advancing levels of disc degeneration (Keller et al, 1989). They supported the concept of adaptive bone remodelling with the alteration in disc loading associated with advanced degeneration. However, for the inferior endplate, a more uniform decrease in strength was seen across the endplate with advancing disc degeneration (Grant et al, 2002).  Despite this recognised change in the lumbar endplate strength profile with advancing disc degeneration (Grant et al, 2002; Keller et al, 1989; Kurowski and Kubo, 1986), Hansson et al. found no difference in overall vertebral strength with varying DDD, and Oxland et al. showed only a moderate effect of DDD on the strength of the bone-implant interface (Hannson et al, 2003; Hansson and Roos, 1981; Oxland et al, 1996).  In contrast to the findings of Keller et al. recent work using precise indentation testing to map the lumbosacral endplates has shown the posterolateral aspect of the lower lumbar and lumbosacral endplates to be significantly stronger than the anterior or central aspects  14  (Grant et al, 2001). It was found that the L3-L5 surfaces were stronger around the periphery than centrally and stronger at the rear margin when compared to the front. Using a similar indentation protocol on the lower lumbar and S I endplates, the findings of Kumar et al. were in agreement with Grant's (Kumar and Doherty, 1994). Edwards et al. compressed cylinders of trabecular bone harvested from standard positions of nine lumbar spines (Edwards et al, 1986). They also found that anterior trabecular bone is weaker than posterior trabecular bone (Edwards et al, 1986). A significant difference in methodology of the studies reported above may help explain the discrepancy found in the vertebral strength profile (Edwards et al, 1986; Grant et al, 2001; Grant et al, 2002; Keller et al, 1989; Kumar and Doherty, 1994; Lang et al, 1998; Oxland et al, 2003). The work by Grant et al. and Kumar et al. utilized a small indenter pressed into the vertebral endplate, while Keller et al., Lang et al., and Edward et al. compressed small cores or cubes of trabecular bone which excluded the endplate (Edwards et al, 1986; Grant et al, 2001; Grant et al, 2002; Keller et al, 1989; Kumar and Doherty, 1994; Lang et al, 1998; Oxland et al, 2003).  The difference in the endplate strength profile would suggest that varying positions of small interbody devices, such as a PLIF or ALIF device, may influence the strength of the implant-bone interface and affect the risk of device subsidence, especially in the face of osteoporosis and DDD. One study has recently investigated the effect of position on the endplate failure loads for PLIF cages (Labrom, 2003). Although no significant difference between three cage positioning constructs was found, the cages positioned on the posterolateral aspect of the endplate tended to fail at higher loads (Labrom, 2003).  It has been shown that the inferior endplates of the lower lumbar spine have significantly more strength than the corresponding superior endplates (Grant et al, 2001; Kumar and Doherty, 1994). However, no significant difference in load to failure was found for trabecular bone located beneath the superior endplate compared to bone located beneath the inferior endplate (Keller et al, 1989).  15  Whether the vertebral body strength changes with spinal column level also remains controversial. Yoganandan et al. compressed T12 to L5 vertebral bodies with their endplate removed to 50% of their starting height (Yoganandan et al, 1988).  No  difference in strength was found between levels (Yoganandan et al, 1988). As well, in the work of Grant et al, no difference in the endplate strength profiles was found for the lower lumbar vertebrae (L3-L5) when the inferior endplates and superior endplates were compared separately (Grant et al, 2001).  Hansson et al. found that the L3 vertebral  body was significantly stronger in compression testing than was the L I vertebral body (Hansson et al, 1980). Singer et al. on the other hand found an increase in vertebral strength in a cranial to caudal direction going from TI to L5 (Singer et al, 1995). Brinckmann et al. showed that strength correlated to cross-sectional area, and that both increased with caudal descent through the lumbar spine (Brinckmann et al, 1989).  1.2.3 1.2.3.1  Implant Factors Area of Bone-Implant Interface  Intuitively, it seems reasonable to believe that the greater the contact surface area of the implant the larger the stress distribution across the bone-implant interface. It has been demonstrated that a minimum of 30-40% implant contact area must be available to avoid subsidence into trabecular bone (Closkey et al, 1993; Pearcy and Evans, 2003).  Gill  suggested that for successful interbody spinal fusion, between 50% and 80% of the vertebral body cross-sectional area (CSA) should be covered by graft (Gill, 1989). Brinckmann et al. found that the strength of the vertebral body increases as the C S A of the vertebral endplate increases (Brinckmann et al, 1989). Hasegawa et al. examined the failure load of titanium mesh cages, with varying degrees of implant surface area in contact with the endplate, on L I and L5 human cadavers (Hasegawa et al, 2001). They found that the C S A of the bone-implant interface significantly affects the failure load (Hasegawa et al, 2001). These findings are in contrast to a biomechanical analysis, in which the vertebral body cross-sectional area contacting the implant was not found to have an independent effect on force to failure of either an implant-endplate or implanttrabecular bone interface (Hollowell et al, 1996).  16  1.2.3.2 Implant Design Design of the implant may affect the strength of the bone-implant interface. For the same reason that interbody fusion devices should be positioned at the strongest aspect of the endplate, so should interbody devices be designed so that load can be distributed through the strongest aspects of the bone interface. It has been hypothesised that a perfect surface match of the implant face with the underlying bony endplate would result in a more uniform stress distribution (Steffen et al, 2000a). As the ring apophysis of the endplate is thought to be the strongest region, the bone-implant interface should intuitively be best located there. It has been suggested that an implant should contact the periphery of the endplate (Kozak et al, 1994; Kumar and Doherty, 1994; Steffen et al, 2000b; Steffen et al, 2000a).  A n implant resting solely on the peripheral endplate offers compressive  strength similar to that of an implant with contact across the entire endplate.(Steffen et al, 2000a)  Edwards et al, studied endplate structure and determined that the central  aspect is more porous surrounded about the perimeter by a dense annular rim (Edwards et al, 2001) However, in biomechanical studies, implant type or design had no significant effect on the measured failure loads (Hoshijima et al, 1997; Jost et al, 1998; Knop et al, 2001; Steffen et al, 2000a). Jost et al. and Hoshijima et al. compared implants designed for a PLIF or TLIF procedure, while Knop et al. and Steffen et al. tested cages required for anterior column reconstruction inserted during an open anterior approach. None of the implants tested were specifically designed to capitalise on the regional variation in the endplate's strength profile.  1.3  Summary  It has been definitively shown that B M D has a significant correlation with vertebral strength and that interbody fusion devices are at an increased risk of subsiding into the bone interface following reconstruction of a spine with poor bone quality (i.e. low B M D ) . In the osteoporotic spine, little can be done to influence this "patient factor " at the time of anterior column stabilization.  However, "surgical and implant factors" can be  optimized to minimize the risk of subsidence. We believe that the "surgical and implant factors" should be optimized to reflect the endplate strength profile as defined by Grant et al. and Kumar et al. (Grant et al, 2001; Kumar and Doherty, 1994). This would translate  17  into a stronger bone-implant interface, therefore reducing the risk of implant subsidence following anterior column stabilization.  The strength profile of the thoracolumbar spine has not been determined as it has for the lower lumbar and sacral endplates.  As well, although general trends exist regarding  regional variation in vertebral density, little work has been performed mapping the thoracolumbar endplate r B M D profile or correlating it with the endplate strength profile. The majority of this work has focused exclusively on the lumbar or lumbosacral spine (Edwards et al, 1986; Lu et al, 2000; Nepper-Rasmussen and Mosekilde, 1989), or has identified a small number of sectors to describe general trends in r B M D (Banse et al, 2001; Edwards et al, 1986; Simpson et al, 2001).  Thoracolumbar spine reconstruction is commonly performed for tumour, infection, trauma, and deformity. In the majority of these situations, i f anterior column stabilization is required, it is performed through an anterior open approach. For this approach, implants are larger and it is technically easier to manipulate the position of interbody devices. Therefore, the opportunity to capitalise on "surgical and implant factors" is greater than when performing a PLIF, TLIF, or ALIF. A better understanding of the thoracolumbar endplate's strength profile and r B M D profile is required in order to improve the strength of the bone-implant interface by manipulating "surgical and implant factors".  18  1.4  Purpose  1. To determine the strength profile of the endplates in the thoracolumbar spine at levels T9, T12, andL2.  2. To determine the regional bone mineral density profile of the trabecular bone immediately adjacent to the subchondral endplate in the thoracolumbar spine as represented by T9, T12, and L2.  3. To compare the thoracolumbar endplate strength profile with the regional bone mineral density profile. 1.5  Hypothesis  We hypothesize that for T9, T12, and L2, the periphery of the endplate will be stronger than the centre. We expect that r B M D will correlate with local indenter failure load, and that the endplate strength profile will be similar to the endplate r B M D profile.  We believe that the mechanical and structural properties of the endplate will vary with the changing sagittal alignment and anatomy of the spine. The posterior aspect of the endplate, between L3-S1, was significantly stronger than the anterior aspect of the endplate (Grant et al 2000). This could be due to the lordosis of the lower lumbar spine in that a greater load is seen across the posterior aspect of the endplate than the anterior aspect, increasing bone density and strength relative to the anterior aspect of the endplate in accordance with Wolffs Law.  Therefore the mechanical and structural properties of  L2 should be similar to those of L3-S1; that is the posterior endplate strength should be greater than the anterior endplate strength. Likewise, the posterior aspect of the L2 subendplate zone should be of higher density than the anterior aspect.  In keeping with this line of reasoning, we hypothesize that because of the thoracic kyphosis, the anterior endplate strength and density of T9 will be greater than the  19  posterior endplate strength or density. As well, for T 1 2 , the profiles of the posterior and anterior endplate will be equal.  2 0  2.0 Materials and Methods 2.1  Specimen Selection  Seven fresh-frozen specimens were obtained from the University of British Columbia Anatomy department through the Body Donation Program. Specimen attributes are listed in the table below, including age and sex (Table 2.1). The mean age of the specimens was 78.8 years. Four specimens were complete spines spanning from vertebral body C l to S5. Three specimens were segments of a spine spanning from vertebral bodies of T9 to S5. A l l specimens were stored at minus 20 degrees Celsius in double plastic bags. These specimens were dethawed and refrozen three separate times; for sectioning and preparation, bone mineral density determination, and finally mechanical testing. During all times that the spines were dethawed they were kept moist with saline spray. Table 2.1: Specimen attributes  Specimen  Level  Gender  Age  Race  1061  T9, T 1 2 , L2  female  73  1067  T 9 . T 1 2 , L2  female  1069  T9, T 1 2 , L2  1081  2.2  Caucasian  Height (cm) 157  Weight (kg) 63  88  Caucasian  152  60  male  67  Caucasian  180  120  T 9 . T 1 2 , L2  female  unknown  172  85  1083  T9, L2  female  88 84  unknown  152  48  1087  T9, T 1 2 , L2  female  75  unknown  168  93  1090  T12  female  73  unknown  unknown unknown  Specimen Preparation  During the sectioning and preparation phase, the spines were stripped of all muscles and soft tissue apart from the interspinous ligaments, capsules, and discs. The four specimens that included the upper thoracic and cervical spines were transected between the disc space of T8 - T9. A l l spines were transected between the L3 - L4 disc. This created seven specimen spinal segments including the vertebral bodies of T9 to L2. During the transection, care was taken to divide the disc evenly between the superior and inferior vertebral bodies. The disc was transected cleanly with minimal trauma and the endplate was not damaged. After D X A scans were performed, the T9, T12, and L2 vertebral bodies were separated using the same transection technique. Using a variety of scalpels,  21  disc and cartilage was carefully removed from the superior and inferior endplates without damaging the endplate  2.3  Dual Energy X-ray Absorptiometry (DXA)  A l l specimens, T9-L2, underwent dual energy x-ray absorptiometry to determine the bone mineral content (BMC).  A large bag of rice, approximately seven centimetres  thick was used to simulate the density of removed soft tissue. Lateral D X A scans were performed for the T9, T12, and L2 vertebral bodies. Lateral D X A scans were selected rather than A P D X A scans to limit the error in B M C values associated with A P D X A scanning. A P D X A scans include the posterior elements, which artificially raise B M C values. The technician defined the area of the vertebral body for which the mean B M C was to be determined. Care was taken not to include the peripheral vertebral osteophytes when defining the area to be included in analysis, as this would artificially increase the B M C value as well. This defined area was also used to estimate volume and calculate the BMD.  2.4  Peripheral Quantitative Computed Tomography (pQCT)  The T9, T12, and L2 vertebral bodies had regional bone mineral densities (rBMD) determined using the STRATEC XCT-2003 pact (Norland Medical Systems Inc, Fort Atkinson, WI, USA). A calibration phantom was used before scanning to serve as a bone mineral analogue. Each vertebral body was positioned within a custom made plastic tube (Figure 2.1). first.  The body was oriented such that the inferior endplate was always scanned  The spinous process and lamina was secured in plasticine so that the anterior  vertebral body wall was directed toward the ceiling. The pQCT emitted a laser reference line onto the anterior vertebral body. Care was taken to ensure that there was no side to side rotation or tilting of the endplate relative to this reference line. This ensured that each CT cut was as parallel to the endplate as possible.  22  Figure 2.1: p Q C T with custom-made tube suspending the vertebrae within the scanning field.  Images were acquired at a speed of 30mm/second and resolution of 0.25 by 0.25 voxels. Each image was 2.4 mm thick. Two masks (groups of CT cuts) were performed of each vertebral body. Each mask was initiated at a predetermined percentage of the entire vertebral height referenced from the inferior endplate. The vertebral height was based on a measurement from the inferior to superior aspect of the anterior vertebral wall. The first mask consisted of five axial CT cuts of the inferior endplate. The first axial cut began at a distance equal to five percent of the entire vertebral height from the most anteriorinferior corner of the vertebral cortex. A distance of 2.5% of the entire vertebral height thereafter separated each cut. Therefore, mask one consisted of five cuts, at 5%, 7.5%, 10%, 12.5%) and 15% of the entire vertebral height, respectively. The second mask consisted of five cuts of the superior endplate, beginning at eighty-five percent of the entire height, and included cuts, at 85%, 87.5%, 90%, 92.5%, and 95% of the entire vertebral height. Each mask was designed so that one of the cuts would capture the entire extent of cancellous bone immediately below the vertebral endplate.  This area was  23  referred to as the sub-endplate zone. The first few cuts invariably would be devoid of bone in the middle aspect of the endplate because of its concave shape (Figure 2.2).  Figure 2.2: M a s k of five axial cuts of the inferior endplate. Each image is numbered in order. In this instance the 5 image was selected as that which first complete captured the trabecular bone within the sub-endplate zone, without including the dense vertebral endplate. t h  2.5  Analysis of Regional Bone Mineral Density  Regional bone mineral density was determined using the software program Bonalyse (Bone and soft tissue analysis software for pQCT and QCT images, Finland) on a 17 inch monitor (Hewlett Packard - 800X600 pixels). The first C T cut which entirely was through the cancellous bone juxtaposed to the endplate (sub-endplate zone) was selected from the five cuts in both masks. Thirty-six CT cuts were hence selected, including CT cuts of cancellous sub-endplate bone juxtaposed to eighteen superior and eighteen inferior endplates. Using the Bonalyse program, these C T cuts were then divided into squares with equal area. Each square had the following dimension: 20% of the endplate's maximum A P dimension, by 15% of the endplate's maximum lateral dimension. This created a grid system of squares, with each square representing a regional bone mineral  24  density (Figure 2.3). Each square within the rectangular grid was constructed in reference to the centre point of the A P and lateral dimensions of the endplate.  A consistent measurement technique was used to ensure the grid was precise and reproducible. A measurement could be performed between two points selected by the cursor. A l l positional measurements were performed at 200% zoom factor to ensure  Figure 2.3: G r i d system of squares, w i t h each square representing a specific r B M D . T h e c e n t r a l square is h i g h l i g h t e d . A l l o t h e r squares a r e referenced relative t o t h e c e n t r a l square.  accuracy. The A P measurement was started at the inner cortex of the posterior vertebral wall (adjacent to the vertebral foramen i f present, or otherwise based at the apex of the posterior wall concavity). The cursor was then extended in a straight line to the inner surface of the cortex of the anterior vertebral body. This represented the start and finish points of the A P measurement, respectively. Each pixel producing the CT image was represented by an X and Y coordinate. The same X coordinate that was registered at the starting point of the A P measurement was selected at the finishing point to ensure a straight line. The difference between the starting Y co-ordinate and the finishing Y coordinate, represented the A P measurement. The A P measurement was then divided into five equal rows, each with a height of 20% of the total A P dimension (Figure 2.4).  25  F i g u r e 2.4: Regional B M D g r i d w i t h 5 rows. Each r o w representing 2 0 % o f t h e endplate A P dimension.  The lateral dimension was measured from the inner cortex of one side to the inner cortex of the opposite side. The Y coordinate corresponding to 20% of the A P dimension going from posterior to anterior referenced this measurement. This was typically the maximum lateral dimension of the endplate. The difference between the starting and finishing X coordinates represented the lateral dimension. The lateral dimension was then divided into seven equal columns, each with a width of 15% of the total lateral measurement (Figure 2.5). 15%  Fi g u r e 2.5 Regional B M D g r i d w i t h 7 columns. Each c o l u m n represents 1 5 % o f the endplate l a t e r a l dimension.  