Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The influence of adiposity on bone quality in children, adolescents and young adults Hoy, Christa Leigh 2012

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

Item Metadata

Download

Media
24-ubc_2012_fall_hoy_christa.pdf [ 4.41MB ]
Metadata
JSON: 24-1.0073104.json
JSON-LD: 24-1.0073104-ld.json
RDF/XML (Pretty): 24-1.0073104-rdf.xml
RDF/JSON: 24-1.0073104-rdf.json
Turtle: 24-1.0073104-turtle.txt
N-Triples: 24-1.0073104-rdf-ntriples.txt
Original Record: 24-1.0073104-source.json
Full Text
24-1.0073104-fulltext.txt
Citation
24-1.0073104.ris

Full Text

THE INFLUENCE OF ADIPOSITY ON BONE QUALITY IN CHILDREN, ADOLESCENTS AND YOUNG ADULTS  by CHRISTA LEIGH HOY BHK, The University of British Columbia, 2010  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  MASTER OF SCIENCE in THE FACULTY OF GRADUATE STUDIES (Experimental Medicine)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  August 2012 © Christa Leigh Hoy, 2012  Abstract Introduction: Overweight children have greater bone mass than their healthy weight peers; however, they sustain more fractures. Thus, there is a need to better understand the relation between body fat and bone strength and aspects of bone quality such as bone microstructure that contribute to bone strength. Methods: I aimed to determine the cross-sectional relationship between fat mass (FM) and aspects of bone quality (strength, geometry, density and microstructure) at the distal radius and distal tibia in boys (n = 137, 15.6 ± 3.3 yrs on average) and girls (n = 157, 14.5 ± 3.9 yrs) in the context of the functional model of bone development (after adjusting for lean mass (LM)). FM and LM were measured using dual energy X-ray absorptiometry and bone quality was measured using high resolution-peripheral quantitative computed tomography. Results: In boys, FM negatively predicted bone strength at the radius but not at the tibia. Conversely, FM did not significantly predict bone strength in girls. In both boys and girls, FM negatively predicted total area at the radius but not the tibia. In girls but not boys, FM positively predicted cortical bone mineral density at the tibia but not the radius. For bone microstructure, FM did not significantly predict many variables; however, FM negatively predicted cortical thickness at the tibia in boys, trabecular thickness at the tibia in girls, and cortical porosity at both the radius and tibia in girls. In nearly all cases, LM mediated the relationship between FM and bone quality, whereby prior to adding LM to regression models FM positively predicted bone quality; however, after adjusting for LM the positive associations became non-significant or negative. Conclusions: The relation between fat and bone is complex and appears to be sex- and sitespecific. My results also highlight the important influence of lean mass on bone strength. Longitudinal studies with larger cohorts and more overweight and obese participants would clarify the sex-specific muscle-fat-bone relationships during growth and into adulthood. The potentially hazardous influence of high levels of fat mass on child and youth bone health must be considered among the adverse consequences of overweight and obesity.  ii  Preface The overall study was designed and implemented under the leadership of Dr. Heather McKay; data were collected in 2009 by the Healthy Bones III Study measurement team. Ethical approval was obtained from the University of British Columbia Clinical Research Ethics Board (H0702013, approved January 2009).  Abstracts Christa Hoy, Heather Macdonald, Heather McKay. Adiposity and cortical bone microstructure in girls and young women: A high-resolution pQCT study. Presented as a poster at the 2012 Canadian Obesity Student Meeting in Edmonton AB, Canada, June 20-23.  Hoy CL, Nettlefold L, Moore SA, Donaldson MG, Macdonald HM, McKay HA. A comparison of trabecular bone microstructure at the distal radius and distal tibia between healthy weight and overweight adolescents: An HR-pQCT study. Presented as a poster at the 2011 The American Society for Bone and Mineral Research Annual Meeting in San Diego CA, United States, September 16-20. Data analyzed for this abstract are separate from the data analyzed for this thesis; however, similar concepts are discussed.  Manuscript Christa Hoy, Heather Macdonald, Heather McKay. How does bone quality differ between healthy weight and overweight adolescents? Accepted for publication in Clinical Orthopedics and Related Research on August 17, 2012. Data analyzed for this manuscript are separate from the data analyzed for this thesis; however, similar concepts are discussed.  iii  Table of Contents  Abstract .......................................................................................................................................... ii Preface ........................................................................................................................................... iii Table of Contents ......................................................................................................................... iv List of Tables .............................................................................................................................. viii List of Figures ................................................................................................................................ x List of Abbreviations .................................................................................................................. xii Acknowledgements ..................................................................................................................... xv Chapter 1 - Introduction .............................................................................................................. 1 Chapter 2 – Literature Review .................................................................................................... 3 2.1  Bone Biology.................................................................................................................... 3  2.1.1  Bone Types ............................................................................................................... 3  2.1.2  Bone Growth ............................................................................................................. 4  2.1.3  Mechanical Integrity of Bone ................................................................................... 7  2.2  Measuring Bone in Children and Adolescents ............................................................... 13  2.2.1  Dual Energy X-Ray Absorptiometry ...................................................................... 14  2.2.2  Peripheral Quantitative Computed Tomography .................................................... 15  2.2.3  High-Resolution pQCT (HR-pQCT) ...................................................................... 16  2.2.3.1 Finite Element Analysis ...................................................................................... 17 2.3  Determinants of Bone Strength in Childhood and Adolescence .................................... 18  2.3.1  Genetics................................................................................................................... 18  2.3.2  Ethnicity .................................................................................................................. 19  2.3.3  Hormones ................................................................................................................ 20  2.3.4  Lifestyle Factors...................................................................................................... 23  2.3.4.1 Physical Activity ................................................................................................. 23 iv  2.3.4.2 Diet ...................................................................................................................... 24 2.3.4  The Mechanostat and the Functional Model of Bone Development ...................... 25  2.3.4.1 The Muscle-Bone Relationship ........................................................................... 26 2.3.4.2 Measuring Muscle ............................................................................................... 28 2.4  The Fat-Bone Relationship............................................................................................. 29  2.4.1  Cellular Mechanisms .............................................................................................. 29  2.4.2  Measuring Adiposity in Children and Adolescents ................................................ 30  2.4.2.1 Body Mass Index ................................................................................................. 31 2.4.2.2 Dual-energy X-ray Absorptiometry .................................................................... 32 2.4.3  Studies of the Fat-Bone Relationship using DXA in Children and Adolescents .... 34  2.4.4  Studies of the Fat-Bone Relationship using pQCT in Children and Adolescents .. 41  2.5  Summary of the Literature ............................................................................................. 49  Chapter 3 - Rationale, Objectives and Hypotheses.................................................................. 50 Chapter 4 – Methods .................................................................................................................. 52 4.1  Study Design .................................................................................................................. 52  4.2  The Healthy Bones III Study Cohort.............................................................................. 52  4.2.1  Cohort Description and Participant Recruitment .................................................... 52  4.2.2  Participant Scheduling and Measurement ............................................................... 53  4.3  Measurements................................................................................................................. 54  4.3.1  Anthropometry ........................................................................................................ 54  4.3.2  Maturity................................................................................................................... 54  4.3.3  Health History Questionnaire ................................................................................. 