The Bonalyse program determined the bone mineral density of each square. Based on the grid system described, a map of regional bone mineral densities for each endplate was created. The grid was labelled in reference to the central region (or square). The row within which the central region was positioned was labelled 0. The two rows anterior and two rows posterior were labelled +1, +2, and -1, -2, respectively. The column within which the central region was positioned was labelled 0. The three columns to the right and the three columns to the left were labelled +1, +2, +3, and -1, -2, -3, respectively.  26  Every region (square) was labelled with respect to the row and column within which it was positioned. For example, a region in the + l row and - 2 st  region (+1, -2).  nd  column was identified as  Using this system, the same region could be compared between  endplates (Figure 2.6). +3  +2 +1  0  -1  -2  -3  F i g u r e 2.6: r B M D g r i d w i t h labeled columns a n d rows relative to t h e c e n t r a l r o w ( p i n k ) a n d c o l u m n (blue). Region (+1,-2) is shaded black.  Due to the shape of the thoracolumbar vertebral endplate, not all regions defined within the grid which overlay the CT cut had bone within their boundaries.  To facilitate a  complete comparison, only the regions which contained cancellous bone, consistently for all of the selected CT cuts, were included for bone mineral density analysis. Thus, row 1 was the only row that included regions from all seven columns. The - 2 , 0, and +1 rows excluded the regions in column +3 and - 3 . The most anterior row (row +2) included only the regions in columns +1, 0, -1 (Figure 2.7). -3  -2  -1  0  +1  +2  +3  +2 +1  /  0  •  -2  warn  p  J  F i g u r e 2.7: Areas shaded blue w i t h i n the r B M D g r i d represent the regions f o r w h i c h B M D was d e t e r m i n e d .  Occasionally, a region on the periphery of the CT cut would include the cortex or overhanging osteophytes.  In these circumstances, the area was decreased until only  cancellous bone was contained within the region of interest by moving the boundary until  27  the cortical bone was excluded. In all cases the decrease in area was only by a few percent. This would ensure that the region's bone mineral density was representative of only trabecular bone. The same technique was consistently used when decreasing the area of a square. Any change in dimension was performed at a zoom factor of 200% to minimize the chance of over correction. The dimension that was decreasing depended upon which row the square was positioned. If the square was in row +2 or - 2 then the AP dimension was decreased while maintaining a constant lateral dimension. If the square was positioned in the +1, 0, or -1 row, then the lateral dimension was decreased and the A P dimension remained unchanged. The tip of the measuring cursor would turn green when it overlay cortical bone. This software feature of the Bonalyse program accurately ensured that cortical bone was not included in the bone mineral density.  To ensure the reliability of the r B M D measurement technique the same endplate was analysed ten times on separate occasions.  For each of the twenty-five regions the  coefficient of variation (standard deviation/mean x 100) was determined as a measure of reliability. The r B M D measurement technique proved to be very reliable. The average coefficient of variation was 0.41%.  A l l B M D measurements were determined at the following parameter in the Bonalyse software: Weight type = unit weight Search Rules for Bone = do not need to "find bone for analysis" Bone Analysis limits = -40 to 2500 gm/cm3 Bone Marrow Analysis limit = -500 gm/cm3 2.6  Indentation Testing  Indentation testing was performed using a low-force servohydraulic-testing machine (Dynamite, Instron Corporation, M A ) . A custom made indenter was secured to the linear actuator of the machine. A load cell on the actuator registered all forces in Newtons. Using the software programs Wave editor and Wave runner (Instron Corporation, M A ) each indentation test was designed and performed by exactly the same protocol. The indenter was manually placed in the starting position just micrometers above the endplate  28  at the indentation site. In no case did the indenter exert pressure on the endplate prior to the indentation test.  From the starting position, the indenter was depressed into the  endplate for a distance of 3mm, at a rate of 0.2mm per second. This depth ensured that the bone beneath the indenter failed mechanically. Upon reaching the maximum depth of 3mm, the indenter was elevated out of the endplate at a rate of 5mm per second. At no time did macroscopic damage occur to the endplate while the indenter was elevated out of the endplate. This indentation protocol has been previously used successfully in our laboratory (Grant et al, 2001).  2.6.1 Indenter The indenter was hemispherical in shape. This shape was chosen to provide a uniform contact profile, and to avoid edge contact between the indenter and a sloped aspect of the endplate (Grant et al, 2001). The indenter had a diameter of 3mm to ensure adequate trabecular coverage of test sites, while at the same time maximizing the number of test sites per endplate (Grant et al, 2001).  Figure 2.8: 3mm hemispherical indenter.  2.6.2  Test Form at  Indentation tests were performed in a rectangular grid system similar to that used for the analysis of regional bone mineral densities. The location of an indentation test was defined by the percentage of the endplate's A P and lateral dimensions.  The A P  29  dimension was measured from the edge of the posterior endplate adjacent to the vertebral foramen i f present, or otherwise at the centre of posterior longitudinal ligament, to the centre of the anterior edge of the endplate.  The lateral dimension represented the  maximum width of the endplate, measured perpendicular to the A P dimension. Care was taken to not include peripheral osteophytes in the measurements of either dimension. The A P and lateral dimensions were divided into increments of 20% and 15% of the maximum dimensions, respectively. Therefore, there were 5 rows and 7 columns of possible indentation sites. The centre column and centre row was defined as the 0% A P dimension and 0% lateral dimension. The two rows anterior and two rows posterior to 0% A P dimension, were defined +20%, +40%, -20%, and -40% A P dimension, respectively. The three columns to the right of the 0% lateral dimension were defined as the +15%), +30%), and +45% lateral dimension, while the three columns to the left were defined as -15%, -30%, and -45% lateral dimension (Fig. 2.9). <  100% L A T  -45% -30% -15% 0% 15% 30%  ». 45%  Figure 2.9: Rectangular grid system used to perform indentation tests.  This grid system allowed the maximum number of indentation tests to be performed on an endplate. Based on the dimensions of the smallest endplate, a minimum of 1.5mm was still present between the outside diameter of the indentations. This was sufficient to prevent interaction between indentation sites (Grant et al, 2001).  In no case did a  fracture propagate into an adjacent test area. Trabecular bone failed only beneath the indenter and not adjacent to it. As with the grid system developed for the regional bone mineral densities, this grid system would allow a direct comparison of the same relative positions between endplates and with regional B M D . For example, the indentation test  30  performed in the +15% lateral dimension and +20% A P dimension corresponded to the regional bone mineral density at row +1, column +1. 2.6.3  Indentation procedure  Each vertebral body was first secured within a custom made holding jig (Fig 2.10) The vertebral body was centred within the holding device and then secured at its anterior and lateral sides by a minimum of three, and maximum of five screws per side. The screws outer diameter was 4.8 mm and the ends were blunted so that they did not penetrate the outer vertebral cortex. This ensured that a penetrating screw did not support cancellous bone, which could artificially increase the failure load. The vertebral body was centred such that the posterior vertebral bodyline was parallel to the posterior border of the holding device (Figure 2.10).  Figure 2.10: Custom made jig used to secure vertebral body during indentation testing.  The holding device was then mounted onto an XY-translating table by three legs (Figure 2.11). The height at which the holding device was secured to each of the three legs could be individually adjusted. Using a tripod bubble level, the holding device was adjusted such that the endplate was level to the floor.  The translating table was then secured  beneath the testing machine (Figure 2.11). Periodically during the testing process checks were performed to ensure that there was no play between the vertebral body and holding device, between the holding device and translating table, or between the translating table and testing machine.  31  By use of the translating table, the indenter was centred on the endplate relative to the endplate's lateral and A P dimensions. The centre of the lateral dimension corresponded to centre of the posterior longitudinal ligament.  The centre of the A P dimension  corresponded to the mid-point of the A P measurement as defined above. The translating table allowed incremental adjustments of 0.25mm, to ensure precise positioning of the indenter along the overlying grid. The first indentation was performed at the centre of the grid (row 0%, column 0%) for all specimens. The indenter was then translated laterally, along the same row, randomly in either a negative or positive direction.  When all  possible indentations were performed in row 0%, the indenter was moved to either the adjacent anterior or posterior row. Indentations were then completed on this row moving laterally from one side to the other. The rest of the rows were then completed in this same manner in a random fashion.  Figure 2.11: The indentation setup consisting of the X Y translating table, custom made holding jig, and instron.  Each indentation test was recorded on the PC by the wave runner program. The indenter depth and force was sampled at a rate of 100 hertz. A total of 24 to 32 indentation tests per endplate were performed and recorded. Data from each test was then transferred to a Window's Excel worksheet. This software generated a load-displacement curve of force in newtons versus displacement in millimetres (Figure 2.12).  32  2.7  Analysis  2.7.1 Curve Analysis Each load-displacement curve was manually analyzed to determine the failure load. The failure load was determined when the slope of the load-displacement curve decreased for the first time (Figure 2.12). Often small blips would occur in the load-displacement curve but the overall slope of the curve on either side of the blip would remain positive. In these instances, the blips would not be recognised as the failure point.  Specimen 1061 - Test 12  Displacement (mm)  Figure 2.12: Indentation test load-displacement curve. 126.09 N .  Failure point is identified to be  2.7.2 Data Analysis: Data from both the indentation tests as well as the pQCT regional B M D were analysed using the same method. The test value, whether it represented an indentation test or QCT r B M D reading, was dependent on the specimen from which it came, the endplate from which it was located (superior vs. inferior), and its location on that endplate as defined by the grid system. Because there were many specimens tested, Analysis of Variance ( A N O V A ) was utilized for statistical analysis. As there were a minimum of 24 tests performed per endplate, a repeated measure A N O V A was required. Therefore, the test value (force or B M D ) for a specific grid position was compared to all the other test values from that same relative grid location on all the other specimens. This comparison  33  was performed for each of the test locations on the grid, using repeated measures ANOVA.  Within the A N O V A , the test location was defined by the A P and lateral  position. The endplate (superior versus inferior) and spinal level also related each test position. This defined endplate and spinal level as independent variables within the A N O V A . Therefore, a repeated measure, three-way, multivariate A N O V A was required for data analysis of the indentation tests and the r B M D results. The statistical software package Statistica (StatSoftn Inc, 1997) was used for all A N O V A statistical analysis.  The shape of the endplates varied both within a spinal level (i.e. T9) and between levels (T9 vs. L2). For this reason, the number of tests (rBMD or indentation test) performed also varied between endplates.  For example, for an endplate that was very wide  anteriorly, 7 indentation tests could be performed in row +20%, while on an endplate which had an average anterior width, only 5 indentation tests could be performed.  In  order to perform the A N O V A , the number of indentation tests had to be consistent across all endplates. Therefore, to facilitate the analysis, an area was defined to maximise the number of tests consistently performed across all endplates (Grant et al, 2001). This was . called the "Maximum profile analysis". It consisted of 4 rows (-40%, -20%, 0%, and +20%) for indentation testing or row +1, 0, -1, and -2 for rBMD) and 5 columns (+30%, +15%, 0%>, -15%, and -30% for indentation testing or column +2, +1, 0, - 1 , and -2 for rBMD) (Figure 2.13).  A deficiency of the Maximum profile analysis is that the most anterior and lateral tests were not included. Therefore, to ensure we characterized the maximum expansion of the endplate in both an anterior-posterior and lateral direction, two further areas were defined for analysis; the "Anterior-Posterior profile analysis" and the "Lateral profile analysis". The Anterior-Posterior profile analysis extended anteriorly to row +40% (row +2 for r B M D analysis), however, the area was limited in width by three columns. Thus, the area defined by the Anterior-Posterior profile analysis included all 5 rows of the grid and the 3 central columns (Figure 2.13).  34  Maximum profile analysis  Anterior-Posterior profile analysis  Lateral profile analysis  Figure 2.13: Maximum, Anterior-Posterior, and Lateral profile analyses.  The Lateral profile analysis extended the maximum width of the endplate, incorporating all seven grid positions in the lateral excursion. The lateral profile analysis only included values from row -20% (or row -2 for r B M D analysis) as this was the only row in which all seven tests was performed consistently for all specimens (Figure 2.13).  In summary, three separate repeated-measures, multivariate A N O V A s were performed to characterize the thoracolumbar endplate's strength profile: the Maximum profile analysis, the Anterior-Posterior profile analysis, and the Lateral profile analysis. Similarly, three repeated-measures,  multivariate  ANOVAs  were  performed  to  characterize  the  thoracolumbar endplate's regional bone mineral density profile. For each of these six analyses, the A N O V A provided an explanation of the endplate profile in relation to an AP position and a lateral position, as well as define any difference in the profile between endplate and spinal level.  Indentation sites along the peripheral aspect of the endplate frequently contacted the apophyseal ring. A further A N O V A was performed on indentation sites, which were found only within the ring. This analysis was performed to determine whether the A P and lateral position effect on the endplate's strength profile occurred despite any influence of the cortical ring. Hence a Cancellous Anterior-Posterior profile analysis was performed (Figure 2.14). A separate analysis for Lateral position was not performed as it only entailed the removal of position "-20%,+45%" and "-20%,-45%" from the endplate map in the Lateral profile analysis.  35  -45%-30%-1.5%  0%  15%  30%  45%  Cancellous Anterior-Posterior profile analysis Figure 2.14: Cancellous Anterior-Posterior profile.  To further investigate the effect of spinal level on the endplate profile, a univariate A N O V A was performed to compare the ratio of the mean strength of the entire posterior row with the mean strength o f the entire anterior row. This analysis was performed primarily because the M a x i m u m profile analysis and Anterior-Posterior profile analysis could not adequately investigate the hypothesis of whether spinal level and sagittal contour influenced the endplate's strength profile.  The effect of spinal level on D X A B M C and D X A B M D was also investigated using a univariate A N O V A .  Endplate r B M D profile maps were compared to the corresponding endplate strength profile maps. Pearson R coefficient was determined to evaluate the correlation between strength and r B M D for the M a x i m u m , Anterior-Posterior, Lateral, Cancellous AnteriorPosterior, and analyses. For each endplate the failure force for the indentation tests was correlated to the respective r B M D result.  36  3.0 Results 3.1  Endplate Strength Profiles  3.1.1 Maximum Profile Analysis 3.1.1.1  Effect of A P Position  The relative position of the indentation sites in the A P direction had a significant effect on the endplate strength profile in the Maximum profile analysis (p<0.0001). The mean failure load decreased between row -40% and row 0%, followed by an increase of row +20% over row 0% (Figure 3.1).  The Newman-Keuls post hoc comparison of means  revealed that each row was significantly different than the other three (p<0.046). 95  -40%  -20%  0%  20%  A P position Figure 3.1: Effect of A P position on the endplate strength profile in the M a x i m u m profile analysis. F=72.68; p<0.0001  3.1.1.2 Effect of Lateral Position Lateral position had a significant effect on the endplate strength profile in the Maximum profile analysis (pO.0001). Overall, there was an increase in the endplate's mean strength from the centre (column 0%) toward the periphery (columns +30% and - 30%) (Figure 3.2). In the post hoc comparison of means, the strength of the three central columns (columns 0%, +15%, and -15%) were not significantly different from each other, but were significantly weaker than the lateral columns (column +30% and -30%) (p<0.043). There was no difference between the two lateral columns +30% and -30%.  37  85  50  i  :  :  :  :  30%  15%  0%  -15%  L_  -30%  Lateral position  Figure 3.2: Effect of lateral position on the endplate strength profile in the Maximum profile analysis. F=7.09; p<0.0001  3.1.1.3 Effect of the Interaction between A P and Lateral Position The interaction between A P and lateral position had a significant effect on the endplate strength profile in the Maximum profile analysis (p<0.0001). A similar endplate strength profile as seen in the overall lateral position analysis above (Figure 3.2) existed for rows -40% and +20%, but was not seen in rows -20%, and 0% (Figure 3.3).  In row -40%,  Newman-Keuls analysis revealed that the mean strength showed a significant increase from the centre to the lateral indentation sites (p=0.05). The central site was significantly weaker than position "-40%,+15%" and "-40%,-15%", which was significantly weaker than the most lateral sites ("-40%,+30%"; "-40%,-30%"). In rows -20% and 0%, there was no difference in mean strength between indentation sites relative to lateral position. For row +20%, the two lateral indentation sites ("+20%,+30%"; "+20%,-30%") had a greater mean strength than the three central sites and all sites in row 0% (p=0.07 0.00003).  The strongest mean failure loads in the Maximum profile analysis were  located at the posterolateral indentation sites ("-40%,+30%"; "-40%,-30%") (pO.OOOl). A l l five columns had a similar strength profile relative to AP position.  38  30%  15%  0%  -15%  -30%  Lateral Position  -40%  -20%  0%  20%  AP position Figure 3.3: The effect of the interaction between AP and lateral position on the endplate strength profile in the Maximum profile analysis. F=7.31; p<0.0001.  3.1.1.4 Effect of Spinal Level Within the Maximum profile analysis, there was no significant difference in endplate strength between level (p<0.29), although, L2 had a somewhat greater mean strength than T9 or T12 (Figure 3.4).  39  85 80  Level  Figure 3.4: Effect of spinal level on the endplate's mean strength in the M a x i m u m profile analysis. F=1.31; p=0.29  3.1.1.5 Effect of the Interaction between AP Position and Spinal Level The A N O V A for the Maximum profile analysis revealed that the A P position and spinal level interacted to produce a significant effect on the endplate strength profile (p<0.007). A similar endplate strength profile as seen in the overall analysis for AP position (Figure 3.1) was seen for all levels (Figure 3.5). Post hoc comparison of means showed that for all three levels, the posterior row (-40%) was significantly stronger that the three anterior rows (pO.OOl). The mean strength of the L2 posterior row (-40%) was significantly greater than that for T12 and T9 (p<0.0004), and T9 was significantly greater than that for T12 (p<0.043). In the two most anterior rows of this analysis (row 0% and +20%) there was no significant difference in strength between indentation sites, except for row 0% at level T12, which had a lower mean strength than row +20% at level L2 and T9 (p<0.04). Therefore the mean strength of L2 was significantly greater then T12 and T9 in the posterior aspects of the endplate but not in the anterior aspect.  40  3.1.1.6 Effect of the Interaction between Lateral Position and Spinal Level The A N O V A examining the effect the interaction between lateral position and spinal level did not prove to be significant for the Maximum profile analysis (Figure 3.6). For all three levels, the mean strength decreased from the periphery to the centre of the endplate. This trend was not found to be statistically significant for level T12 or L2. For the T9 level, only the mean strength of column -30% was significantly stronger than the centre of the endplate (column 0%). 90 -o85  -o{  80 75  z O O LL  T9  j  T12  •~o - L2  j  "  "S  'o~...  70  ^  •/  f-  65 60  _.. _ _ - 6- -""  55 50  30%  15%  0%  -15%  -30%  Lateral position Figure 3.6: The effect of the interaction between spinal level and lateral position on the endplate strength profile in the M a x i m u m analysis profile. F=0.6; p<0.78.  41  3.1.1.7 Effect of Endplate No significant difference was detected between the mean strength of the superior endplate and that for the inferior endplate in the Maximum profile analysis (p<0.18). However, a trend existed, in that the overall strength of the inferior endplate was slightly greater than that for the superior endplate (Figure 3.7). 80 75 70 CD  o o  65 60 55  Superior  Inferior Endplate  Figure 3.7: Effect of endplate on the endplate's mean strength in the M a x i m u m profde analysis. F=1.94; p<0.18.  3.1.1.8 Effect of the Interaction between AP Position and Endplate No significant interaction existed between A P position and endplate for the endplate strength profile in the Maximum profile analysis (p<0.48). Both the superior and inferior endplates had a similar strength profile (Figure 3.8) to the profile seen for the overall A P position in the Maximum profile analysis (Figure 3.1). There was no difference in the shape of the strength profile between the superior and inferior endplates.  42  110 ioo  i  j  I-  _  <  >  —  Superior  A P position  Figure 3.8: 2-way interaction between A P position and endplate on the endplate strength profile for the Maximum profile analysis. F=0.84; p< 0.48.  3.1.1.9 Effect of the Interaction between Lateral position and Endplate A significant interaction existed between lateral position and endplate for the Maximum profile analysis, resulting in a different shape between the superior and inferior endplate strength profile maps (p<0.006).  The Newman-Keuls post hoc comparison of means  revealed no difference with respect to lateral position on the superior endplate, however the inferior endplate strength profile (Figure 3.9) was similar to that in Figure 3.2. The mean strength of the lateral indentation sites on the inferior endplate (column+30% and 30%) were significantly greater than were those in the centre (column +15%, -15% and 0%) (p=0.05). This was similar to the trend seen in the endplate strength profile for overall lateral position (Figure 3.2).  43  30%  15%  0%  -15%  -30%  Lateral position  Figure 3.9: Effect of interaction between lateral position and endplate on the endplate strength profile in the Maximum profile analysis. F=3.85; p=0.006  3.1.1.10 Effect of the Interaction between Endplate and Level There was no significant interaction between spinal level and endplate on the strength profile for the Maximum profile analysis (p<0.98). Although not significant, the mean strength of the inferior endplate was greater than that for the superior endplate for all three levels (Figure. 3.10). This trend was consistent with that seen independently for both spinal level and endplate in the Maximum profile analysis (Figures. 3.4 and 3.7). 85, 80  ,.-o  !  75 70 " o  V  '  ^  ^  ^  ^  65 •  ^  ^  '  ^  ' "  ^  60  -o- T12  55 50  - o - T9  ........ L2 Superior  Inferior Endplate  Figure 3.10: Effect of the interaction between endplate and spinal level had on the endplate strength profile in the Maximum profile analysis F=0.02; p<0.98.  44  3.1.2  Anterior-Posterior Profile Analysis  3.1.2.1 Effect of APPosition AP position had a significant effect on the endplate strength profile for the AnteriorPosterior profile analysis (p <0.0001). The overall effect was a decrease in the mean strength from the posterior aspect of the endplate to the centre of the endplate, with a substantial increase in the mean strength of the most anterior row (Figure 3.11). The Newman-Keuls post-hoc analysis showed that the most posterior row (row -40%) had a significantly greater mean failure strength than did the row directly anterior to it (row 20%>) (p<0.002). The mean strength for both row -40% and row -20% were significantly greater than the two central rows mean strength, row 0% and row +20% (p<0.003). The Newman-Keuls post-hoc test found no significant difference between the mean failure loads for row 0%> and row +20%. The most anterior row, row +40%o, had the greatest mean strength, being significantly stronger than all other rows (p<0.0002).  -40%  -20%  0%  20%  40%  AP position  Figure 3.11: Effect of AP position on the endplate strength profile in the Anterior-Posterior profile analysis. F=37.87; p>0.0001  3.1.2.2 Effect of the Interaction between AP and Lateral Position A P and lateral position interacted to create a significant influence on the endplate strength profile for the Anterior-Posterior profile analysis (p<0.02). In all three columns (columns 0%, +15%, and -15%>) there was a decrease in the mean strength of the indentation tests moving from the posterior aspect of the endplate to the centre of the endplate (Figure 3.12). In the most anterior row (row +40%>), the mean strength for the  45  indentation sites in all three columns significantly increased over the mean strength of the central rows (rows 0% and +20%). The endplate strength profile created by AP position for all three columns (columns 0%; +/- 15%) is a similar profile to that seen for the overall AP position in the Anterior-Posterior profile analysis (Figure 3.11). The post-hoc analysis showed a significant side to side difference in row —40% (p<0.04). The mean strength for the indentation sites at position "-40%,+15%" and "-^10%,-15%", were significantly stronger than that for the central site at ' M 0 % , 0 % " .  However, the  Newman-Keuls test found no side to side difference in any other rows except the most anterior row, row +40%.  Here, as seen in Figure 3.12, the mean strength for the  indentation site "40%,+15%" was significantly greater than that for the other two positions in the same row (p<0.0004). 110 p—  •  •  •  .  .  ,  AP position Figure 3.12: Two-way interaction between AP and lateral position on the endplate strength profile in the Anterior-Posterior profile analysis. F=2.27; p<0.02.  46  3.1.2.3 Effectof Spinal Level No difference existed in the endplate's mean strength between spinal level (p<0.56). As was the case in the Maximum profile analysis, the mean strength of L2 tended to be greater than the mean strength of T9 and T12 (Figure 3.13). 85 80 75 0) 70 o o  LL  65  55  T9  T12  L2  LEVEL  Figure 3.13: Effect of spinal level on the endplate's mean strength in the Anterior-Posterior profile analysis. F=0.58; p<0.56.  3.1.2.4 Effect of the Interaction between AP Position and Spinal Level There was no significant difference in the endplate strength profile relative to AP position between spinal level in the Anterior-Posterior profile analysis (Figure 3.14). Each level's endplate strength profile was similar to the strength profile found for the overall A P position (Figure 3.11). The mean indentation strength of row -40% for L2 tended to be greater than for T12 and T9, while in the row +40%, the opposite trend was true, the mean strength of T9 was greater than T12 and L2. Furthermore, for L2, no significant difference in mean strength was found between row +40%> and row -40%o, while for T9 row +40%> was significantly stronger than row -40%.  47  3.1.2.5 Effect of Endplate A n A N O V A was performed to determine i f a difference existed between the mean strength of the superior endplate versus that of the inferior endplate.  No significant  difference between the endplate's mean strength was found (Figure 3.15); the mean strength of the inferior endplate was slightly greater than the superior endplate's (p<0.38). 80,  :  ;  ,  55 l  i Superior  i  1  Inferior ENDPLATE  Figure 3.15: Effect of endplate on the endplate's mean strength for the Anterior-Posterior profile analysis. F=0.79; p<0.38.  48  3.1.2.6 Effect of the Interaction between A P Position and Endplate A n A N O V A investigating the interaction between A P position and endplate revealed no significant effect.  The strength profile for the A P position was very similar between  superior and inferior endplates (Figure 3.16). Both endplate strength profiles closely resembled the general profile seen for the Anterior-Posterior profile analysis (Figure 3.11).  -°—  Superior  -o-  Inferior  0)  20%  40%  AP position  Figure 3.16: Effect of the interaction between AP position and endplate on the endplate strength profile for the Anterior-Posterior profile analysis. F=0.47; p<0.76.  3.1.2.7 Effect of the Interaction between Spinal Level and Endplate The A N O V A revealed no significant interaction between spinal level and endplate for the Anterior-Posterior profile analysis (p<0.99). Across all three levels the mean strength of the inferior endplate appeared similarly stronger than the superior endplate (Figure 3.17).  49  Figure 3.17: Two-way interaction between spinal level and endplate on the endplate strength profile in the Anterior-Posterior profile analysis. F=0.01; p<0.99.  50  3.1.3  Lateral Profile Analysis  3.1.3.1 Effect of Lateral Position Lateral position had a significant effect on the endplate strength profile in the Lateral profile analysis (p<0.024). The overall trend was that the most lateral indentation sites ("-20%,+45%" and "-20%,-45%") had a mean strength greater than the central indentation sites (Figure 3.18). However, Newman-Keuls post-hoc analysis revealed that only indentation site "-20%,45%" had significantly greater mean failure strength than the more central indentation sites (p<0.04). No significant difference existed between site " 20%,+45%" and "-20%,-45%". There was no difference in the mean strength of the five central indentations  o  45%  30%  15%  0%  -15%  -30%  -45%  Lateral position  Figure 3.18: The effect of lateral position on the endplate strength profile in the Lateral profile analysis. F=2.52; p<0.03.  3.1.3.2 Effect of Spinal Level In the Lateral profile analysis, level did not have a significant effect. L2 tended to have a greater mean strength then did T9 and T12 (Figure 3.19)  51  3.1.3.3 Effect of the Interaction between Lateral Position and Spinal Level No effect was detected from the interaction between spinal level and lateral position on the endplate strength profile. The endplate strength profile for each level (Figure 3.20) was similar to the overall profile resulting from lateral position seen in Figure 3.18. L2 was stronger than T9, which was stronger than T12. -o-  100 -  T12  \- I  90 j  45%  \ I  30%  T9  L2 1  15%  0%  -15%  -30%  -45%  Lateral position  Figure 3.20: Two-way interaction between lateral position and spinal level on the endplate strength profile in the Lateral profile analysis. F=0.84; p<0.61.  52  3.1.3.4 Effect of Endplate Although there was no significant difference between the superior and inferior endplate's mean strength (p<0.18), a similar trend as seen in the Maximum and Anterior-Posterior profile analyses was detected (Figure 3.7 and 3.15). The interior endplate's mean failure load was greater than that for the superior endplate's (Figure 3.21). 78  60  I  !  •  Superior  Inferior  ;  1  Endplate  Figure 3.21: Effect of endplate on the endplate strength profile in the Lateral profile analysis. F=1.89; p<0.18.  3.1.3.5 Effect of the Interaction between Lateral Position and Endplate The interaction between lateral position and endplate was significant (p<0.03). For the Lateral profile analysis, the inferior endplate adopted a similar trend to that seen in Figure 3.18 which was the overall effect lateral position had of the endplate strength profile. For this endplate, the indentation sites mean strength decreased from the lateral edge towards the centre (Figure 3.22).  Newman-Keuls post-hoc test revealed that the far lateral  position "-20%,+45%" and position "-20%,+30%" had a significantly greater mean strength than did the central position "-20%, 0%" (p<0.0003 and p<0.05 respectively), but that no other indentation site on the inferior endplate had a mean strength significantly different from the other sites. Newman-Keuls test showed no significant difference in mean strength between indentation sites on the superior endplate for the lateral profile analysis (Figure 3.22). The indentation site with the greatest mean strength was the centre position "-20%,0%", which is opposite to the finding for the inferior endplate (Figure 3.22).  53  45%  30%  15%  0%  -15%  -30%  -45%  Lateral position  Figure 3.22: Effect of lateral position and endplate on the endplate strength profile in the Lateral profile analysis. F=2.41; p<0.03.  3.1.3.6 Effect of the Interaction between Endplate and Spinal Level The A N O V A of the Lateral profile analysis found no significant effect from the interaction between endplate and spinal level. The same general trend existed for each level (Figure 3.23). As was present for the overall endplate effect (Figure 3.21), the inferior endplate was not significantly stronger than the superior endplate. As well, for both the superior and inferior endplate, the mean strength was non-significantly greater for L2 over T12 and T9.  QQ  —Superior  T9  J  T12  L2  Level  Figure 3.23: Effect of endplate and spinal level on the endplate strength profile in the Lateral profile analysis. F=0.06; p<0.94.  54  3.1.4  Cancellous Anterior-Posterior Profile Analysis  3.1.4.1 Effect of AP Position AP position had a significant effect on the endplate strength profile in the Cancellous Anterior-Posterior profile analysis (pO.OOOl). Newman-Keuls post-hoc test showed that the mean strength of row -20% was significantly stronger than the mean strength of the more anterior rows 0%> and +20%) (p<0.002). No difference existed in the mean strength of rows 0%> and +20% (Figure 3.24). 68  -20%  0%  20%  AP position  Figure 3.24: Effect of AP position on the endplate strength profile in the Cancellous Anterior-Posterior profile analysis. F=20.93; pO.OOOl  3.1.4.2  Effect of the Interaction between AP and Lateral Position  There was no significant interaction between A P and lateral position in the Cancellous Anterior-Posterior endplate profile analysis (Figure 3.25).  Post hoc analysis using  Newman-Keuls test confirmed that all three indentation tests performed in row -20%) had a mean strength greater than those performed in rows 0%> and +20%> (p<0.0002). No difference existed between the mean failure loads for any of the indentation sites in rows 0% or +20%). No difference existed between the three columns (column 0%>, +15%, and -15%)  55  75  15%  0%  -15%  Lateral position  0.  , ;  -20%  0%  o  20%  A P position F i g u r e 3.25: Effect o f the i n t e r a c t i o n between A P a n d l a t e r a l position on the endplate strength p r o f i l e f o r the Cancellous A n t e r i o r - P o s t e r i o r p r o f i l e analysis. F=1.17; p<0.33.  3.1.4.3 Effect of Spinal Level No significant effect of spinal level was found in the endplate's mean strength in the Cancellous Anterior-Posterior profile analysis (Figure 3.26).  The trend of L2 being  stronger than T9 and T12 was consistent with a similar trend detected in the Maximum, Anterior-Posterior, and Lateral profile analysis (Fig. 3.4, 3.13, 3.19)  5 6  3.1.4.4 Effect of the Interaction between AP Position and Spinal Level No difference in the endplate strength profile was seen between levels for the Cancellous Anterior-Posterior profile analysis. The general trend was a decrease in mean strength progressing from posterior to anterior (Figure 3.27). However, post-hoc analysis using Newman-Keuls test showed that only in L2 and T12 was this trend significant (p<0.05).  -20%  0%  20%  A P position F i g u r e 3.27: E f f e c t o f the i n t e r a c t i o n between A P position and spinal level on the endplate strength p r o f i l e f o r the Cancellous A n t e r i o r - P o s t e r i o r p r o f i l e analysis. F=0.46; p=0.77.  57  3.1.4.5 Effect of Endplate The A N O V A examining the influence of endplate on the endplate strength did not find an effect for the Cancellous Anterior-Posterior profile analysis (p<0.43). The mean strength for the inferior endplate was greater than that for the superior endplate (Figure 3.28), which is consistent with all previous analysis. 621  •  1  Superior  Inferior  ,  61  Endplate  Figure 3.28: Effect of endplate on the endplate strength profile for the Cancellous AnteriorPosterior profile analysis. F=0.64; p<0.43.  3.1.4.6 Effect of the Interaction between AP Position and Endplate No significant effect was detected from the interaction between A P position and endplate (Figure 3.29). For both the superior and inferior endplates, row -20% was significantly stronger than the two more anterior rows (p<0.03). The shapes were the same but the inferior endplate strength profile was stronger than was the superior endplate strength profile. These findings are consistent with the individual effects AP position and endplate had in the Cancellous Anterior-Posterior profile analysis (Figures 3.24 and 3.27. respectively).  58  45  1  -20%  '  0%  20%  AP position  Figure 3.29: Effect of AP position and endplate on the Endplate strength profile in the Cancellous Anterior Posterior profile analysis. F=0.12; p<0.89.  3.1.4.7 Effect of the Interaction between Spinal Level and Endplate The A N O V A revealed no significant effect from the interaction between spinal level and endplate on the endplate strength profile in the Cancellous Anterior-Posterior profile analysis (Figure 3.30).  No significant difference existed between level or between  endplate. Again L2 tended to be stronger than T9 and T12. 70  45'  j  -  >  - o - T9  1  Superior  Inferior Endplate  Figure 3.30: The effect of the interaction between spinal level and endplate on the endplate strength profile for the Cancellous Anterior-Posterior profile analysis. F=0.1; p<0.91.  59  3.1.5  Ratio posterior endplate / anterior endplate analysis  3.1.5.1  Effect of Level  The ratio of the mean strength of the posterior aspect of the endplate (row -40%) to the mean strength of the anterior aspect of the endplate (row +40%) was significantly different between levels (p<0.04). The ratio for L2 was 1.35, which infers that the mean strength of the posterior aspect of the endplate is 35% greater than that for the anterior aspect. The ratio for T12 was 0.97, which is close to one and corresponds to an endplate with almost equal mean strength between the anterior and posterior region. T9 had a ratio of less than one, 0.91, which corresponds to an endplate that is 9%> stronger in the front in comparison to the back. This finding is graphically displayed in Figure 3.31. The L2 ratio was significantly different from the ratios for both T12 and T9 (p<0.05). No difference existed in the ratio between T9 and T12. 1.4  0.8 I  !  !  !  T9  T12  L2  1  Level  Figure 3.31: Effect of Level on the ratio of posterior endplate / anterior endplate. F=3.72; p<0.05.  