55  4.3.4  Body Composition .................................................................................................. 55  4.3.5  Bone Strength, Geometry, Density and Microstructure.......................................... 57  4.3.5.1 HR-pQCT Scan Acquisition ............................................................................... 57 4.3.5.2 Scan Analysis ...................................................................................................... 58 v  4.4  Statistical Analysis ......................................................................................................... 63  4.4.1  Inclusion and Exclusion Criteria ............................................................................. 63  4.4.2  Data Exploration ..................................................................................................... 63  4.4.3  Descriptive Analyses .............................................................................................. 64  4.4.4  Regression Analyses ............................................................................................... 64  Chapter 5 – Results ..................................................................................................................... 66 5.1  Cohort Characteristics .................................................................................................... 66  5.1.1  Participants .............................................................................................................. 66  5.1.2  Descriptive Characteristics of the Sample .............................................................. 68  5.2  Relationship Between Body Fat and Bone Quality ........................................................ 70  5.2.1  Unadjusted .............................................................................................................. 70  5.2.2  Regression Models .................................................................................................. 73  5.2.2.1 Bone Strength ...................................................................................................... 73 5.2.2.2 Bone Geometry ................................................................................................... 74 5.2.2.3 Bone Density ....................................................................................................... 74 5.2.2.4 Bone Microstructure ............................................................................................ 75 Chapter 6 - Discussion ................................................................................................................ 81 6.1  Overview of Findings ..................................................................................................... 81  6.2  Adiposity and Bone Quality ........................................................................................... 81  6.2.1  Bone Strength.......................................................................................................... 81  6.2.2  Bone Geometry ....................................................................................................... 84  6.2.3  Bone Density........................................................................................................... 86  6.2.4  Bone Microstructure ............................................................................................... 87  6.3  Implications of Findings................................................................................................. 89  6.3.1 6.4  Utility of HR-pQCT in Paediatric Bone Research .................................................. 89  Limitations ..................................................................................................................... 91 vi  6.5  Unique Contributions and Strengths .............................................................................. 92  6.6  Future Directions ............................................................................................................ 93  Chapter 7 - Summary and Conclusions .................................................................................... 94 7.1  Primary Objective: Bone Strength ................................................................................. 94  7.2  Secondary Objective: Bone Geometry, Density and Microstructure ............................. 94  7.3  Conclusions .................................................................................................................... 95  References .................................................................................................................................... 96 Appendices ................................................................................................................................. 111 Appendix A: Consent Forms ................................................................................................... 111 Appendix B: Questionnaires ................................................................................................... 126 Appendix C: Additional Data .................................................................................................. 137  vii  List of Tables Table 1. Summary of advantages and disadvantages of body composition measurement methods. ....................................................................................................................................................... 31 Table 2. Summary of studies that compared percent body fat (%BF) measured by DXA with %BF measured by 4-compartment (4C) models. .......................................................................... 33 Table 3. Summary of studies that used DXA to assess the fat-bone relationship. ....................... 36 Table 4. Summary of studies that used pQCT to assess the fat-bone relationship in children and adolescents. ................................................................................................................................... 43 Table 5. Demographic, maturity and anthropometric outcomes for boys and girls. Values are mean (SD), minimum and maximum unless otherwise indicated. ............................................... 68 Table 6. Bone outcomes (assessed by HR-pQCT) for boys and girls at the distal radius and distal tibia. Values are mean (SD), minimum and maximum. ............................................................... 69 Table 7. Model 1: Hierarchical multivariable linear regression models to demonstrate the independent contribution of fat mass to bone variables at the distal radius and distal tibia in boys (controlled for age, limb length and ethnicity) and girls (controlled for centered age, centered age squared, limb length and ethnicity: menarcheal status added to models for Tt.BMD, Ct.BMD, Ct.Po). ........................................................................................................................................... 78 Table 8. Model 2: Hierarchical multivariable linear regression models to demonstrate the independent contribution of fat mass to bone variables prior to adjusting for lean mass at the distal radius and distal tibia in boys (controlled for age, limb length, ethnicity and lean mass) and girls (controlled for centered age, centered age squared, limb length, ethnicity and lean mass: menarcheal status added to models for Tt.BMD, Ct.BMD, Ct.Po). ............................................. 79 Table 9. Model 3: Hierarchical multivariable linear regression models to demonstrate the independent contribution of lean mass to bone variables at the distal radius and distal tibia in boys (controlled for age, limb length, ethnicity and fat mass) and girls (controlled for centered age, centered age squared, limb length, ethnicity and fat mass: menarcheal status added to models for Tt.BMD, Ct.BMD, Ct.Po). ...................................................................................................... 80 Table 10. Univariate regression R values of fat mass, lean mass, height, weight, limb length, age and ethnicity with bone variables at the distal radius and distal tibia for BOYS. Ethnicity is Caucasian compared with Asian. ................................................................................................ 137  viii  Table 11. Univariate regression R values of fat mass, lean mass, height, weight, limb length, age and ethnicity with bone variables at the distal radius and distal tibia for GIRLS. Ethnicity is Caucasian compared with Asian. ................................................................................................ 138 Table 12. Unstandardized multivariable regression model coefficients (95% confidence interval) and R2 for each model at the distal radius for BOYS. ................................................................ 139 Table 13. Unstandardized multivariable regression model coefficients (95% confidence interval) and R2 for each model at the distal tibia for BOYS. ................................................................... 140 Table 14. Unstandardized multivariable regression model coefficients (95% confidence interval) and R2 for each model at the distal radius for GIRLS. ............................................................... 141 Table 15. Unstandardized multivariable regression model coefficients (95% confidence interval) and R2 for each model at the distal tibia for GIRLS. .................................................................. 142  ix  List of Figures Figure 1. Schematic of a long bone during embryonic development, including the growth plate. The right image shows the proliferation of chondrocytes in the growth plate. Figure from Wallis GA. Curr Biol. 1996;6(12):1577-1580, [35] with permission. ....................................................... 5 Figure 2. Hierarchical structural organization of bone. Figure from Rho J, et al. Med Eng Phys. 1998,20:92-102, [41] with permission. ........................................................................................... 8 Figure 3. Stress-strain curve. E = elastic modulus, which is the slope of the stress-strain curve before yield. Figure from Currey JD. Calcif Tissue Int. 2001,68:205-210, [43] with permission. 