60  3.2  D X A Scan Results  3.2.1  Vertebral Bone Mineral Content (BMC) and Bone Mineral Density (BMD)  The B M C was assessed for all vertebral bodies utilized in this research project. Results are listed in table 3.1 below by specimen and level. Table 3.1: D E X A values for each specimen Specimen 1061 1067 1069 1081 1083 1087 1061 1067 1069  3.2.2  Level T9 T9 T9 T9 T9 T9 T12 T12 T12  B M C (g) B M D (g/cm2) Specimen Level 2.7 0.445 1081 T12 1.89 0.343 1087 T12 2.59 0.459 1090 T12 2.97 0.418 1061 L2 2.29 0.375 1067 L2 3.26 0.473 1069 L2 3.71 0.531 1081 L2 2.46 0.38 1083 L2 3.82 0.522 1087 L2  B M C (G) B M D (g/cm2) 3.75 0.423 3.89 0.493 2.99 0.366 3.44 0.454 2.47 0.372 4.27 0.535 4.54 0.466 3.06 0.407 3.61 0.461  Effect of Spinal Level on Vertebral Bone Mineral Content (BMC):  Level had a significant influence on the vertebral B M C (p<0.04) (Figure 3.32). NewmanKeuls test revealed that the average B M C for T9 vertebral bodies was significantly less than the average B M C for T12 and L2 vertebral bodies (p<0.05).  No significant  difference was found between the B M C values between T12 and L2. 3-8 ,  ,  .  r—  .  3.6  2.4 '  •  •  T9  T  1  2  L  i 2  1  Level Figure: 3.32: Effect of the interaction between B M C and spinal level. F=4.12; p<0.04.  61  3.2.3  Effect of Spinal Level on Vertebral Bone Mineral Density (BMD)  There was no significant difference between level in D X A bone mineral density (p<0.58). Although not significant, T9 had a lower mean B M D than did T12 or L2 (Figure 3.33).  o ,3, 0.435 Q  0.430  m  0.425 0.420 .0.415 0.410  T9  T12  L2  Level  Figure 3.33: The effect of the interaction between spinal level and D X A B M D . F=0.57;p< 0.58.  62  3.3  Endplate Regional Bone M i n e r a l Density Profiles  3.3.1 3.3.1.1  Maximum Profile Analysis Effect of AP Position  AP position had a significant effect on the endplate mean r B M D profile in the Maximum profile analysis (p<0.0001). The mean r B M D for each row decreased from the posterior aspect of the endplate towards the anterior aspect of the endplate (Figure 3.34). Post hoc Newman-Keuls test revealed that rows -2 had a mean r B M D , which was significantly greater than the mean r B M D for rows 0 and +1 (p<0.005). The mean r B M D for row -1 was also significantly greater than were those for row 0 and +1 (p<0.04 and p<0.0005, respectively). The difference in mean r B M D for rows -2 and -1 approach statistical significance on the post hoc analysis (p=0.07), as did the difference in mean r B M D for rows 0 and +1 (p=0.07). 185  +3 +2 +1 0  -1  0  +1  A P position  Figure 3.34: Effect of A P position on the endplate r B M D profile for the M a x i m u m profile analysis. F=12.99; PO.OOOl.  3.3.1.2 Effect of Lateral Position The relative lateral position had a significant effect on the endplate r B M D in the Maximum profile analysis (p<0.006).  Post hoc analysis revealed that no significant  difference existed between the centre and periphery of the endplate, but that the sides were not symmetrical (Figure 3.35).  The mean r B M D of columns -1 and - 2 were  significantly greater than the mean r B M D in column +2 (p<0.01).  A non-significant  63  decrease in mean r B M D occurred on the positive side relative to the centre, while a nonsignificant increase occurred on the negative side. 190 1 8 5 _  co  +3 +2 +1 0  -1 -2 -3  1 8 0  E  T5>  CQ  175 1 7 0 165 160  +2  +  1  0  Lateral position  Figure 3.35: Effect of lateral position on the endplate r B M D profile for the M a x i m u m profile analysis. F=3.83; p<0.006.  3.3.1.3  Effect of Interaction between the AP and Lateral Position  Analysis of the endplate r B M D profile revealed a significant interaction between the A P and lateral positions for the Maximum profile analysis (p<0.0001).  The lateral r B M D  profile was significantly different dependent on the relative A P position (row). A l l A P r B M D profiles showed a typical decrease in density from posterior to anterior regardless of the relative lateral position (Figure 3.36). Newman-Keuls analysis found that in the posterior row (row -2) the peripheral regions had a significantly greater B M D then did the centre (p<0.001). Furthermore, the endplate's posterolateral corners had the highest B M D of all the subcortical regions of the endplate. Region "-2, -2" was significantly more dense than any other region of the endplate, and the opposite corner, region "-2, +2" was significantly more dense than the anterior endplate (row +1), and the lateral aspects of the more central columns (column 0 and -1) (Appendix B). In the central rows, rows -1 and 0 the opposite trend was discovered. The centre was stronger than the periphery. This however reached a significant level on only one side (column +2), with the p value ranged from 0.0002 to 0.07. In the anterior aspect of the endplate no difference in the mean r B M D was found except laterally in column - 2 , which was significantly greater than the opposite side column +2 (p<0.02). The two more anterior columns displayed the  64  lateral asymmetry found in the overall lateral endplate r B M D profile described in section 1.3.1.2. while the two posterior columns were more symmetrical. 205  j  •  j  •  ' +1  ' 0  • - 1  .  ,  200  155 150'  -  2  AP position 205 200  155 150'  • +2  • + 1  • 0  • - 1  -  2  Lateral position Figure 3.36: Effect of the interaction between AP and lateral position on the endplate rBMD profile in the Maximum analysis profile. F=5.93; p<0.0001.  3.3.1.4 Effect of Spinal Level There was no significant difference in the mean B M D between the three levels of study (p<0.75). However, the mean B M D for L2 was less than that for both T9 and T12 (Figure 3.37).  65  lateral asymmetry found in the overall lateral endplate r B M D profile described in section 1.3.1.2. while the two posterior columns were more symmetrical. 205 200 195 190  n 185 O  180  a  175  CO  a  >vi - - >  •- *  170  ~ ~D- .  |  165  f<^  ""~*  160 155 150  +1  0  -1  A P position  205 200 195 190 185 o 180 3 175 Q  170  CQ  165  -a-'""  T^Hr  /  " i j >  160  :  155 150  2  +  +1  0  -1  .2  Lateral position  Figure 3.36: Effect of the interaction between A P and lateral position on the endplate r B M D profile in the M a x i m u m analysis profile. F=5.93; p<0.0001.  3.3.1.4  Effect of Spinal Level  There was no significant difference in the mean B M D between the three levels of study (p<0.75). However, the mean B M D for L2 was less than that for both T9 and T12 (Fig ure 3.37).  65  160 '  :  T9  : T12  L2  Level  Figure 3.37: The effect of spinal level on mean B M D for the M a x i m u m profile analysis. F=0.3; P<0.75.  3.3.1.5  The Effect of the Interaction between AP Position and Spinal Level  Spinal level had a significant effect on the endplate r B M D profile relative to AP position for the Maximum profile analysis (p<0.003). For T9 and T12, no differences in the mean r B M D were found between rows (Figure 3.38). However, for L2 there was a similar decrease in the mean r B M D from a posterior to anterior position was shown in Figure 3.34. Newman-Keuls test showed that for L2 the mean r B M D for the anterior most row (row +1) was significantly less than that for the other three rows located posterior (p<0.004).  66  Figure 3.38: The effect of the interaction between A P position and spinal level. F=3.79; p<0.003.  3.3.1.6 Effect of the Interaction between Lateral Position and Spinal Level In the comparison of r B M D profiles relative to lateral position between spinal level no significant differences were seen (Figure 3.39).  190  +3 +2 +1 0  +1  -1 -2 -3  0 Lateral position  Figure 3.39: The effect of the interaction between lateral position and spinal level on the endplate r B M D profile for the M a x i m u m profile analysis. F=11.37; p<0.22.  67  3.3.1.7  Effect of Endplate  There was no significant difference observed between the mean r B M D between the superior and inferior endplates in the Maximum profile analysis (p<0.57), although the inferior endplate had somewhat higher B M D values (Figure 3.40). 185 r  f  ;  180  -  • -  175  -  -  !--  j  -y Q  ;  -  -  +3 +2 +1 0 -1 -2 -3  -  -  ^ ^ ^ ^ 170  :  CO 165  i  160 I  I  •  •'  Superior  Inferior  Endplate Figure 3.40: The effect of endplate on the mean rBMD for the Maximum profile analysis. F=0.33; p<0.57.  3.3.1.8  Effect of the Interaction between AP Position and Endplate  There was no significant difference in the endplate r B M D profile relative to A P position between the superior and inferior endplates (p<0.73).  For both endplates there was a  decrease in the mean r B M D from the posterior to anterior position (Figure 3.41).  68  \  i  t  1 8 5  \ I  -  -  •o - S u p e r i o r o  - I n f e r i o r +3 +2 +1  155  !  s  -2  L  -1  L  0  0  -1 -2 -3  +1  A P position  Figure 3.41: The effect of the interaction between A P position and endplate on the rBMD profile for the Maximum profile analysis. F=0.44; p<0.73.  3.3.1.9 Effect of the Interaction between Lateral Position and Endplate Endplate did not significantly affect the endplate r B M D profile relative to lateral position (p=0.99). The profiles were very similar between superior and inferior endplates as shown in Figure 3.42. Again, as seen from the overall effect lateral position had on the endplate r B M D profile (Figure 3.35), for both endplates the mean r B M D for column +2 was less than the centre mean rBMD, while column - 2 was greater. 190  '  -o- Superior isrj  „. --o- Inferior  |  |  • i  i  j  [ I  j  ,  +3 +2 +1 _  co E  180  -  I  -1 -2 -3  4  i  \  j  175  Q  0  1  7  0  f  \  :  165  160  !  +2  +1  0  -1  -2  Lateral position  Figure 3.42: The effect of the interaction between lateral position and endplate on the endplate rBMD profile in the Maximum profile analysis. F=.04; p=0.99.  69  3.3.1.10 Effect of the Interaction between Spinal Level and Endplate There was no significant interaction effect between spinal level and endplate (p<0.98). Across all levels there was no significant difference in the mean r B M D between the superior and inferior endplates (Figure 3.43). 185,  160  1  ;  :  Superior  ;  .  1  1  Inferior  Endplate  Figure 3.43: The effect of the interaction between the spinal level and endplate on the mean r B M D for the M a x i m u m profile analysis. F=0.02; p<0.98.  70  3.3.2  Anterior-Posterior Profile Analysis  3.3.2.1 Effect of AP Position AP position had a significant effect on the endplate r B M D in the Anterior-Posterior profile analysis (p<0.01). The general profile was a decrease in the mean r B M D from the posterior of the endplate towards the anterior aspect, except at the most anterior row, which revealed an increase in mean r B M D equivalent to that in the posterior rows (Figure 3.44). The mean r B M D in row +1 was less than that for any other column. NewmanKeuls post hoc analysis revealed that the endplate trabecular bone was significantly more dense in the posterior rows (row - 2 and row -1) than that in anterior row +1 (p<0.022). The most anterior row, row +2, had a mean r B M D , which was significantly greater than the mean density in row +1 (p<0.04). No significant difference existed in mean r B M D between row 0 and row +1. 185  160 '  ;  -2  !  -1  •' 0  • +1  :  ;  1  +2  A P position Figure 3.44: The effect of A P position on the endplate r B M D profde in the AnteriorPosterior profile analysis. F=3.39; p<0.01.  3.3.2.2  Effect of the Interaction between AP and Lateral Position  A significant interaction was detected between A P and lateral position by the A N O V A for the Anterior-Posterior profile analysis (p<0.0007). There was a decrease in the mean r B M D from row -2 to row +1 for all the three columns, except for the centre column (column 0) in the posterior row (row -2) (Figure 3.45). In this region, Newman-Keuls test revealed the mean density was significantly lower than any other region in the two posterior rows in the Anterior-Posterior profile analysis (range: p<0.004 - 0.04). 71  190 185 180  <  +3 +2 +1  ft  0  -1  -2 -3  cO~  E  _o  175  O)  Q  170  \  /  1  \  CQ 165 160  -2  -1  0  +2  +1  AP position  155  Figure 3.45: The effect of the interaction between AP and lateral position on the endplate rBMD profile in the Anterior-Posterior profile analysis. F=3.57; p<0.0007.  3.3.2.3 Effect of Spinal Level There was no significant difference in the mean r B M D between level in the AnteriorPosterior profile analysis (p<0.40). The mean r B M D was less for L2 than for T9 or T12 but this was a non-significant trend (Figure 3.46). 190  185  ^  +3 +2 +2  180  CO  +1  E o  "5)  175  CQ  170  0  i  -1  -2 -3  i  N  -1 •  -2  r —"  165  160  /  0  T9  T12  L2  Level  Figure 3.46: The effect of spinal level on the mean rBMD for the Anterior-Posterior profile analysis. F=0.97; p<0.40.  72  1  3.3.2.3  Effect o f Spinal Level  There was no significant difference in the mean r B M D between level in the AnteriorPosterior profile analysis (p<0.40). The mean r B M D was less for L2 than for T9 or T12 but this was a non-significant trend (Figure 3.46).  160  1  • T9  T12  'L2  Level  Figure 3.46: The effect of spinal level on the mean r B M D for the Anterior-Posterior profile analysis. F=0.97; p<0.40.  72  3.3.2.4  Effect of the Interaction between AP Position and Spinal Level  A significant interaction occurred between A P position and spinal level for the AnteriorPosterior profile analysis (p<0.0001).  In the case of T9 and T12 the endplate r B M D  profile for A P position was different than the profile of L2 (Figure 3.47). For both T9 and T12 there existed a gradual decrease in mean r B M D from row -2 toward row +1, followed by an increase in mean r B M D for row +2 relative to all rows more posterior. Newman-Keuls post hoc analysis showed that there was no significant difference between mean densities within either level. In the case of L2, the two anterior most rows (row +2 and row +1) had a significantly lower mean density than rows -2 through row 0 (range: p<0.03-0.0003), except for row - 2 relative to row +1 which was not significantly different (p<0.05).  - o - T9 150  -D-  140 •• 130  T12 L2 -1  0  +1  +2  A P position  Figure 3.47: The effect of the interaction between A P position and spinal level on the endplate r B M D profile in the Anterior-Posterior profile analysis. F=5.51; p<0.0001.  3.3.2.5  Effect of Endplate  There existed no significant difference in the mean density of the superior or inferior endplate in the Anterior-Posterior profile analysis (p<0.35).  The mean density of the  superior endplate was lower than that for the inferior endplate (Figure 3.48).  73  185  §  180  i  175  !  6  •£  ^  J \  j  O)  Q  +3 +2 +1 0 -1 -2 -3  i  170  CO.  ^  !  *  1-  165  160 I  |  -  ]  • Superior  i Inferior  Endplate  Figure 3.48: The effect of endplate on the mean rBMD for the Anterior-Posterior profde analysis. F=0.93; p<0.35.  3.3.2.6 Effect of the Interaction between AP Position and Endplate A significant interaction was found between A P position and endplate for the AnteriorPosterior profile analysis (p<0.05). For both superior and inferior endplates the r B M D profile exhibited a general decrease in mean r B M D from the posterior two rows (row -2 and row -1) towards the adjacent anterior two rows (row 0 and row +1) (Figure 3.49). However, there was an obvious difference in both the magnitude of this decline, as well as the increase in mean r B M D between row +1 to row +2 between superior and inferior endplates. Post hoc analysis revealed that the decline between row - 2 to row +1 in the inferior endplate resulted in no significant difference in mean r B M D , while for the superior endplate, the difference between both the mean r B M D for row -2 and row -1 were almost significantly greater than that for row +1 (p<0.06).  Furthermore, the  increase in mean r B M D between the two most anterior rows was not significant for the superior endplate, but was for the inferior endplate (p=0.05).  74  195  150  -  ;  -2  !  -1  0  -  • +2  !  +1  A P position F i g u r e 3.49: T h e effect o f the i n t e r a c t i o n between A P position a n d endplate f o r the r B M D p r o f i l e i n the A n t e r i o r - P o s t e r i o r p r o f i l e analysis. F=2.48; p<0.05.  3.3.2.7 Effect of the Interaction between Spinal Level and Endplate There is no significant interaction between spinal level and endplate on the mean r B M D for the Anterior-Posterior profile analysis (p<0.91). 195 -°190 185 0 0  E o  15)  Superior  --Q-- Inferior  +2 +1 0  180 175  CO  F i g u r e 3.50: T h e effect o f the i n t e r a c t i o n between spinal level and endplate f o r the A n t e r i o r Posterior p r o f i l e analysis. F = 0 . 1 ; p<0.91.  75  3.3.3  Lateral Profile Analysis  3.3.3.1 Effect of Lateral Position Lateral position had a significant effect on the endplate r B M D profile in the Lateral profile analysis (pO.OOOl).  As seen in Figure 3.51, the trabecular bone adjacent to the  endplate at its periphery had a lower mean r B M D than did that in the centre. The mean r B M D decreased from the centre towards the periphery. However, Newman-Keuls test showed that only on the positive side was the density significantly greater than the centre (p<0.0003), and in fact, the mean density at position +3 was significantly greater than that on the opposite side as well (p< 0.03) (Appendix B).  +3 +2 +1  0  -1  -2  -3  +2 +1 0 -1  /  I  111111  -2  /  +1  0  J  -1  Lateral position  Figure 3.51: The effect of lateral position of the endplate r B M D profile in the Lateral profile analysis. F=5.98; p<0.0001.  3.3.3.2 Effect of Spinal Level There was no significant difference in the mean r B M D between level for the Lateral profile analysis (p<0.86). In this analysis, T9 was less than T12 and L2, which were similar in mean density (Figure. 3.52).  76  Figure 3.52: The spinal effect of level on the mean r B M D in the Lateral profile analysis. F=0.16; p<0.86.  3.3.3.3  Effect of the Interaction between Lateral Position and Spinal Level  The shape of the endplate r B M D profile was not significantly different between levels for the Lateral analysis profile (p<0.11).  The overall trend was an increase from the  periphery towards the centre of the endplate in mean r B M D (Figure 3.53), which is also seen as the overall r B M D profile for lateral position (Figure 3.51). However, T12 on the negative side of the periphery does not decrease, but rather increases. Newman-Keuls post hoc analysis does not reveal any significant difference in mean density within levels, except within L2, where position -1 is significantly more dense than position +3 (p<0.04).  77  +3 +2 +1 0 +2  v  +1 0 J -1  -1 -2 -3  \  I  Figure 3.53: The effect of the interaction between lateral position and spinal level on the endplate r B M D profile in the Lateral profile analysis. F=1.56; p<0.11.  3.3.3.4  Effect of Endplate  No significant difference existed in the mean r B M D between superior and inferior endplates (p<0.60). The mean density was greater for the inferior endplate as compared to the superior endplate (Figure 3.54). 185 i  1  :  180 CO  E o  O)  175  Q  CO 170  165  Superior  Inferior  Endplate  Figure 3.54: The effect of endplate on the mean r B M D for the Lateral profile analysis. F=0.29; p<0.60.  78  3.3.3.5  Effect of the Interaction between Lateral Position and Endplate  The was no significant effect resulting from the interaction between lateral position and endplate for the Lateral profile analysis (p<0.98).  In both the superior and inferior  endplate r B M D profiles the central mean densities were greater than the lateral densities (Fig. 3.55).  Newman-Keuls test revealed that a significant difference existed within  endplate profiles for only the superior endplate; position +3 relative to positions +1, 0, and -1 (p<0.02). 190 185 1 180 P>  175  ^5)  170  E o  Q  /  j  h  j  - <p j \  165  CO 160 155 150  ! ! ! i  Y !  |  !  +3  +2  +1  i  !  -o-  Superior  •--  Inferior  0  Lateral  Figure 3.55: The effect of the interaction between lateral position and endplate for the Lateral profile analysis. F=0.21; p<0.98.  3.3.3.6  Effect of the Interaction between Spinal Level and Endplate  No significant difference exists between spinal level for either the superior or inferior endplate in the Lateral profile analysis (p<0.95). For all three levels the superior endplate has a lower mean density than the inferior endplate. This is less pronounced for T12 than for L2 or T9 (Figure 3.56).  