8 Figure 4. Schematic of a long bone and its properties measured by DXA and pQCT. The bending (CSMIx, CSMIy) and torsional (CSMIz) cross-sectional moments of inertia are calculated as the sum of the products of the area of each pixel (Ai) and the squared distance (dx, dy, dz) to the corresponding bending (x,y) or torsion (z) axis. Adapted from Macdonald HM. Thesis, University of British Columbia. 2006, [47] with permission. ...................................................... 10 Figure 5. The relation between bone geometry and bone strength. Figure adapted from Leonard MB, et al. Bone. 2004;34(6):1044-1052, [46] with permission.................................................... 10 Figure 6. Stress-strain curve of destructive loading of cadaveric distal radii to determine linear and elastic failure regions. P = platen force. Figure from MacNeil JA, et al. Bone. 2008;42:12031213, [18] with permission. .......................................................................................................... 11 Figure 7. Sample DXA spine, whole body and hip scan (A); sample pQCT distal tibia scan (B); sample HR-pQCT tibia scan (integral, C) and showing the cortical and trabecular structures (D). ....................................................................................................................................................... 13 Figure 8. Influence of parathyroid hormone on serum calcium. Figure from Biomedical Hypertextbooks http://www.vivo.colostate.edu/hbooks/pathphys/endocrine/thyroid/pth.html, retrieved July 15, 2012 with permission from author. .................................................................. 21 Figure 9. Functional model of bone development based on mechanostat theory. Adapted from Rauch F, et al. Pediatr Res. 2001;50(3):309-314, [7] with permission. ....................................... 26 Figure 10. Velocity curves that depict total body lean body mass (LBM) and bone mineral content (BMC) accrual during the pubertal growth spurt. Note the difference in magnitude with boys’ accrual being higher than girls’ at peak and the difference in timing with girls achieving peak values almost 2 years in advance of boys. Figure from Rauch F, et al. Bone. 2004;32:364377, [105] with permission. .......................................................................................................... 27 Figure 11. Sample image of an analyzed total body DXA scan of a 12-year old girl. ................ 56 x  Figure 12. Set up for HR-pQCT radius (A, C) and tibia (B, D) scans. ........................................ 57 Figure 13. Scout view images of the distal radius (A) and distal tibia (B) including the relative regions of interest (ROI). .............................................................................................................. 58 Figure 14. Sample images of a motion grade 1 scan (left) and a motion grade 4 scan (right). ... 59 Figure 15. Summary of the filters and parameters implemented in the dual threshold algorithm for analysis of a 3D dataset. Figure from Buie HR, et al. Bone. 2007;41(4):505-515, [63] with permission. .................................................................................................................................... 61 Figure 16. Flow chart that describes participant exclusions from data analysis. ......................... 67 Figure 17. Scatter plots of fat mass and lean mass vs. failure load (FLoad), total area (Tt.Ar), total density (Tt.BMD) and trabecular bone volume fraction (BV/TV) at the radius for BOYS. 71 Figure 18. Scatter plots of fat mass and lean mass vs. failure load (FLoad), total area (Tt.Ar), total density (Tt.BMD) and trabecular bone volume fraction (BV/TV) at the radius for GIRLS. 72  xi  List of Abbreviations Abbreviation  Definition  4C  Four-compartment model of measuring body composition.  μCT  Micro computed tomography.  %BF  Percent body fat (%).  aBMD  Areal bone mineral density (g/cm2) as measured by DXA.  AS! BC  Action Schools! British Columbia.  BA  Bone area (cm2) as measured by DXA.  BMAD  Bone mineral apparent density (g/cm3) as measured by DXA.  BMC  Bone mineral content (g) as measured by DXA and pQCT.  BMD  Volumetric bone mineral density (g/cm3) as measured by pQCT and HRpQCT.  BMI  Body mass index (kg/m2).  BSI  Bone strength index (mg2/mm4) as calculated with pQCT outcomes of Tt.Ar and Tt.BMD (BSI = Tt.Ar ∙ Tt.BMD2).  BV/TV  Trabecular bone volume to tissue volume, or trabecular bone volume fraction as measured by HR-pQCT.  CDC  Centers for Disease Control.  CSMI  Cross-sectional moment of inertia (cm4) as measured by HSA or pQCT.  Ct.Ar  Cortical area (mm2) as measured by pQCT and HR-pQCT.  Ct.BMD  Cortical bone mineral density (mgHA/cm3) as measured by pQCT and HRpQCT.  Ct.Po  Cortical porosity (%) as measured by FEA.  Ct.Th  Cortical thickness (mm) as measured by pQCT and HR-pQCT.  DXA  Dual energy X-ray absorptiometry. The Hologic QDR 4500W model is used in this thesis. (Hologic QDR 4500 W, Hologic Inc., Waltham, MA, USA)  E  Average stiffness or Young’s modulus.  EC  Endosteal circumference.  FA  Fat cross-sectional area (mm2) as measured by pQCT.  FEA  Finite element analysis.  FLoad  Failure load (N) as measured by FEA. xii  Abbreviation  Definition  FM  Fat mass (kg) as measured by DXA.  GH  Growth hormone.  HA  Hydroxyapatite.  HBS III  Healthy Bones III Study.  HR-pQCT  High-resolution peripheral quantitative computed tomography. The XtremeCT; Scanco Medical, Brüttisellen, Switzerland is used in this thesis.  HW  Healthy weight. Classification based on BMI-for-age or % body fat.  HSA  Hip structure analysis. Refers to the computer algorithm applied to proximal femur DXA images to estimate bone structural variables.  IGF-1  Insulin-like growth factor-1.  Imax  Maximal moment of inertia.  Imin  Minimal moment of inertia.  IOTF  International Obesity Task Force.  LM  Bone mineral free lean mass (kg) as measured by DXA.  Load:Str  Load-to-strength ratio. Calculated as fall load/FLoad.  MCSA  Muscle cross-sectional area (mm2) as measured by pQCT.  MES  Minimum effective strain.  MRI  Magnetic resonance imaging.  NHANES  National Health and Nutrition Examination Survey.  NHES  National Health Examination Survey.  OB  Obese. Classification based on BMI-for-age or % body fat.  OW  Overweight. Classification based on BMI-for-age or % body fat.  PC  Periosteal circumference.  PHV  Peak height velocity.  pQCT  Peripheral quantitative computed tomography.  PTH  Parathyroid hormone.  PVE  Partial volume effects.  ROI  Region of interest.  SSIp  Polar strength-strain index (mm3) as measured by pQCT.  Tb.BMD  Trabecular bone mineral density (mg HA/cm3) as measured by pQCT and HRpQCT. xiii  Abbreviation  Definition  Tb.N  Trabecular number (1/mm) as measured by HR-pQCT.  Tb.Sp  Trabecular separation (mm) as measured by HR-pQCT.  Tb.Th  Trabecular thickness (mm) as measured by HR-pQCT.  Tt.Ar  Total bone cross-sectional area (mm2) as measured by pQCT and HR-pQCT.  Tt.BMD  Total bone mineral density (mgHA/cm3) as measured by pQCT and HRpQCT.  UStress  Ultimate stress (MPa), as assessed by FEA.  VCHRI  Vancouver Coastal Health Research Institute.  xiv  Acknowledgements This thesis would not have been possible without the expert guidance and support of a number of individuals. First and foremost, I need to thank my two supervisors, Dr. Heather McKay and Dr. Heather Macdonald. Thank you both for constantly pushing me to find a deeper level of understanding. Your passion and dedication to your work is truly inspiring. I also need to thank my committee member, Dr. PJ Naylor. Thank you for your guidance and feedback from afar throughout this process. I am indebted to Dr. Penny Brasher for her statistical expertise. Thanks Penny, for the whiteboard tutorials and stats chats. Finally, I would like to thank Dr. Carolyn Gotay for serving as my external examiner. I am also incredibly thankful for the years of hard work from the Healthy Bones III Study measurement teams. Of note, thank you to Dr. Kerry MacKelvie-O’Brien, Dr. Heather Macdonald, and Sarah Moore for recruiting the hundreds and hundreds of participants who partook in this study over the past 13 years. Thanks also to Deetria Egeli and Douglas Race for their hard work coordinating data collection. I would be remiss not to acknowledge the dedication and enthusiasm of the Healthy Bones III Study participants, without whom I would have no thesis. I would also like to acknowledge the Canadian Institutes of Health Research for their generous financial support of this study. I am so glad to have spent the past two years getting to know so many great people who formed my support crew on the 6th floor. Kira Frew, thanks for being such a great friend. Thank you for making MEDI 501 and 502 slightly more bearable, for always lending a listening ear and helping me find solutions to my problems. Dr. Sophie Kim, thank you for always having time to chat about anything from bone microstructure to vampires and for always telling me ‘you are not alone’. Dr. Lindsay Nettlefold, thank you for being such a great role model and mentor and for constantly telling me that I can do it. Lynsey Hamilton, thank you for keeping my brain fueled with Percy Pigs and for all the orange stool chats. Sarah Moore, thanks for roping me into grad school and for all the support along the way. Dr. Danmei Liu, thank you for all of your guidance with the XtremeCT, for helping me put things into perspective and for all the laughs along the way. Last, but certainly not least, I owe my family an enormous thank you for their unwavering support and encouragement over the past two years. Mom, Dad, Michael, Andrew and Nissa, thank you for always being just a phone call away.  xv  Chapter 1 – Introduction  Chapter 1 – Introduction The prevalence of childhood and adolescent overweight and obesity has risen dramatically in recent years and is cause for concern [1]. Obesity is linked with many chronic conditions, including orthopaedic complications [2]. The relationship between excess fat mass and bone health in children and adolescents is poorly understood, yet given the rise in the prevalence of obesity the association between these tissues warrants further study. Overweight children sustain more fractures [2] and are at greater risk for forearm fractures than their healthy weight peers [3]. The incidence of forarm fractures has increased concurrently with the escalating levels of child and adolescent overweight and obesity [4]. Since childhood and