79  185  +2  +3  +1  0  -1  -2 -3  +2 +1  0  -o- T9 -Q- T12 L2 Superior  I  -1 ]  -2  |  /  y  Inferior  Endplate Figure 3.56: The effect of the interaction between spinal level and endplate on the mean r B M D for the Lateral profile analysis. F=0.05; p<0.95.  80  3.3.4  Ratio Posterior Endplate /Anterior Endplate Analysis  3.3.4.1 Effect of Level A significant difference existed between the L2 and T9 as well as L2 and T12 (p<0.002). There was no difference between the ratio for T9 and T12. Both T9 and T12 had a ratio of essentially one (0.99 and 1.02, respectively), implying no difference in the mean r B M D between the most anterior and posterior aspects of the vertebral endplate. The ratio for L2 was 1.33, which was interpreted as the mean r B M D being 33% greater in the posterior row compared to the anterior row (Figure 3.57). 1.40 1.35 1.30 1.25  .2  1 2 0  I 1.15 1.10 1.05 1.00 0.95  T 9  T 1 2  L2  Level Figure 3.57: Effect of spinal level on the ratio of Posterior endplate / anterior endplate. F=3.72; p<0.04.  81  3.4  Correlation between Endplate rBMD and Force  3.4.1 Maximum Profile Analysis A significant correlation existed between r B M D of the sub-endplate zone and force to failure on the vertebral endplate (p=0.05). The Pearson r coefficient was 0.56, which represented a rather low correlation.  \  j  i  :  |  !  °80  120  160  240  ;  o  \o  200  °  !  I  \  \  240  280  \  I  I  !  320  360  rBMD (g/cm3) Figure 3.58: The correlation between r B M D and force for the M a x i m u m profile analysis. Pearson r=0.56; p<0.05.  3.4.2 Anterior-Posterior Profile Analysis The Pearson r coefficient for the correlation between indentation failure force and r B M D for the Anterior-Posterior Profile was 0.57. This was a significant (p=0.05) but again rather low correlation.  82  1  280 240 200 160 O  120 80 40 0 80  120  160  200  240  280  320  360  400  rBMD (g/cm3) Figure 3.59: The correlation r B M D and force for the Anterior-Posterior profile analysis. Pearson r = 0.57; p<0.05.  3.4.3  Lateral Profile Analysis  A significant correlation was found between the indentation failure force and r B M D in the Lateral profile analysis (p=0.05). The Pearson r coefficient was 0.48, which represented a low correlation. 280  " O  0  100  °  OcPOXiO  o ob  140  180  220  260  Regression 95%confid. | 300  340  rBMD (g/cm3) Figure 3.60: The correlation between r B M D and force for the Lateral profile analysis. Pearson r=0.48; p<0.05.  83  3.4.4  Cancellous Anterior-Posterior Profile Analysis  A significant correlation was found between the indentation failure force and r B M D in the Cancellous Anterior-Posterior profile analysis (p=0.05).  This was a stronger  correlation than that found for the Anterior-Posterior profile analysis. The Pearson r was 0.68, which represented a moderate correlation between force and density.  Z  240  \  I  |  200  I  I  I  -  |  \  \  \  I  \  i  \  160  r B M D (g/cm3) Figure 3.61: The correlation between force and r B M D for the Cancellous Anterior-Posterior profile analysis. Pearson r=0.68; p<0.05.  84  4.0 Discussion Anterior interbody fusion procedures are common in the spine surgeon's armentarium and important in the treatment of a diversity of clinical pathology. The theory behind anterior column reconstruction is sound and the clinical results confirm this. Failure of the reconstruction occurs most commonly due to subsidence of the interbody device into the endplate.  This is particularly a problem for the osteoporotic patient, which is  concerning when one considers the changing demographics of the developed world, towards an increasing proportion of the population being comprised of elderly. In the Introduction, the factors attributing to the cause and prevention of implant subsidence were analysed.  The purpose of this research was to further advance our current  understanding of these factors, by defining the strength and r B M D profile of the thoracolumbar endplate, as well as the correlation existing between them. It is hoped that such an understanding will improve our ability to influence the "surgical and implant factors", thereby decreasing the risk of implant subsidence.  4.1 4.1.1  The Thoracolumbar Endplate Strength Profile Effect of Level  No significant differences in the mean endplate strength between level were detected by any of the profile analyses (Maximum, Anterior-Posterior, Lateral, and Cancellous Anterior-Posterior). In all profile analyses, the same general trend was seen that L2 was strongest, and T9 was slightly stronger than T12. Previous studies have conflicting results with respect to strength difference between levels (Brinckmann et al, 1989; Grant et al, 2001; Hansson et al, 1980; Singer et al, 1995; Yoganandan et al, 1988). Singer et al. showed the vertebral strength increased from the upper thoracic spine to the lower lumbar spine (Singer et al, 1995). This has been confirmed for the lumbar spine (Brinckmann et al, 1989; Hansson et al, 1980). In these studies, whole body vertebral strength was analysed. Therefore, failure load was dependent not only on vertebral structure but also vertebral size. It has been shown that vertebral strength correlates directly with cross-sectional area (Brinckmann et al, 1989; Singer et al, 1995).  85  A different analysis technique was utilized in this study and in that by Grant et al., which was the mean strength of multiple indentation sites (Grant et al, 2001). These studies found no difference in strength between spinal level following the analysis of thoracolumbar or lumbar spine. The contrasting results may be explained by the difference in the studies methods; the mean strength of multiple indentation sites is different than vertebral body failure load following entire vertebral body compression testing. Yoganandan et al. found no difference in vertebral strength relative to level following full body compression testing (Yoganandan et al, 1988).  Although their  findings were not consistent with those of other studies utilising full body compression testing, their testing technique differed significantly. They removed the endplate prior to testing where as the others did not. It has been shown that the subchondral endplate is stronger than the sub-endplate trabecular bone (Brinckmann et al, 1989; Grant, 2000; Kumar and Doherty, 1994; Lim et al, 2001; Oxland et al, 2003; Zaccaria, 2003).  4.1.2 4.1.2.1  Effect of Position A P Position  The findings of this study are consistent with those for the lower lumbar and lumbosacral spine (Grant et al, 2001; Kumar and Doherty, 1994). Endplate mapping revealed that the strongest aspects of the endplate in the A P excursion were at the back and front, with the centre being weakest. The most anterior aspect of the thoracolumbar endplate was the strongest in the A P profile analysis. There was a gradual decrease in endplate strength from the posterior of the endplate towards its centre seen in both the Maximum and Anterior-Posterior profile analysis.  Overall, these findings occurred consistently across all three levels, for both the Maximum and Anterior-Posterior profile analysis. Although L2 and T9 did not have an overall significantly different endplate profile, in the Maximum profile analysis, a significant difference existed in the mean strength of the posterior aspect of the endplate (row -40% and row -20%>), while at the more anterior aspect no difference existed. This implies that there is a relatively greater strength differential between the anterior and posterior aspect of the endplate for L2 than for T9 or T12 (Figure 3.5).  This lends  86  indirect support to the hypothesis that the endplate strength profile is influenced by the position of the endplate relative to the sagittal balance of the spine. Furthermore, in the A P profile analysis, which allows for inspection of the most anterior and posterior aspects of the endplate (Maximum profile analysis excludes the most anterior row), the trend discussed above is further demonstrated. The anterior aspect of the T9 endplate had a significant increase in strength relative to its posterior aspect, while no difference was seen for L2. This trend accounts for only the three most central columns of indentation sites, as that defined the width of the Anterior-Posterior profile analysis. This analysis excluded the posterolateral corners, the strongest aspect of the endplate. To compensate for this deficiency, a ratio of the mean strength of posterior row indentation sites relative to those in the anterior row was compared between levels. This comparison revealed a significant difference. As presented in the results chapter, the ratio for L2 was 1.35, which infers that the mean strength of the posterior aspect of the endplate is greater than that for the anterior aspect.  The ratio for T12 was 0.97, which is close to one and  corresponds to an endplate with equal mean strength between the anterior and posterior aspects. T9 had a ratio of less than one, 0.91, which corresponds to an endplate that is relative stronger in the front in comparison to the back. The L2 ratio was significantly different from the ratio for T12 and T9 (p<0.05). This significant difference and the trends described for the Maximum and Anterior-Posterior profile analysis above, supports that sagittal balance does influence the endplate strength profile relative to A P positioning.  A detailed mapping of the thoracolumbar endplate strength profile, to our knowledge, has not been performed before. The lower lumbar endplates have been mapped, but the influence of sagittal alignment on the profiles has not been described (Grant et al, 2001; Kumar and Doherty, 1994). It stands to reason, that just as the endplate strength profile is thought to be influenced by changes in loading characteristics with progressive disc degeneration in accordance to Wolffs Law (Grant et al, 2002; Keller et al, 1989; Kurowski and Kubo, 1986; Shirazi-Adl et al, 1984), so should changes in sagittal alignment have a similar influence. Sagittal alignment is known to vary widely between individuals but overall the thoracic spine is kyphotic, the thoracolumbar spine (T12-L1)  87  close to neutral, and the lumbar spine lordotic (Bernhardt and Bridwell, 1989; Jackson and McManus, 1994; Stagnara et al., 1982). A plumb line dropped from the centre of C7, runs anterior to the thoracic spine, transects the thoracolumbar junction, and lies posterior to the lumbar spine. The endplate loading characteristics and strength profile should be reflective of this changing alignment in accordance with Wolffs Law. It is hypothesised that the anterior aspect of T9 was found to be relatively stronger than the posterior aspect because T9 is in the kyphotic curve, posterior to the plumb line, where relatively more load is transmitted through the anterior aspect of the body. The strength profile of L2 and T12 were similarly found to reflect their relative position in the spinal column's overall sagittal balance. For T12, load is equally distributed across the endplate relative to A P position, which is reflected by a posterior/anterior ratio close to 1.0. The L2 vertebral endplate is stronger in the posterior aspect relative to the anterior aspect, which in accordance with Wolffs Law, is thought to reflect the posterior vertebral loads seen in the lordotic spine.  4.1.2.2  Lateral Position  The endplate strength profile showed that the lateral aspects of the endplate were significantly stronger than the central aspects. This effect was consistent across level. The Maximum analysis endplate strength profile revealed that the mean strength for indentation sites in columns +30% and -30%) (the lateral extent of the analysis) was significantly greater than the mean strength of the more central columns. In the Lateral analysis endplate strength profile, the mean strength of columns +30%> and -30% were not greater than that for the more central columns. However, a significant increase was detected in columns +45% and -45%, which is the lateral extent of the Lateral profile analysis.  At first glance this appears to be a discrepancy, but is resolved by  understanding the effect of the interaction between A P and lateral position on the endplate strength profile in the Maximum profile analysis. In this analysis, only rows -40%) and +20% had a side to centre difference, where as for the central rows -20% and 0% lateral position did not have an effect. Therefore, the finding from the Lateral profile analysis that columns +30% and -30% are not significantly stronger than the centre, is  88  consistent with the findings in the Maximum profile analysis, as the Lateral profile analysis included only indentation sites from row -20%.  The difference in the strength profile between the central rows and rows -40% and +20% (which is the effect of the interaction between A P and lateral position) in the Maximum profile analysis is explained by the general morphology of the endplate. In the centre of the endplate, where rows -20% and 0% are located, the endplate is wider than it is in its anterior and posterior aspects. Therefore, in the central rows, the most lateral indentation sites are excluded from the Maximum profile analysis because they extend beyond the defined area for the analysis. In the anterior and posterior rows (row +20% and —40%) the analysis includes indentation sites located at the most lateral aspects of the endplate. This suggests that it is the strength of the endplate's most lateral aspects, which creates the effect lateral position had in the Maximum profile analysis, as only rows —40% and 20% showed a difference between the centre and lateral sides. This observation is further supported by the Lateral profile analysis, in which only indentation sites positioned at "-20%,+45%", which is the furthest lateral extent of the row, had a mean strength significantly stronger than the more central positions. 4.1.2.3  Rationale for "Cancellous Endplate Analysis" in the Anterior-Posterior Profile  As discussed in the summary above, the effect of AP and lateral position on the endplate strength profile was detected due to a relative difference in mean strength between the stronger peripheral indentation sites compared to those positioned in the centre. These indentation sites correspond to the location of the "cortical ridge" surrounding the periphery of the endplate. This ridge is the ossified apophyseal ring. A further A N O V A was performed on indentation sites found only within the apophyseal ring. This analysis was performed to determine whether the effect of AP and lateral position on the endplate strength profile occurred without the influence of this cortical ridge. Hence a Cancellous Anterior-Posterior profile analysis was performed. Interestingly, with all indenter sites located on the apophyseal ring removed from the analysis, there remained a significant difference relative to A P position. Row -20% was significantly stronger than rows 0% or +20%. This trend was present when looking at all levels independently, however, it was  89  only significantly different for L2 and T12. No difference was seen in endplate strength relative to lateral position when indenter sites located on the apophyseal ring ("20%,+45%", and "-20%,-45%") were removed from the analysis. Therefore, the cortical ring of the endplate fully explains the strength variation between the anterior, anterolateral, and lateral aspects with the central endplate, but does not fully explain the increased strength in the posterior aspect.  This confirms the importance of the  apophyseal ring in resisting implant subsidence previously discussed by other authors (Kozak et al, 1994; Kumar and Doherty, 1994; Steffen et al, 2000b; Steffen et al, 2000a).  4.1.3  Effect of Endplate  There was no significant difference in the mean strength between the superior and inferior endplates.  The inferior endplate was generally stronger than the superior  endplate, but this was not found to be significant in any two-way A N O V A (Maximum, Anterior-Posterior, Lateral, or Cancellous profile analysis). These findings are consistent with those of Grant et al., and Kumar et al. for the endplates of the lower lumbar spine (Grant et al, 2001; Kumar and Doherty, 1994).  This trend was consistent at all levels. The endplate strength profile did not change relative to A P position between the superior and inferior endplates. However, the profile relative to lateral position was very different between the superior and inferior endplates. The superior endplate displays no variation in strength moving from the centre towards the periphery in the Maximum profile analysis, and a non-significant decrease in strength in the Lateral profile analysis. On the other hand, the inferior endplate strength profile reveals the general trend of increasing strength from the centre towards the periphery. In the Lateral profile analysis, the apophyseal ring effect discussed above is present for both superior and inferior endplates. It can be hypothesized that this difference in the endplate strength profile reflects the ability of the disc to distribute stress across the superior endplate. Where as, for the inferior endplate, the majority of load is distributed from the vertebral cortex wall, apophyseal ring, and endplate periphery, into the underlying disc. According to Wolffs Law this difference in stress would be reflected by a difference in  90  bone density and correlate with a difference in strength. Having said that, no difference in the r B M D profile was seen between the superior and inferior endplates, which casts serious doubt on this hypothesis.  4.1.4  Summary of the Endplate Strength  Profile  The strongest aspect of the thoracolumbar endplate was in the posterolateral corners. The weakest aspect was found in the centre of the endplate, surrounded by the apophyseal ring. The anterior rim was the second strongest aspect of the endplate, especially for T9. The anterolateral corners, midposterior (-40%, 0%), far lateral, and posterocentral (row 20%) aspects of the thoracolumbar endplate were significantly stronger than the centre of the endplate.  Unfortunately, the majority of interbody fusion devices transfer load  through the weakest aspects of thoracolumbar endplate, as defined by this experiment.  4.2 4.2.1  The Thoracolumbar r B M D Profile Effect of level  In all three analyses there existed no difference in mean r B M D between levels (Maximum, Anterior-Posterior, Lateral profile analyses). However, a trend that L2 was less dense than both T12 and T9 was present for all three analyses. Two other studies examining regional bone mineral density with QCT have found no significant difference between level (Antonacci et al, 1997; Banse et al, 2002). No significant difference existed between levels in the D X A B M D either.  The D X A B M C increased in a craniocaudal direction. T9 had a significantly lower B M C value than either T12 or L2. This intuitively makes sense, as a L2 vertebra is larger than a T12 vertebra, which is larger than a T9 vertebra.  In this study no comparison was performed between D X A B M D and mean r B M D . The measurements are different as D X A includes both cortical and cancellous bone in its B M D measurement and pQCT includes only trabecular bone. Using QCT, cortical and cancellous B M D have been shown to be affected by changing level independently from each other (Brassow et al, 1982; Singer and Breidahl, 1990). Brassow et al. found no  91  difference in the mean cancellous B M D of thoracic and lumbar vertebrae, but showed an increasing mean cortical B M D with caudal descent (Brassow et al, 1982). While Singer et al. found the mean D X A B M D to increase when moving from thoracic spine to lumbar spine, but inversely found QCT trabecular B M E D to decrease (Singer and Breidahl, 1990).  4.2.2  Effect of Position  4.2.2.1 A P Position The density of the trabecular bone beneath the thoracolumbar endplate decreased from the posterior aspect toward the centre. The lowest mean r B M D was found to be in the centre of the subcortical endplate for both the Maximum and Anterior-Posterior profde analysis. Based on the Anterior-Posterior profile analysis, no significant difference exists between the posterior and anterior mean density.  The r B M D endplate profile was  different in the posterior half when comparing the Maximum and Anterior-Posterior profile analysis. In the Maximum profile analysis there was an obvious decline in density between each row going from the most posterior aspect of the endplate (row -2) toward the centre (row +1).  However, for Anterior-Posterior profile analysis, there was no  difference in density between the two most posterior rows (row -2 and row -1). This difference in the r B M D endplate profile is related to the posterolateral corners of the endplate, which is the densest part of the endplate.  The Maximum profile analysis  includes the posterolateral corner in the analysis, while the Anterior-Posterior profile analysis does not. Therefore, the posterolateral corners produce the decrease in mean r B M D between row -2 and row -1 seen in the Maximum profile analysis, but not the Anterior-Posterior profile analysis.  The difference in regional density relative to A P position identified in this study is fairly consistent with the literature (Edwards et al, 1986; Flynn and Cody, 1993; Lu et al, 2001; Matsui, 1991; Nepper-Rasmussen and Mosekilde, 1989; Simpson et al, 2001). Lu et al. found that the posterolateral and posterocentral aspects of the sub-endplate zone to be the most dense part of the endplate (Lu et al, 2001). This was confirmed in our study. However, a number of authors have shown that the anterior vertebral body is significantly  92  less dense than is the posterior aspect, while this study found no regional difference in density of the sub-endplate zone (Banse et al, 2001; Edwards et al., 1986; Matsui, 1991; Nepper-Rasmussen and Mosekilde, 1989). For example, Singer et al. analysed three ROI with QCT in the thoracolumbar spine and found that the density of the anterior body to be greater than the posterolateral body (Singer and Breidahl, 1990). This discrepancy may be explained by the differences in imaging technique. The regions of interest identified for analysis in the other studies were much larger than were those used in this study. This study found that the least dense row was row +, and the most anterior row significantly increased in density over that one. As the other studies used much larger ROI, the two most anterior rows may well have both been included as the "anterior ROI", producing an averaging effect. This averaging effect would decrease the mean r B M D of the anterior ROI causing it to be less dense than the posterior ROI. A further difference in technique is that this study examined the sub-endplate zone, while the other studies utilized a QCT slice through the central aspect of the vertebral body (Nepper-Rasmussen and Mosekilde, 1989; Singer and Breidahl, 1990), or averaged multiple axial cuts spanning the entire vertebral body (Banse et al., 2001; Edwards et al, 1986).  There existed a significant difference between the endplate r B M D profile between thoracolumbar levels. For the T9 and T12 endplates no significant difference existed between mean r B M D relative to A P position. In the case of the L2 endplate, a significant difference did exist between the most anterior and posterior aspects of the endplate. The posterior aspect was significantly more dense than was the anterior aspect. There was a decrease in mean B M D from the posterior to anterior aspect of the sub-endplate zone. This was confirmed by the endplate r B M D profile for both the Maximum and AnteriorPosterior profile analysis, as well as by the ratio of posterior endplate density to anterior endplate density. This difference in the endplate r B M D profile between level supports the hypothesis that sagittal alignment is influential on the regional bone mineral density. The L2 regional density differential can be attributed to its position in the lordotic spine. As discussed in the endplate strength profile section above, in accordance to Wolffs Law, the regional density difference may reflect endplate load distribution. The increase seen in density at the anterior aspect of the T9 and T12 endplates is consistent with anterior  93  body loading in the kyphotic thoracic spine. With disc degeneration, osteophyte formation is similarly known to occur at the anterior aspect of the vertebral endplate of T9 and T12, but not L2. To our knowledge, little work as been conducted previously on the difference in regional bone mineral density with changing spinal level.  4.2.2.2  Lateral Position  When looking at changes in regional sub-endplate trabecular bone density, an asymmetric profile was discovered. One side of the endplate appeared to be significantly more dense than the opposite side. Therefore, the relationship between the density of the central subendplate zone versus the lateral sub-endplate zone appears to be unknown, as on one side the endplate is of greater density than the centre, while on the other side it is of lesser density. This finding is not intuitive and is in contrasts with previous literature which has consistently shown symmetry in bone mineral density between sides (Antonacci et al, 1997; Banse et al, 2001; L u et al, 2000).  A more detailed analysis of the Maximum  profile map reveals that although asymmetry exists in the overall endplate r B M D profile relative to lateral position, important trends are identified. The posterolateral corners are the densest parts of the subcortical endplate. In the posterior row (row -2) the periphery is more dense than the centre.  In the central rows (row -1 and row 0), although  significant on only one side, the opposite trend was found. The centre is denser than is the periphery. This is confirmed in the Lateral profile analysis, which found a decrease in r B M D from the centre of the endplate toward the periphery. Unfortunately, in the anterior row of the Maximum profile analysis, the difference between the centre and periphery of the endplate can not be determined because the two peripheral regions are significantly different from one another.  However, the most anterior row of the  endplate, as identified in the Anterior-Posterior profile analysis, reveals no significant difference between the sides and the centre, although the periphery tended to be of less density than did the central aspect of the sub-endplate zone. Therefore, the periphery of the sub-endplate trabecular bone appears to be less dense than the centre, except for in the most posterior aspect, where the posterolateral corners cause the opposite trend. This summation of endplate symmetry is now consistent with previous literature (Antonacci et al, 1997; Banse et al, 2001; Lu et al, 2000).  94  4.2.3  Effect of Endplate  In all profile analyses the mean r B M D of the superior and inferior subcortical endplates were not significantly different.  4.2.4  Summary of the Endplate rBMD Profile  The density of the sub-endplate trabecular bone decreased from the posterior and anterior aspects towards the centre of the endplate. Generally, the density decreased from the centre towards the lateral periphery except in the posterior row, where the posterolateral corners were the densest of the entire endplate. Spinal level was shown to influence the endplate r B M D profile.  4.3  Correlation between the Endplate Strength Profile and r B M D Profile  The correlation between vertebral body strength, vertebral trabecular bone strength, and endplate strength with D X A B M D has been well established (Grant et al, 2001; Hasegawa et al, 2001; Hollowell et al, 1996; Hoshijima et al, 1997; Jost et al, 1998; Kumar and Doherty, 1994; Lin, 1985; Oxland et al, 1996; Oxland and Lund, 2000; Singer et al, 1995).  Q C T trabecular B M D has also been strongly correlated with  vertebral body strength in compression (Brassow et al, 1982; Brinckmann et al, 1989; Cody et al, 1991; Eriksson et al, 1989; Knop et al, 2001; McBroom et al, 1985; McCubbrey et al, 1995; Singer et al, 1995; Veenland et al, 1997). More recently, it has been shown that the use of multiple regional cancellous densities correlate more highly with vertebral body failure than does a single vertebral body density (Cody et al, 1991; Mirsky and Einhorn, 1998). Attempts have been made to correlate cores of trabecular bone density with failure load (Edwards et al, 1986; Keller et al, 1989; Lang et al, 1998). This study is the first to correlate small regions of sub-endplate trabecular B M D with local endplate indentation failure loads. A small but significant positive correlation was found for all three endplate profile analyses. When the influence of the apophyseal ridge was removed from the endplate strength profiles, the correlation was even stronger. Other factors then bone mass, such as microstructure and material properties of the  95  mineralised tissue itself, also contribute to skeletal strength and explain the low to moderate correlation found between density and strength.  Comparison of strength and r B M D profile maps, showed very similar trends relative to AP position. Both density and strength decreased from the posterior aspect of the endplate toward the centre, followed by a significant increase at the most anterior aspect of the endplate. The Maximum profile analysis identified the posterolateral corners to be the strongest and densest of the entire endplate.  Similar to previous authors, we attribute  this to the effect of the pedicle (Grant, 2000; L u et al, 2000). Banse et al. described a large vertical inhomogenity of trabecular density in the posterior aspect of the body (Banse et al, 2001). They attributed this to the effect of the pedicles, suggesting that the increase in trabecular density of the inferior body be required to compensate for the structural enhancement of the superior cortical body shell produced by the pedicle insertion.  This would be consistent with the findings of this study.  Although not  significantly different, in all profile analyses the inferior endplate tended to be stronger and of greater density than was the superior endplate.  Comparison of endplate profiles relative to lateral position showed similar trends for the majority of rows.  In the most posterior row the posterolateral corners were of  significantly greater density and strength relative to the posterocentral aspect.  No  difference in strength relative to lateral position was found for row -20% or row 0%>, while the density tended to decrease from the centre toward the periphery. For row +20%, the anterolateral corners were stronger than the centre of the endplate.  This  difference was attributed to the to the influence of the apophyseal ring. Unfortunately, the asymmetry on density between sides found for row +1, prevents its comparison with strength within this row.  Sagittal alignment had a similar influence on both the endplate strength and r B M D profiles. L2 was found to a have a greater density and strength in the posterior aspect of the endplate relative to the front, while for T9 the opposite trend was found. As discussed  96  above, Wolffs Law may account for the regional difference in endplate density and strength relative to spinal level.  One major discrepancy found when comparing r B M D and strength profiles for the thoracolumbar endplate is the effect produced by the interaction between lateral position and endplate.  As described earlier in the discussion, a significant difference exists  between the endplate strength profile relative to lateral position for the superior endplate compared to the inferior endplate.  While no significant difference in the shape of the  endplate r B M D profile was found. This again illustrates the importance of vertebral structure, both macroscopic and microscopic, in influencing vertebral strength (Andresen et al, 1999; Banse et al, 2002; Singer et al, 1995; Thomsen et al, 2002; Veenland et al, 1997), as well as highlights the fact that bone density does not completely explain vertebral strength.  In summary, this study found only a low to moderate correlation between r B M D and local strength.  However, comparison of endplate and r B M D profiles should much  similarity. The densest areas of the sub-endplate zone are the strongest aspects of the vertebral endplate. Furthermore, sagittal alignment influenced the endplate profiles in a similar manner. For L2 the posterior aspect of the endplate was significantly stronger and denser than the anterior aspect. The relative difference did not exist in the upper and middle thoracolumbar spine as represented by T9 and T12 respectively.  4.4  Study Strengths a n d Weaknesses  This study is novel in its mapping of the vertebral endplate r B M D profile and is the first to generate the thoracolumbar strength profile using indentation testing. To date no study has attempted to divide the endplate into small regions and correlate local density with local strength.  This study utilised only osteoporotic specimens, as they are most  applicable to the clinical problem of implant subsidence in the osteoporotic spine.  The pQCT imaging protocol contained some weaknesses.  The speed that the images  were acquired (30mm/sec) was slightly faster than the standard speed (lOmm/sec) used  97  for a 0.25 voxel size. This means that the images were slightly noisier (blurry) than ideal, and bone mineral density readings as a result had a larger standard deviation.  A  correlation was performed between rBMDs obtained from two images acquired at lOmm/sec and an image acquired at 30 mm/sec. The correlation between the images acquired at different speeds was very high with a Pearson r of 0.94 and 0.93. We were therefore satisfied with the quality of the images.  Another weakness in the methodology was that of human error. Some error occurred in the positioning of the vertebral body relative to the common X , Y , and Z axis during both the indentation testing and pQCT. As this error is a random error and a large number of specimens and tests were performed, the error would be expected to be normally distributed. However, this error may have resulted in some of the anomalies found in the analysis, such as the asymmetry in sides of the endplate r B M D profile relative to lateral position. A further confounding bias not accounted for in the methodology, is the effect of degenerative disc disease and osteophyte formation. Although specimens with gross sagittal and coronal malalignment or pathologic processes (such as tumour or infection) were excluded based on A P and lateral X-rays, degenerative disc disease was not considered. Degenerative disc disease has been shown to directly influence changes in bone density and strength (Antonacci et al, 1997; Grant et al, 2001; Hansson et al, 1980; Keller et al, 1989; Oxland et al, 1996; Simpson et al, 2001), and may have had a confounding influence.  Although only three primary questions were investigated in this study (what is the thoracolumbar endplate strength and r B M D profiles and what is the correlation between them?) multiple analyses were performed. It is well recognised that multiple analyses increases the risk of committing a type I error.  However, the use of A N O V A is  appropriate in this circumstance and is a significant improvement over the use of multiple student t - tests. A further concern regarding analysis methodology is the question of a study's power.  The number of specimens and endplates required for this study was  estimated based on the previous study of Grant et al (Grant, 2000). It was determined that five specimens per level were required. Therefore this study did meet the sample  98  size requirements. Although, type II errors are a concern when answers to multiple questions are sought and multiple analyses are performed.  Finally, the specimens were fresh frozen and only approximate a real in vivo situation. However, a previous comparison of thawed and fresh specimens has found no differences in stiffness or hysteresis (Smeathers and Joanes, 1988).  4.5  Future Work and Clinical Implications  Like any successful research, this body of work answers a number of important questions, but inspires even more. With a better understanding of the thoracolumbar endplate profile, the "surgical and implant factors" can be addressed. Anterior interbody fusion devices should now be designed with novel shapes and be better situated on the endplate to match the strength profile in hopes of improving the strength of the boneimplant interface. Implants should be shaped to distribute load into the posterolateral and anterolateral corners of the endplate.  This would maximize the strength of the bone-  implant interface to better prevent subsidence. Furthermore, implants should be positioned or designed to distribute load onto the ossified apophyseal ring, as this is one of the strongest regions of the endplate. The influence of sagittal alignment on the endplate strength and density profile can be further investigated by mapping the upper thoracic and cervical spine.  It is hypothesized that a further decline in the  posterior/anterior density and strength ratio will be found at the apex of the thoracic kyphosis.  99  5.0 Conclusion In conclusion, the thoracolumbar endplate strength profile revealed that 1. The anterior and posterior aspects of the endplate are significantly stronger than the centre. 2.  The lateral periphery is stronger than the centre.  3. The posterolateral corners are the strongest part of the thoracolumbar endplate. 4. The apophyseal ring is important in explaining the difference in strength between the centre and periphery of the endplate, apart from the increase in strength between the anterocentral aspect of the endplate and the posterior aspect of the endplate, which occurs within the confines of the apophyseal ring. 5. No significant difference exists in the mean strength of the vertebral endplate between level or superior/inferior endplate. 6. The strength profile is influenced by the sagittal alignment of the spine. L2 is relatively stronger in the posterior of the endplate, while T9 is relatively stronger in the anterior of the endplate.  In conclusion, the thoracolumbar endplate rBMD profile revealed that 1. The anterior and posterior aspects of the sub-endplate zone are significantly more dense than is the centre. 2. The posterolateral corners are the densest part of the thoracolumbar endplate 3. In the centre of the endplate no significant difference in density exists relative to lateral position 4. No significant difference exists in the mean density of the vertebral subendplate zone between level and endplate. 5. The r B M D profile is significantly influenced by the sagittal alignment of the spine. In the lumbar spine the anterior sub-endplate zone is relatively less dense than is the posterior aspect, while in the thoracic spine no significant differential in density relative to AP position existed.  100  In conclusion, a very low to moderate  correlation was  found between a  thoracolumbar endplate strength profile and reciprocal thoracolumbar endplate r B M D profile. The maps o f the endplate strength and r B M D profile are similar in shape, despite a rather low correlation.  101  6.0 Reference List Andresen,R., Haidekker,M.A., Radmer,S., and Banzer,D. ( 1 9 9 9 ) Computerized analysis of gray-value profiles in spongy and cortical bone. Clinical experience. Invest Radiol. 34, 572-578.  Antonacci,M.D., Hanson,D.S., Leblanc,A., and Heggeness,M.H. ( 1 9 9 7 ) Regional variation in vertebral bone density and trabecular architecture are influenced by osteoarthritic change and osteoporosis. Spine 22, 2 3 9 3 - 2 4 0 1 . Augat,P., Gordon,C.L., Lang,T.F., Iida,H., and Genant,H.K. ( 1 9 9 8 ) Accuracy of cortical and trabecular bone measurements with peripheral quantitative computed tomography (pQCT). Phys.Med.Biol 43, 2 8 7 3 - 2 8 8 3 . Banse,X., DevogelaerJ.P., and Grynpas,M. ( 2 0 0 2 ) Patient-specific microarchitecture of vertebral cancellous bone: a peripheral quantitative computed tomographic and histological study. Bone 30, 8 2 9 - 8 3 5 . Banse,X., Devogelaer,J.P., Munting,E., Delloye,C, Cornu,0., and Grynpas,M. ( 2 0 0 1 ) Inhomogeneity of human vertebral cancellous bone: systematic density and structure patterns inside the vertebral body. Bone 28, 5 6 3 - 5 7 1 . Bernhardt,M. and Bridwell,K.H. ( 1 9 8 9 ) Segmental analysis of the sagittal plane alignment of the normal thoracic and lumbar spines and thoracolumbar junction. Spine 14,717-721.  Boden,S.D. and Sumner,D.R. ( 1 9 9 5 ) Biologic factors affecting spinal fusion and bone regeneration. Spine 20, 1 0 2 S - 1 1 2 S . Brassow,F., Crone-Munzebrock,W., Weh,L., Kranz,R., and Eggers-Stroeder,G. ( 1 9 8 2 ) Correlations between breaking load and CT absorption values of vertebral bodies. EurJ.Radiol. 2, 9 9 - 1 0 1 . Brinckmann,P., Biggemann,M., and F£ilweg,D. ( 1 9 8 9 ) Prediction of the compressive strength of human lumbar vertebrae. Spine 14, 6 0 6 - 6 1 0 . BurkusJ.K. ( 2 0 0 2 ) Intervertebral fixation: clinical results with anterior cages. Orthop.Clin.North Am. 33, 3 4 9 - 3 5 7 . Burkus,J.K., DorchakJ.D., Sanders,D.L., and et al. ( 1 9 9 9 ) Single-level anterior lumbar interbody fusion using threaded cortical bone allografts. North American Spine Society Meeting, Miami. Closkey,R.F., Parsons,J.R., Lee,C.K., Blacksin,M.F., and Zimmerman,M.C. ( 1 9 9 3 ) Mechanics of interbody spinal fusion. Analysis of critical bone graft area. Spine 18, 1011-1015.  102  CodyJD.D., Goldstein,S.A., Flynn,M.J., and Brown,E.B. (1991) Correlations between vertebral regional bone mineral density (rBMD) and whole bone fracture load. Spine 16, 146-154. Dvorak,M.F. and Fisher,C.G. (1999) Revision spine surgery in the presence of osteroporosis. In Revision Spine Surgery (Edited by Margulies,J.Y., Aebi,M., and FarcyJ.C.) Pp. 646-653. Mosby Inc., St. Louis. Edwards,W.T., McBroom,R.C, Ffayes,W.C, Goldberg,R., and White,A.A. (1986) Variation of density in the vertebral body measured by quantitative computed tomography. Trans.Orthop.Res.Soc. 11,205. Edwards,W.T., Zheng,Y., Ferrara,L.A., and Yuan,H.A. (2001) Structural features and thickness of the vertebral cortex in the thoracolumbar spine. Spine 26, 218-225. Eriksson,S.A., Isberg,B.O., and LindgrenJ.U. (1989) Prediction of vertebral strength by dual photon absorptiometry and quantitative computed tomography. Calcif. Tissue Int. 44, 243-250. EvansJ.H. (1985) Biomechanics of lumbar fusion. Clin.Orthop. 38-46. Flynn,M.J. and Cody,D.D. (1993) The assessment of vertebral bone macro architecture with X-ray computed tomography. Calcif.Tissue Int. 53 Suppl 1, S170-S175. GalanteJ., Rostoker,W., and Ray,R.D. (1970) Physical properties of trabecular bone. Calcif.Tissue Res. 5, 236-246. Gill,K. (1989) Introduction to interbody fusion. In Lumbar interbody fusion (Edited by Lin,P.M. and Gill,K.) Pp. 3-7. Aspen, Maryland. GrantJ.P. (2000) Mapping the structural properites or the lumbosacral vertebral endplates. Master of Applied Science University of British Columbia. Grant,J.P., Oxland,T.R., and Dvorak,M.F. (2001) Mapping the structural properties of the lumbosacral vertebral endplates. Spine 26, 889-896. GrantJ.P., Oxland,T.R., Dvorak,M.F., and Fisher,C.G. (2002) The effects of bone density and disc degeneration on the structural property distributions in the lower lumbar vertebral endplates. J.Orthop.Res. 20, 1115-1120. Groll,0., Lochmuller,E.M., Bachmeier,M., WillneckerJ., and Eckstein,F. (1999) Precision and intersite correlation of bone densitometry at the radius, tibia and femur with peripheral quantitative CT. Skeletal Radiol. 28, 696-702. Guglielmi,G., Gluer,C.C, Majumdar,S., Blunt,B.A., and Genant,H.K. (1995) Current methods and advances in bone densitometry. Eur.Radiol. 5, 129-139.  103  Hacker,RJ. (2003) Comparison of interbody fusion approaches for disabling low back pain. Spine 22, 660-665. Hannson,T.H., Keller,T.S., and Panjabi,M.M. (2003) A study of the compressive properties of lumbar vertebral trabeculae: effects of tissue characteristics. Spine 11, 5662. Hansson,T. and Roos,B. (1981) The relation between bone mineral content, experimental compression fractures, and disc degeneration in lumbar vertebrae. Spine 6, 147-153. Hansson,T., Roos,B., and Nachemson,A. (1980) The bone mineral content and ultimate compressive strength of lumbar vertebrae. Spine 5, 46-55. Harkey,H.L. (1996) Osteoporosis, Osteomalacia and Paget's Disease. In Principles of Spinal Surgery (Edited by Menezes,A.H. and Senntag,V.K.H.) Pp. 493-503. McGrawHill, New York. Hasegawa,K., Abe,M., Washio,T., and Hara,T. (2001) A n experimental study on the interface strength between titanium mesh cage and vertebra in reference to vertebral bone mineral density. Spine 26, 957-963. Heggeness,M.H. (1997) Surgery of the osteoporotic spine. In The Adult Spine: Principles and Practice (Edited by FreymoyerJ.W.) Pp. 1639-1664. Lippincott-Raven Publishers, Philadelphia. Holdsworth,F. (1970) Fractures, dislocations, and fracture-dislocations of the spine. J.Bone Joint Surg.Am. 52, 1534-1551. HollowellJ.P., Vollmer,D.G., Wilson,C.R., Pintar,F.A., and Yoganandan,N. (1996) Biomechanical analysis of thoracolumbar interbody constructs. How important is the endplate? Spine 21, 1032-1036. Holte,D.C, O'BrienJ.P., and Renton,P. (1994) Anterior lumbar fusion using a hybrid interbody graft. A preliminary radiographic report. Eur.Spine J. 3, 32-38. Hoshijima,K., Nightingale,R.W., Yu,J.R., Richardson,W.J., Harper,K.D., Yamamoto,H., and Myers,B.S. (1997) Strength and stability of posterior lumbar interbody fusion. Comparison of titanium fiber mesh implant and tricortical bone graft. Spine 22,11811188. Jackson,R.P. and McManus,A.C. (1994) Radiographic analysis of sagittal plane alignment and balance in standing volunteers and patients with low back pain matched for age, sex, and size. A prospective controlled clinical study. Spine 19, 1611-1618. Jost,B., Cripton,P.A., Lund,T., Oxland,T.R., Lippuner,K., Jaeger,P., and Nolte,L.P. (1998) Compressive strength of interbody cages in the lumbar spine: the effect of cage shape, posterior instrumentation and bone density. Eur.Spine J. 7, 132-141.  104  Keller,T.S., Hansson,T.H., Abram,A-C, Spengler,D.M., and Panjabi,M.M. (1989) Regional variations in the compressive properties of lumbar vertebral trabeculae. Effects of disc degeneration. Spine 14, 1012-1019. Knop,C, Lange,U., Bastian,L., Oeser,M., and Blauth,M. (2001) Biomechanical compression tests with a new implant for thoracolumbar vertebral body replacement. Eur.Spine J. 10, 30-37. KostuikJ.P. and Heggeness,M.H. (1997) Surgery of the osteoporotic spine. In The Adult Spine (Edited by FrymoyerJ.W.) Pp. 1639. Lippincott-Raven Publishers, Philidelphia. KozakJ.A., Heilman,A.E., and O'BrienJ.P. (1994) Anterior lumbar fusion options. Technique and graft materials. Clin.Orthop. 45-51. Kumar,A. and Doherty,B.J. (1994) Biomechanical testing of vertebral endplate strength: a cadaver study. Kurowski,P. and Kubo, A. (1986) The relationship of degeneration of the intervertebral disc to mechanical loading conditions on lumbar vertebrae. Spine 11, 726-731. Kuslich,S.D., Ulstrom,C.L., Griffith,S.L., Ahern,J.W., and DowdleJ.D. (1998) The Bagby and Kuslich method of lumbar interbody fusion. History, techniques, and 2-year follow-up results of a United States prospective, multicenter trial. Spine 23, 1267-1278. Labrom,R.D. (2003) The effect of cage positioning on lumboscaral vertebral endplate failure in compression. Master of Science (Surgery) University of British Columbia. LaneJ.M. and Sandhu,H.S. (1997) Osteoporosis of the spine. In Orthopaedic Knowlegde Update. Spine (Edited by Garfin,A.R. and Vaccaro,A.R.) Pp. 227-234. American Academy of Orthopaedic Surgeons, Rosemont IL. Lang,S.M., Moyle,D.D., Berg,E.W., Detorie,N., Gilpin,A.T., Pappas,N.J., Jr., Reynolds,J.C, Tkacik,M., and Waldron,R.L. (1988) Correlation of mechanical properties of vertebral trabecular bone with equivalent mineral density as measured by computed tomography. J.Bone Joint Surg.Am. 70, 1531-1538. Lang,T., Augat,P., Majumdar,S., Ouyang,X., and Genant,H.K. (1998) Noninvasive assessment of bone density and structure using computed tomography and magnetic resonance. Bone 22, 149S-153S. Lim,T.H., Kwon,H., Jeon,C.H., Kim,J.G., Sokolowski,M., Natarajan,R., An,H.S., and Andersson,G.B. (2001) Effect of endplate conditions and bone mineral density on the compressive strength of the graft-endplate interface in anterior cervical spine fusion. Spine 26, 951-956. Lin,P.M. (1985) Posterior lumbar interbody fusion technique: complications and pitfalls. Clin.Orthop. 90-102.  105  Lu,W.W., Zheng,Y., Holmes,A., Zhu,Q., Luk,K.D., Zhong,S., and Leong,J.C. (2000) Bone mineral density variations along the lumbosacral spine. Clin.Orthop. 255-263. Lu,Y., Genant,H.K., ShepherdJ., Zhao,S., Mathur,A., Fuerst,T.P., and Cummings,S.R. (2001) Classification of osteoporosis based on bone mineral densities. J.Bone Miner.Res. 16, 901-910. Matsui,M. (1991) [An evaluation of bone mineral mass and trabecular distribution on cross sections of thoracic and lumbar vertebral bodies]. Nippon Seikeigeka Gakkai Zasshi 65, 9-17. Mazess,R.B., Gifford,C.A., BisekJ.P., Barden,H.S., and HansonJ.A. (1991) D E X A measurement of spine density in the lateral projection. I: Methodology. Calcif. Tissue Int. 49, 235-239. McAfee,P.C. (1999) Interbody fusion cages in reconstructive operations on the spine. J.Bone Joint Surg.Am. 81, 859-880. McBroom,RJ., Hayes,W.C, Edwards,W.T., Goldberg,R.P., and White,A.A., I l l (1985) Prediction of vertebral body compressive fracture using quantitative computed tomography. J.Bone Joint Surg.Am. 67, 1206-1214. McCubbrey,D.A., Cody,D.D., Peterson,E.L., KuhnJ.L., Flynn,M.J., and Goldstein,S.A. (1995) Static and fatigue failure properties of thoracic and lumbar vertebral bodies and their relation to regional density. J.Biomech. 28, 891-899. Mirsky,E.C. and Einhorn,T.A. (1998) Bone densitometry in orthopaedic practice. J.Bone Joint Surg.Am. 80, 1687-1698. Myers,B.S., Arbogast,K.B., Lobaugh,B., Harper,K.D., Richardson,WJ., and Drezner,M.K. (1994) Improved assessment of lumbar vertebral body strength using supine lateral dual-energy x-ray absorptiometry. J.Bone Miner.Res. 9, 687-693. Nepper-Rasmussen,J. and Mosekilde,L. (1989) Local differences in mineral content in vertebral trabecular bone measured by dual-energy computed tomography. Acta Radiol. 30, 369-371. Osteoporosis Society of Canada . About Osteoporosis, www.osteoporosis.ca . 2003. Ref Type: Electronic Citation Oxland,T.R., GrantJ.P., Dvorak,M.F., and Fisher,C.G. (2003) Effects of endplate removal on the structural properties of the lower lumbar vertebral bodies. Spine 28, 771777. Oxland,T.R. and Lund,T. (2000) Biomechanics of stand-alone cages and cages in combination with posterior fixation: a literature review. Eur.Spine J. 9 Suppl 1, S95-101.  106  Oxland,T.R., Lund,T., Jost,B., Cripton,P., Lippuner,K., Jaeger,P., and Nolte,L.P. (1996) The relative importance of vertebral bone density and disc degeneration in spinal flexibility and interbody implant performance. An in vitro study. Spine 21, 2558-2569. Pearcy,M.J. and EvansJ.H. (2003) The load bearing capacity of vertebral cancellous bone in interbody fusion of the lumbar spine. Engineering in Medicine 12, 183-184. Ray,C.D. (1997) Threaded titanium cages for lumbar interbody fusions. Spine 22, 667679. Sandhu,H.S., Turner,S., KaboJ.M., Kanim,L.E., Liu,D., Nourparvar,A., Delamarter,R.B., and Dawson,E.G. (1996) Distractive properties of a threaded interbody fusion device. A n in vivo model. Spine 21, 1201-1210. Shirazi-Adl, A. (1992) Finite-element simulation of changes in the fluid content of human lumbar discs. Mechanical and clinical implications. Spine 17, 206-212. Shirazi-Adl,S.A., Shrivastava,S.C, and Ahmed,A.M. (1984) Stress analysis of the lumbar disc-body unit in compression. A three-dimensional nonlinear finite element study. Spine 9, 120-134. Sievanen,H., Koskue,V., Rauhio,A., Kannus,P., Heinonen,A., and Vuori,I. (1998) Peripheral quantitative computed tomography in human long bones: evaluation of in vitro and in vivo precision. J Bone Miner Res 13, 871-882. Simpson,E.K., Parkinson,I.H., Manthey,B., and Fazzalari,N.L. (2001) Intervertebral disc disorganization is related to trabecular bone architecture in the lumbar spine. J.Bone Miner.Res. 16,681-687. Singer,K., Edmondston,S., Day,R., Breidahl,P., and Price,R. (1995) Prediction of thoracic and lumbar vertebral body compressive strength: correlations with bone mineral density and vertebral region. Bone 17, 167-174. Singer,K.P. and Breidahl,P.D. (1990) Vertebral body trabecular density at the thoracolumbar junction using quantitative computed tomography. A post-mortem study. Acta Radiol. 31, 37-40. SmeathersJ.E. and Joanes,D.N. (1988) Dynamic compressive properties of human lumbar intervertebral joints: a comparison between fresh and thawed specimens. J.Biomech. 21,425-433. Stagnara,P., De MauroyJ.C., Dran,G., Gonon,G.P., Costanzo,G., DimnetJ., and Pasquet,A. (1982) Reciprocal angulation of vertebral bodies in a sagittal plane: approach to references for the evaluation of kyphosis and lordosis. Spine 7, 335-342. Steffen,T., Tsantrizos,A., and Aebi,M. (2000a) Effect of implant design and endplate preparation on the compressive strength of interbody fusion constructs. Spine 25, 10771084.  107  Steffen,T., Tsantrizos,A., Frufh,L, and Aebi,M. (2000b) Cages: designs and concepts. Eur.SpineJ. 9 Suppl 1, S89-S94. Takada,M., Engelke,K., Hagiwara,S., Grampp,S., and Genant,H.K. (1996) Accuaracy and precision study in vitro for peripheral quantitative computed tomography. Osteoporos Int 6, 207-212. Thomsen,J.S., Ebbesen,E.N., and Mosekilde,L. (2000) A new method of comprehensive static histomorphometry applied on human lumbar vertebral cancellous bone. Bone 27, 129-138. ThomsenJ.S., Ebbesen,E.N., and Mosekilde,L. (2002) Zone-dependent changes in human vertebral trabecular bone: clinical implications. Bone 30, 664-669. Veenland,J.F., Link,T.M., Konermann,W., Meier,N., GrashuisJ.L., and Gelsema,E.S. (1997) Unraveling the role of structure and density in determining vertebral bone strength. Calcif.TissueInt. 61, 474-479. White,A.A. and Panjabi,M.M. (1990) Clinical biomechanics of the spine. J.B. Lippincott Company, Philadelphia. Yoganandan,N., Maiman,DJ., Pintar,F., Ray,G., MyklebustJ.B., Sances,A., Jr., and Larson,S.J. (1988) Microtrauma in the lumbar spine: a cause of low back pain. Neurosurgery 23, 162-168. Zaccaria,N. (2003) Biomechanical evaluation of troughed and untroughed vertebral bodies in anteiror thoracolumbar burst fractrue constructs. Zheng,Y., Lu,W.W., Zhu,Q., Qin,L., Zhong,S., and Leong,J.C. (2000) Variation in bone mineral density of the sacrum in young adults and its significance for sacral fixation. Spine 25, 353-357.  108  7.0 Appendix 1: Indentation Tests 7.1  Maximum Profile Analysis  1061  T9  s  52.3  50.8  38  46.7  70.1  25.7  1067  T9  28.8  s  62.2  57.7  64.1  51.9  64.5  1069  T9  49.6  s  139.9  37.6  160.8  137.7  86.4  159.4  46.9 81.1  1081  T9  81.6  82.3  s  67.7  68.8  45.8  75.2  120.1  39.3  1083  T9  58.1  s  71.5  89.5  71.9  78.7  76  126.5  55.2  1087  T9  43.6  s  54.6  78.2  69.4  79.3  99.1  80.7  1061  63.8  T9  51  72.3  I  30.6  58.4  38.3  70.8  63.9  1067  48.3  T9  52.2  30.5  I  138.1  80  45.7  53.8  60.6  1069  47.9  T9  66.2  57.6  I  132  157.7  70.3  176.4  168.7  113.2  1081  T9  123.8  90  I  83.5  62  77.6  51.7  51.2  1083  T9  55.6  I  95.6  94.4  64.5 54  75.2  144.3  63.9  78.4  1087  T9  47.8 63.5  96.4  106.4  110.1  188.5  71.7  1061  T12  I s  140.3  68.9  59.2  79.3  69.3  54.3  72.2  58.6  31  1067  T12  47.9  s  57.8  57.2  63.3  52.3  52.8  1069  T12  50.5  53.2  s  27  162.9  118.3  132.9  91.7  48.3 109.6  96.8  1081  84  T12  s  98.7  36.2  66.2  69.2  66.2  51.7  33.5  1087  T12  s  93.2  66.2  78  89.6  76.1  59.1  45 66.9  1090  T12  88.5  s  43  30.2  43.7  28.1  59.9  19.4  1061  T12  28.9  31.2  I  58.6  99.3  50  50.1  65.5  31  52.4  1067  T12  67.5  I  74.8  98.9  58.1  82.6  90.2  35.8  1069  51.8  T12  29.3  1  155.3  124  75.4  143.7  140  112.3  136.8  1081  118  T12  1  110.7  61.7  67.2  64.6  71.4  65.3  1087  55.3  T12  69.4  1  142.5  72.4  58.78  56.4  116.8  78.4  65.2  1090  T12  32.8  50.9 73.1  49.2  50.7  69.2  48.4  30.2  27.7  29.3 46.2  29.7  62  1061  L2  1 s  42.6  46  65.1  81.2  31.2  54.4  1067  L2  s  67.6  47.1  76.4  52.4  55  49.1  1069  L2  51.5  s  62.3  91.5  106.5  144.7  83.9  141.9  67.2  111.3  1081  L2  153.6  s  162.5  166.2  117.7  66.2  79.8  81.6  1083  L2  78.9  65.7  s  103.3  76.8  68.4  115.2  133.2  53.6  1087  52.6  L2  65.1  s  133.4  110.4  82  87.1  122.7  52.1  1061  L2  101.8  48.4  67.7  78.6  126.1  45.6  1067  62.8 41.4  62.4  1  L2  1  171.1  144.8  62.9  75.3  128.5  59.6  48.2  44.4  60.9  1069  L2  1  152.7  117.2  95.6  109.6  157.2  43.6  1081  74.6  L2  82.7  1  228.7  97  93.3  87.3  101.6  1083  131.6  L2  135.3  92.6  1  243.1  76.8  62  79.3  145.9  185.8  1087  L2  109.3  44  1  155.7  74.9  107.3  147.2  176.2  134.7  85  76.7  109  1061  T9  s  34.9  29.1  54.5  32.4  31.7  28.4  30.5  1067  T9  131.3  s  35.5  39.6  23.3  41.2  35.8  52.6  26.9  40.8  57.6  38.3  1069  T9  s  76.5  81.9  76.9  87.9  73.8  58.8  1081  129.6  T9  98.6  s  50.8  56  43  42.2  50.2  37.2  36.5  37.8  48.5  32.2 40.8  1083  T9  s  52.7  38.7  43.6  44.1  41.5  44.5  1087  38.2  T9  41.6  s  61.9  63.2  46.4  46.8  53.9  1061  56.6  T9  53.5  68.2  56  I  60.3  37.2  75.9  42.9  44.3  32.6  1067  70.4  T9  138.4  61  I  78  35.2  48  58.6  58.6  61.5  1069  68.4  T9  85.5  32.1  I  120.3  151.4  99.5  108.6  80.5  123  124.1  1081  T9  92.1  95.6  I  67.7  60.1  42.7  39.3  36.2  28.8  1083  T9  43.2  69.4  I  84.3  45.6  70.7  53.4  55.5  67.7  67.1  61.4  1087 1061  59.7  T9 T12  42.8  I s  51.1  46.9  44.7  30  38.4  42  74.2  38.3  42.8  34.3  38.1  33.5  57.5  51.7 31.5  28.9  1067  T12  71.4  s  41.7  28.4  35.9  32.9  50.6  38.6  37.9  45.9  1069  T12  83  s  144.3  35.6  91.8  62.5  83  64.6  61.8  82.5  83.1  87.8  1081  T12  s  25.5  25.1  37.7  35.2  24.5  T12  60.9  s  56.5 74.9  29.6  1087  44.6  67.4  49  39  63.5  63.1  50.7  1090  T12  76  s  49  29.4  20.3  14.7  20.2  25.4  1061  17.6  T12  27  61.5  20  72.4  49.1  41.8  28.8 40.2  51.2  63.6  1067  49.4  T12  48.6  1069  T12  1081  T12  1087  T12  1090  T12  1061  HHHH I HHH I  48.6  37.4  26.2  25.6  26.6  26.3  34.4  31.4  28.2  108.8  104.8  127.7  135.2  100.2  95.9  68.7  168.6  126  53.1  58.8  52.9  49.8  31.4  50.9  36.7  55.9  40.6  67.1  73.1  73.4  50.5  36.6  47.5  47.8  151.6  91.6  30.1  22.2  15.3 31.9  23.5  24.6  26.2  34.7  19.6  36  34.9  41.4  32.8  27.4  L2  I s  46.2  20.8 49.7  1067  L2  s  64.4  35.1  28.9  46  47.2  57.3  26.6  32.3  1069  L2  28.2  s  121.5  95.7  90.2  81.8  114.72  77.7  82.3  61.2  1081  155.2  L2  s  75.2  73.8  101.7  78.1  58.3  53.9  49.2  1083  L2  s  74.9  84.1  28.4  43.2  22.6  49.8  57  123.5 30.2  1087  L2  24.5  s  65.7  60.6  45.2  79.3  47.5  74.3  38.9  1061  64.3  L2  77.9  I  75.4  54.6  43.3  39.8  42.5  35.4  39.2  30.9  1067  34.2  L2  I  30.5  48.8  30.3  41  34  37.8  31.8  51.2  29.4  27.3  95.5  1069  L2  I  121.1  132.3  44.3  78.8  72.8  50.1  106.1  52.6  1081  L2  121.6  I  115  106.9  91.2  92.5  74.7  97.5  82.7  1083  L2  116.9  52.3  I  66.4  71.5  75.6  72.6  50  59.8  66  1087  73.6  L2  58.3  1  111.7  120  62.9  65  56.1  67.4  73.9  75.1  49.1  110  1061  T9  s  33.6  23.6  1067  s  40.7  36.7  82.2  1069  T9 T9  s  79.8  100.1  145.7  1081  T9  s  38.9  40.2  1083  T9  s  26.4  32.8  52.8 52.7  1087  T9  s  53  49.5  37.5  1061  T9  I  29.8  36.7  36.1  1067  T9  I  56.2  63.1  88  1069  T9  I  89.1  110.9  171.1  1081  T9  I  59.3  T9  I  33.3 41.9  47.2  1083  58.1  61.2  1087 1061  T9 T12  I s  47.6 50.8  55.2 33.4  62.8 37.8  1067  T12  s  35.7  39.9  52.7  1069  T12  s  79.4  75.2  81.1  1081  T12  s  40.2  43.4  64.4  1087  T12  s  40.1  61.3  48.8  1090  T12  s  18  26.4  30.6  1061  T12  I  45.9  30.4  44.9  1067  T12  I  28.4  25.1  56.5  1069  T12  I  128.6  142  175.5  1081  T12  I  33.4  36.2  50.8  1087  T12  I  34.3  44.6  56  1090  T12  I  21.9  18.1  43  24.2  1061  L2  s  46.8  36.7  37.6  1067  L2  s  30.6  31.1  46.4  1069  L2  s  121.9  94.4  111.37  1081  L2  s  66.7  58.4  60.9  1083  L2  s  41.5  35.7  56.3  1087  L2  44.3  44.5  45  1061  L2  s I  43.7  34.5  35.9  1067  L2  I  36.4  33.7  34.4  1069  L2  I  158.7  103.1  94.8  1081  L2  I  82.2  62.9  48.1  1083  L2  I  70.2  56.8  62.7  1087  L2  I  53  48.7  95  7.2  Anterior-Posterior Profile Analysis  1061  T9  s  50.8  38  46.7  28.8  29.7  1067  T9  34.9  s  32.4  57.7  31.7  64.1  28.4  49.6  37.6  35.5  35.8  52.6  26.9  1069  T9  s  160.8  137.7  51.9 86.4  81.6  82.3  76.5  1081  T9  s  87.9  68.8  73.8  45.8  75.2  58.8  58.1  71.5  56  50.2  37.2  36.5  1083  T9  s  71.9  78.7  76  43.6  54.6  1087  T9  52.7  s  44.1  69.4  41.5  79.3  44.5  99.1  1061  T9  58.4  61.9 60.3  53.9  70.8  72.3 30.5  46.8  38.3  51 52.2  42.9  44.3  56.6 32.6 61.5  1067  T9  1069  T9  1081  T9  I  1083  T9  I  94.4  1087  T9  96.4  I  80  45.7  53.8  66.2  57.6  78  157.7  58.6  70.3  58.6  176.4  123.8  90  120.3  108.6  62  80.5  64.5  123  77.6  55.6  47.8  67.7  39.3  36.2  54  75.2  28.8  78.4  63.5  84.3  55.5  67.7  67.1  106.4  110.1  68.9  59.2  51.1  30  38.4  51.7  1061  T12  I s  69.3  54.3  72.2  47.9  57.8  1067  T12  38.3  s  33.5  63.3  57.5  52.3  31.5  53.2  27  1069  T12  s  28.4  118.3  50.6  132.9  52.8 91.7  37.9  84  98.7  1081  T12  144.3  s  83  66.2  38.6 64.6  69.2  66.2  45  62  1087  T12  s  56.5  25.1  66.2  37.7  78  35.2  89.6  66.9  88.5  74.9  1090  T12  s  39  30.2  63.5  43.7  63.1  28.1  31.2  1061  T12  29.4  20.2  99.3  50  67.5  61.5  1067  T12  41.8  28.8 40.2  25.4  50.1  28.9 52.4  98.9  58.1  82.6  51.8  29.3  48.6  25.6  124  26.6  26.3  61.8  51.2  1069  T12  75.4  143.7  136.8  117.96  1081  T12  I  108.8  135.2  61.7  100.2  67.2  95.9  64.6  55.3  69.4  1087  T12  53.1  31.4  I  72.4  49.8  58.78  56.4  65.2  32.8  67.1  1090  T12  50.5  50.7  50.9 47.5  29.3  L2  30.1  42.6  54.4  46.2  46.2  1067  L2  15.3 31.9  23.5  46  69.2 65.1  27.7  1061  I s  49.2  36.6  s  47.1  76.4  52.4  51.5  62.3  64.4  1069  L2  s  106.5  144.7  83.9  111.3  153.6  121.5  1081  L2  s  162.5  117.7  66.2  78.9  65.7  75.2  36  24.6 34.9  46  47.2  57.3  81.8  114.72  77.7  78.1  58.3  53.9  1083  L2  s  76.8  68.4  115.2  52.6  65.1  74.9  1087  L2  43.2  s  22.6  110.4  49.8  82  87.1  62.8  62.4  65.7  1061  L2  48.4  47.5  67.7  74.3  78.6  41.4  60.9  75.4  1067  L2  I I  79.3 39.8  42.5  35.4  62.9  75.3  48.2  44.4  30.5  41  34  37.8  1069  L2  144.8 117.2  95.6  109.6  74.6  82.7  121.1  1081  L2  78.8  72.8  97  50.1  93.3  87.3  135.3  115  1083  L2  92.5  76.8  74.7  62  97.5  79.3  109.3  92.6 44  66.4  1087  L2  72.6  50  74.9  59.8  107.3  147.2  85  76.7  111.7  65  56.1  67.4  I  112  1061 1067  T9 T9  s  39.6  33.6  23.6  79.6  38.7  55  s  38.3  40.7  36.7  114.9  136.3  153.8  157.6  1069  T9  s  50.8  79.8  100.1  134.8 134.7  1081  T9  s  32.2  38.9  40.2  75.6  69.8  105.5  1083  T9  s  40.8  26.4  32.8  41.7  80.1  87.2  1087  T9  s  56  53  49.5  131  73.9  89.1  1061  T9  I  61  29.8  36.7  130.8  93.4  50.6  1067  T9  I  32.1  56.2  63.1  97.1  92.9  67.5  179.8  167.9  176.7  1069  T9  I  95.6  89.1  110.9  1081  T9  I  45.6  47.2  58.5  48.9  86  1083  T9  I  42.8  33.3 41.9  58.1  116.6  112.7  84.6  1087  T9  47.6  84.9  72.4  41.7  50.8  55.2 33.4  108.1  T12  I s  42.8  1061  77.7  54.5  39.7  1067  T12  s  35.6  35.7  39.9  109.5  115.4  72.7  1069  T12  s  87.8  79.4  75.2  103.4  113.8  135.7  1081  T12  s  44.6  40.2  43.4  85.5  82  61.6  1087  T12  s  49  40.1  61.3  66  52.1  60.6  1090  T12  s  20  18  26.4  40.2  50.7  75.5  1061  T12  I  48.6  45.9  30.4  63.6  66.1  60.2  1067  T12 T12  I  28.2  28.4  25.1  50.4  69.8  65.7  1069  I  126  128.6  142  209.1  224.2  223.3  1081  T12  I  40.6  33.4  36.2  76.9  49  40.1  1087  T12  I  91.6  34.3  44.6  203  132.5  81.2  1090  T12  I  19.6  21.9  18.1  40.2  48.7  62.4 47.8  1061  L2  s  27.4  46.8  36.7  43.1  48.3  1067  L2  s  28.2  30.6  31.1  72.2  49.6  61  1069  L2  s  155.2  121.9  94.4  142.9  129.5  97.7  1081  L2  s  95.5  66.7  58.4  246.3  114.4  74  1083  L2  s  24.5  41.5  35.7  43.3  41.4  37.7  1087  L2  s  77.9  44.3  44.5  86.6  74.1  64  1061  L2  I  34.2  43.7  34.5  99.5  57  99  1067  L2  I  29.4  36.4  33.7  59.7  39.8  41.3  1069  L2  I  121.6  158.7  103.1  232.5  184.7  212.6  1081  L2  I  52.3  82.2  62.9  85.1  90  91.5  1083  L2  I  58.3  70.2  56.8  70.8  85.2  80.3  1087  L2  I  49.1  53  48.7  106.9  107.1  91.2  7.3  L a t e r a l Profile Analysis  1061  T9  s  25.7  28.8  29.7  29.1  46.9  49.6  37.6  1069  T9 T9  s  34.9  1067  35.5  23.3  81.1  81.6  82.3  1081  T9  81.9  51.1  39.3  58.1  71.5  1083  T9  s  76.5  56  43  40.7  55.2  43.6  54.6  52.7  38.7  52.4  63.8  51  72.3  61.9  63.2  35.1  60.1  48.3  52.2  30.5  60.3  37.2  38.4  47.9  66.2  57.6  78  35.2  s s  s  61.8  1087  T9  1061  T9  1067 1069  T9 T9  1  111.9  113.2  123.8  90  1081  120.3  T9  151.4  1  52.7  51.2  55.6  47.8  1083  T9  67.7  60.1  71.4  53.3  1  63.9  78.4  63.5  1087  84.3  T9  70.7  63.4  71.7  68.9  59.2  51.1  T12  1 s  80  1061  46.9  45.9  108.7  31  47.9  57.8  38.3  1067  T12  34.3  81.8  31.8  50.5  53.2  27  1069  28.4  T12  s  35.9  87  59.7  96.8  84  98.7  1081  T12  91.8  43.6  104.5  33.5  45  62  1087  56.5  T12  s  144.3  s  29.6  64  55.4  59.1  66.9  88.5  1090  T12  67.4  63.8  1061  T12  s  74.9  1067  T12  1069  T12  1081  T12  1087  T12  1090  T12  1061  L2  1 s  I  s  133.6  36.7  19.4  28.9  31.2  29.4  20.3  42  54  1  31  52.4  67.5  61.5  72.4  26.4  35.8  51.8  29.3  48.6  37.4  81  1  86.2 156.7  112.3  136.8  118  108.8  104.8  96.9  1  95.4  65.3  55.3  69.4  53.1  58.8  47  1  145.6  78.4  65.2  32.8  67.1  73.1  82.7  41.4  30.2  27.7  29.3  30.1  20.8  55.1  53.3  31.2  54.4  46.2  46.2  49.7  60.7  s  1067  L2  65.3  49.1  51.5  62.3  64.4  1069  L2  35.1  s  92.8  103.3  67.2  111.3  153.6  1081  95.7  L2 L2  107.6  116.1  81.6  78.9  65.7  1083  s  121.5 75.2  73.8  s  89.2  86.9  53.6  52.6  65.1  74.9  1087  L2  s  84.1  90.7  103.3  52.1  62.8  62.4  65.7  60.6  82.1  1061  L2  1  60.6  45.6  41.4  60.9  75.4  1067  L2  54.6  61.7  1  100.1  59.6  48.2  44.4  30.5  48.8  85.7  1069  L2  1  73.2  43.6  74.6  82.7  121.1  1081  L2  132.3  138.8  1  138.3  131.6  135.3  92.6  1083  115  L2  106.9  99.1  1  171.7  185.8  109.3  44  66.4  1087  L2  71.5  117  1  139.2  134.7  85  76.7  111.7  120  84.4  I  114  7.4  Cancellous Anterior-Posterior Profile Analysis  1061  T9  s  28.8  29.7  34.9  32.4  31.7  28.4  39.6  1067  33.6  T9  s  49.6  37.6  35.5  35.8  52.6  26.9  38.3  1069  36.7  T9  s  40.7  81.6  82.3  76.5  87.9  73.8  58.8  50.8  1081  T9  79.8  s  100.1  58.1  71.5  56  50.2  37.2  36.5  32.2  1083  s  38.9  T9  40.2  43.6  54.6  52.7  44.1  41.5  44.5  40.8  1087  T9  26.4  s  32.8  51  72.3  61.9  46.8  53.9  56.6  56  1061  T9  I  53  49.5  52.2  30.5  60.3  42.9  44.3  32.6  61  1067  29.8  T9  36.7  I  66.2  57.6  78  58.6  58.6  61.5  32.1  56.2  1069  63.1  T9  I  123.8  90  120.3  108.6  80.5  123  95.6  89.1  110.9  23.6  1081  T9  I  55.6  47.8  67.7  39.3  36.2  28.8  45.6  T9  I  33.3  1083  47.2  78.4  63.5  84.3  55.5  67.7  67.1  42.8  1087  41.9  58.1  T9  68.9  59.2  51.1  30  38.4  51.7  42.8  47.6  55.2  1061  T12  I s  47.9  57.8  38.3  33.5  57.5  31.5  41.7  50.8  33.4 39.9  1067  T12  s  53.2  27  28.4  50.6  38.6  37.9  35.6  T12  s  35.7  1069  84  98.7  144.3  64.6  61.8  87.8  79.4  1081  T12  75.2  s  45  62  56.5  83 25.1  37.7  35.2  44.6  1087  43.4  T12  s  40.2  66.9  88.5  74.9  39  63.5  63.1  49  40.1  1090  T12  61.3  s  28.9  31.2  29.4  20.2  28.8  25.4  20  18  26.4  1061  T12  I  52.4  67.5  61.5  41.8  40.2  51.2  48.6  45.9  1067  T12  30.4  I  51.8  29.3  48.6  25.6  26.6  26.3  28.2  28.4  T12  I  25.1  1069  136.8  118  108.8  135.2  100.2  95.9  126  128.6  1081  T12  142  I  55.3  69.4  53.1  49.8  31.4  50.9  40.6  33.4  36.2  1087  T12  I  65.2  32.8  67.1  50.5  36.6  47.5  91.6  T12  I  34.3  1090  44.6  27.7  29.3  30.1  15.3  23.5  24.6  19.6  21.9  L2  s  18.1  1061  54.4  46.2  46.2  31.9  36  34.9  27.4  46.8  36.7  1067  L2  s  51.5  62.3  64.4  46  47.2  57.3  28.2  30.6  L2  s  31.1  1069  111.3  153.6  121.5  81.8  114.7  77.7  155.2  121.9  94.4  1081  L2  s  78.9  65.7  75.2  78.1  58.3  53.9  95.5  66.7  L2  s  58.4  1083  52.6  65.1  74.9  43.2  22.6  49.8  24.5  41.5  35.7  1087  L2  s  62.8  62.4  65.7  79.3  47.5  74.3  77.9  44.5  L2  I  44.3  1061  41.4  60.9  75.4  39.8  42.5  35.4  34.2  43.7  34.5  1067  L2  I  48.2  44.4  30.5  41  34  37.8  29.4  36.4  33.7  1069  L2  I  74.6  82.7  121.1  78.8  72.8  50.1  121.6  158.7  103.1  I  135.3  92.6  115  92.5  74.7  97.5  52.3  82.2  62.9  I I  109.3  44  66.4  72.6  50  59.8  58.3  70.2  56.8  85  76.7  111.7  65  56.1  67.4  49.1  53  48.7  1081  L2  1083  L2  1087  L2  115  7.5  Ratio Posterior Endplate/Anterior Endplate Analysis  1061  T9  s  0.892902481  1067  T9  s  0.466943005  1069  T9  s  0.920242098  1081  T9  s  0.902989239  1083  T9  s  1.27062201  1087  T9  s  0.83  1061  T9  I  0.572052402  1067  T9  I  1069  T9  I  0.881242718 0.806750572  1081  T9  I  1.052637022  1083  T9  I  0.88595094  1087  T9  I  1.450715901  1061  T12  s  1.164746946  1067  T12  s  0.552217742  1069  T12  s  1.046302069  1081 1087  T12  s  T12  s  0.758184199 1.353441522  1090  T12  s  1061  T12  I  0.738822115 1.022116904  1067  T12  I  1.305863367  1069  T12  I  1081  T12  0.58336887 1.357590361  1087  T12  I  0.643455724  1090  T12  I  1.064375413  1061  L2  s  1.327586207  1067  L2  s  0.9797593  1069  L2  s  0.921642799  1081  L2  s  0.817625949  1083  L2  s  2.435784314  1087  L2  s  1.430173565  1061  L2  I  0.992407045  1067  L2  I  2.482670455  1069  L2  I  0.602381708  1081  L2  I  1.368117029  1083  L2  I  1.541515023  1087  L2  I  1.300065531  8.0 Appendix 2: Regional Bone Mineral Density 8.1  M a x i m u m Profile Analysis  1061  T9  s  177  164  150  154  136  154  1067  T9  s  170  145  156  160  152  165  171  167  144  149  139  170  1069  T9  s  300  281  290  305  301  242  1081  T9  s  225  199  251  178  149  263  167  163  134  1083  T9  s  155  179  187  153  164  163  182  180  164  1087 1090  T9 T9  s s  194  T9  I  169 158 169  181  1061  202 143 166  191  156 132 164  T9  185 174 177  172  156  170 146 177  144  197 136 150 187  153 160 155  1067  183 148 185 167  157  1069  T9  176  160  I  310  251  174 142 193 159  202  276  291  1081  T9  269  219  176  165  273  I  166  218  159  183  144  1083  T9  190  196  169  I  187  166  168  173  172  177  1087  T9  180  224  177  I  198  178  197  203  205  185  1090  T9  162  174  193  138  130  133  139  137  1061  T12  I s  139  133  176  123  218  125  207  207  256  191  1069  T12  181  s  243  255  268  186  207  225  272  1081  T12  232  s  233  132  225  157  255  134  151  143  136  144  1087  T12  s  139  155  144  160  180  163  170  163  1090  T12  s  179  173  191  145  183  134  139  151  141  1061  T12  174  138  170  134  176  148  182  169  161  182  195  185  1067  T12  1069  T12  1081  179  179  164  180  183  156  152  146  222  132  I  254  206  264  317  243  T12  253  260  I  165  262  171  187  180  200  157  1087  T12  180  160  145 142  171  156  I  187  227  140  157  159  162  !  1090  T12  142  263 130  136  150  114  1061  127  L2  s  124  166  137  167  148  184  174  135  1067  L2  149  s  162  179  194  156  150  154  173  162  199  1081  L2  165  s  154  163  150  193  220  185  157  1083  L2  159  s  221  215  243  174  166  149  204  179  1087  L2  181  163  s  203  155  206  164  L2  206  s  218  130  226  137  228 141  220  1090  193 112  165  141  136  1061 1067  L2 L2  125  205  136  I  199  185  199  198  205  236  169  219  I  169  194 184  215  185  136  179  1081  L2  194  210  200  I  192  155  179  214  233  265  203  231  1083  L2  221  142  129  193  256  181  154  1087  L2  158  178  I  193  177  163  172  218  158  1090  L2  188  194  1  149  198  138  138  167  196  144  162  177  162  117  1061  T9  s  130  142  155  169  162  147  1067  142  T9  130  s  163  125  128  141  153  149  168  137  123  139  1069  T9  s  256  211  155  184  217  233  1081  T9  223  222  s  221  156  134  127  134  168  1083  173  T9  140  s  133  164  140  147  169  165  163  154  1087 1090 1061  T9 T9 T9  123  s s  115  153 142  124  133 142  155 140  221 141  270 151  129 242  146  151 166  I  151  191 141  160  181 152 167  158  182  1067  T9  151  162  150  134  I  171  185  177  166  136  1069  T9  185  206  234  197  I  233  246  222  238  224  264  265  254  1081  T9  I  157  144  154  151  161  T9  I  149  1083  175  168  168  169  181  191  168  178  1087  175  T9  180  227  180  199  I  157  162  169  171  174  1090  T9  140  152  135  173  130  138  126  123  122  147  144  169  148  1061  T12  I s  187  198  222  187  190  1069  T12  148  s  179  174  270  230  217  229  229  221  1081  T12  177  s  241  141  144  223  152  151  151  142  1087  149  T12  145  s  155  172  167  172  170  188  175  1090  T12  147  157  s  131  153  128  148  141  136  130  131  144  138  1061  T12  I  182  149  160  181  162  1067  160  T12  120  148  150  145  I  146  168  141  149  126  1069  173  T12  167  141  292  259  264  289  284  263  1081  T12  287  267  157  290  172  170  167  157  148  1087  186  T12  164  164  198  138  156  157  170  213  1090  T12  187  199  224  138  120  130  133  130  132  1061  140  L2  135  s  142  160  127  131  171  176  133  1067  L2  133  130  s  142  149  164  185  145  141  122  102  1081  L2  140  s  135  190  136  143  144  186  203  130  125  146  I  1083  L2  s  150  182  150  143  141  132  1087  L2  148  137  s  137  198  161  220  239  253  219  1090  L2  178  s  196  141  207  121  134  138  134  121  124  1061  L2  123  164  I  203  177  188  219  213  1067  185  L2  116  130  I  157  149  128  162  168  197  150  1081  132  L2  126  148  176  186  221  231  213  152  142  160  167  1083  L2  I  192  157  140  154  172  156  1087  126  L2  129  133  I  192  158  162  174  172  1090  L2  170  172  153  1  144  171  124  149  164  174  180  122  118  129  118  1061  T9  s  145  200  1067  T9  s  119  154  1069  T9  s  199  181  1081  T9  s  150  152  1083  T9  s  128  124  1087 1090  s s I  227 126 148  229 166 248  1067  T9 T9 T9 T9  I  185  170  1069  T9  I  239  198  1081  T9  I  173  183  1083  T9  I  187  184  1061  1087  T9  I  157  152  1090  T9  I  136  123  1061  T12  s  135  173  1069  T12  217  289  1081  T12  s s  154  172  1087  T12  s  148  151  1090  T12  s  132  135  1061  T12  I  144  158  1067  T12  I  154  157  1069  T12  I  269  279  1081  T12  I  150  168  1087  T12  I  251  269  1090  T12  I  116  122  1061  L2  s  159  119  1067  L2  s  115  113  1081  L2  s  187  215  1083  L2  s  126  121  1087  L2  s  232  198  1090  L2  s  123  124  1061  L2  I  129  160  1067  L2  I  134  122  1081  L2  I  158  157  1083  L2  I  138  133  1087  L2  I  175  167  1090  L2  I  121  139  8.2  Anterior-Posterior Profile Analysis  1061  T9  s  164  150  154  170  156  1067  T9  s  152  160  155  169  165  162  171  149  139  170  141  153  149 217  1069  T9  s  281  290  305  225  251  263  1081  s  155  178  184  149  167  155  153  1083  T9 T9  s  163  187  127  164  134  182  168  194  191  1087  T9  181  s  169  169  165  156  163  183  185  170  1090  T9  s  174  158  132  155  181  148  221  174  146  1061  T9  142  140  164  141  I  169  152  185  177  177  1067  T9  193  144  167  I  156  160  167  158  176  160  1069  T9  I  159  185  251  177  202  166  276  219  218  1081  T9  273  246  I  176  222  165  238  159  190  166  1083  T9  169  154  196  151  168  161  I  173  180  178  1087  T9  177  224  191  197  168  203  178  162  174  1090  T9  193  162  138  169  130  171  133  133  123  125  138  126  123  1061  T12  I s  218  207  207  181  243  1069  T12  s  186  198  268  222  207  187  225  233  225  1081  T12  255  s  157  217  134  229  151  229  144  139  1087  T12  144  s  152  160  151  151  180  163  179  191  183  172  170  188  1090  T12  s  145  134  139  174  138  134  1061  T12  141  136  148  182  182  195  185  1067  T12  HHH  148  176  160  179  181  164  162  180  152  146  132  1069  T12  168  141  149  I  254  206  264  253  262  1081  T12  260  171  264  289  187  284  180  180  160  156  1087  T12  I  170  167  171  157  263  187  157  159  1090  T12  162  156  130  170  136  127  124  137  1061  L2  I s  142  157  130  133  167  130  148  184  149  162  194  131  171  176  1067  L2  s  156  150  154  199  165  154  1081  L2  185  s  145  141  150  193  220  159  215  1083  L2  243  s  143  174  144  166  186  149  181  163  155  1087  L2  150  s  143  206  141  193  228  206  218  1090  L2  226  s  220  137  239  112  253  141  136  125  136  1061  L2  134  194  134  I  199  138  185  205  236  219  188  169  219  184  213  215  179  194  200  162  168  192  197  155  179  265  203  231  221  231  213  1067  L2  1081 1083  L2 L2  I  142  129  193  154  158  1087  L2  I  178  140  154  177  172  163  172  188  194  1090  L2  198  162  138  172  I  138  174  167  162  177  162  149  164  174  I  120  1061  T9  s  130  125  145  142  116  132  1067  T9  s  123  139  119  157  161  150  1069  T9  s  222  221  199  199  211  184  1081  T9  s  133  140  150  136  148  145  1083  T9  s  115  124  128  142  142  135  1087  T9  s  151  191  227  124  180  171  1090  T9  s  166  141  126  295  301  221  1061  T9  I  150  134  148  177  181  217  1067  T9  I  206  197  185  171  193  186  1069  T9  I  265  254  239  261  281  303  1081  T9  I  168  169  173  168  179  178  1083  T9  I  227  199  187  353  301  285  1087  T9  I  152  173  157  159  162  198  1090  T9  I  144  148  136  168  162  159  1061  T12  s  179  174  135  140  137  160  1069  T12  s  241  223  217  271  234  256  1081  T12  s  145  155  154  170  165  170  1087  T12  s  157  153  148  140  149  133  1090  T12  s  144  138  132  141  123  126  1061  T12  I  148  145  144  130  137  169  1067  T12  I  167  141  154  186  225  191  1069  T12  I  267  290  269  329  376  342  1081  T12  I  164  164  150  203  230  251  1087  T12  I  199  224  251  245  291  327  1090  T12  I  135  142  116  170  150  145  1061  L2  s  130  149  159  104  116  102  1067  L2  s  140  135  115  120  126  140  1081  L2  s  125  146  187  121  180  214  1083  L2  s  137  137  126  113  136  129  1087  L2  s  196  207  232  167  188  209  1090  L2  s  123  164  123  127  139  127  1061  L2  I  130  149  129  127  117  138  1067  L2  I  126  148  134  114  114  124  1081  L2  I  160  167  158  153  0  157  1083  L2  I  129  133  138  116  115  107  1087  L2  I  153  171  175  164  149  161  1090  L2  I  118  129  121  126  114  115  8.3  L a t e r a l Profile Analysis:  1061  T9  s  161  154  170  156  152  130  134  1067  T9  s  119  144  149  139  170  163  137  1069  T9  s  217  242  225  251  263  256  238  1081  T9  s  143  134  155  153  163  156  137  1083  T9  s  122  164  194  191  181  164  115  1087  T9  s  128  153  185  170  174  153  137  1090  T9  s  169  160  174  146  142  142  119  1061  T9  I  165  155  177  177  193  146  161  1067  T9  I  163  157  176  160  159  162  163  1069  T9  I  248  269  219  218  273  234  250  1081  T9  I  126  144  190  166  169  157  186  1083  T9  I  148  177  180  178  177  168  157  1087  T9  I  160  185  162  174  193  180  148  1090  T9  167  137  133  123  125  135  125  1061  T12  I s  175  191  181  243  186  169  269  1069  T12  s  194  232  233  225  255  270  181  1081  T12  s  137  136  144  139  144  141  140  1087  T12  s  131  179  191  183  172  172  1090  T12  s  125  163 141  174  138  134  131  136  1061  T12  160  161  182  195  185  182  174  1067  T12  I  182  156  152  146  132  150  168  1069  T12  I  234  243  253  262  260  292  291  1081  T12  I  170  157  180  160  156  157  167  1087  T12  I  139  140  157  159  162  198  202  1090  T12  I  149  114  127  124  137  138  167  1061  L2  s  172  135  149  162  194  160  185  1067  L2  s  178  162  199  165  154  142  141  1081  L2  s  148  157  159  215  243  190  227  1083  L2  s  159  179  181  163  155  150  165  1087  L2  s  180  164  206  218  226  198  204  1090  L2  s  141  141  136  125  136  141  141  1061  L2  I  162  198  205  236  219  203  155  1067  L2  I  141  136  179  194  200  157  L2  I  155  1081  132  233  265  203  231  176  173  1083  L2  I  182  181  154  158  178  192  177  1087  L2  I  166  158  188  194  198  192  184  1090  L2  1  135  144  162  177  162  144  135  122  8.4  Anterior-Posterior Profile Analysis  1061  T9  s  1.201538462  1067  T9  s  1.035897436  1069  T9  s  1.491919192  1081  s  1.197202797  1083  T9 T9  s  1.277326969  1087  T9  s  1.145684211  1090  T9 T9  s I  0.526560588  1061 1067  T9  I  0.901090909  1069  T9  I  0.944378698  1081  T9  I  0.970285714  1083  T9  I  0.572523962  1087  T9  I  1.187283237  1090  T9  I  0.833128834  1061  T12  s  1.460869565  1069  T12  s  0.967411301  0.87026087  1081  T12  s  0.851881188  1087  T12  s  1.177251185  1090  T12  s  1.141538462  1061  T12  I  1.162844037  1067  T12  I  0.882059801  1069  T12  I  0.723782235  1081  T12  I  0.792105263  1087  T12  I  0.690382387  1090  T12  I  0.903225806  1061  L2  s  1.563354037  1067  L2  s  1.262176166  1081  L2  s  1.061359223  1083  L2  s  1.450793651  1087  L2  s  1.117021277  1090  L2  s  1.045801527  1061  L2  I  1.542408377  1067  L2  I  1.571590909  1081  L2  I  1.153846154 1.670414201  1083  L2  I  1087  L2  I  1.16835443  1090  L2  I  1.331830986  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0091010/manifest

Comment

Related Items