adolescence are crucial periods for skeletal development [5], it is essential to better understand conditions (i.e. excess adiposity) that may negatively impact development of a healthy skeleton. Many factors influence bone health in children and adolescents and among these the role of muscle is paramount. The mechanostat theory [6] and functional model of bone development [7] propose that bone responds to external mechanical loads by adding bone mineral in areas that experience high strain. Further, they suggest that physiological loads from muscle provide the greatest loads on bones, and it is these loads that drive bone to maintain their functional structural integrity and strength. Given this central relationship, weight bearing physical activity is key to development of a healthy skeleton [8]. Importantly, a myriad of genetic, endocrine and lifestyle factors including diet also interact to develop and maintain bone health during childhood and adolescence. Current literature that investigated the fat-bone relationship in children and adolescents is thus far equivocal. Studies indicated positive [9-11], negative [10,12,13], or non-significant [11,14,15] relations between measures of adiposity and bone variables such as bone mineral content (BMC) measured with dual energy X-ray absorptiometry (DXA) or estimates of bone strength measured with peripheral quantitative computed tomography (pQCT). However, across these studies investigators did not consistently adjust for surrogates of muscle force. This is essential given the strong influence of muscle force on bone as per the functional model of bone development [7]. Although studies that used DXA and pQCT to assess bone mass and structure advanced our understanding of the fat-bone relationship, these technologies are unable to measure detailed aspects of trabecular and cortical bone microstructure, which are key determinants of bone 1  Chapter 1 – Introduction strength [16,17]. With the recent advent of high-resolution peripheral quantitative computed tomography (HR-pQCT) it is now possible to directly measure parameters such as trabecular number and thickness and cortical porosity that previously could only be assessed through invasive bone biopsies. Further, the resolution of HR-pQCT permits application of finite element analysis to estimate bone strength at the distal radius and distal tibia. To date, no study has investigated the influence of adiposity on bone microstructure or strength as assessed by HRpQCT, in children or adolescents [18]. Thus, the primary purpose of my thesis is to evaluate the relationship between adiposity and bone strength in children and adolescents in the context of the functional model of bone development using a novel imaging tool, HR-pQCT. The secondary aim is to determine the relationship between adiposity and bone geometry, density and the microstructural variables that underpin bone strength. I provide relevant background literature in Chapter 2 including bone biology, bone measurement, determinants of bone strength, and the fat-bone relationship. In Chapter 3, I provide the rationale, objectives and hypotheses for this thesis. Chapter 4 describes the recruitment of study participants, methods employed and statistical analyses. Chapter 5 summarizes the results of this thesis and in Chapter 6 I discuss the results and propose future directions for pediatric research focusing on the fat-bone relationship. In Chapter 7 I summarize my findings and provide conclusions.  2  Chapter 2 – Literature Review  Chapter 2 – Literature Review 2.1 Bone Biology Bone is a multi-purpose dynamic tissue. The primary function of bone is to provide structural support to the body while simultaneously withstanding the loads imposed on it by external and internal forces [19-22]. It also serves an important metabolic role as a calcium reservoir to maintain homeostasis of plasma calcium and phosphate [19,21]. In addition to these key functions, bone assists in the protection of internal organs, facilitates movement and accounts for the majority of posture and stature [19,23]. In order to withstand the imposed demands, bones must be strong and resist injury while being sufficiently lightweight to be moved easily and with minimal energy expenditure [24]. Because of their hollow structure, bones are strong, durable, and able to withstand axial forces. Simultaneously, bones’ hollow structure contributes to minimal bone weight and increased efficiency of movement [19].  2.1.1 Bone Types There are two types of bone: woven and lamellar. Woven bone is immature bone found in fetal development, at sites of fracture or healing and in response to mechanical loads [19,20]. Woven bone is characterized by irregular patterns of collagen fibre orientation with osteocytes somewhat randomly dispersed in the matrix [19,20,23,25]. It is laid down quickly and forms the initial scaffolding of bone, which is later resorbed and replaced by lamellar bone [23,25]. In contrast to woven bone, lamellar bone is mature and is the most abundant type of bone found in mammals [21]. It is laid down much slower than woven bone and has a precise and defined structure [25]. Lamellar bone is organized in lamellae: sheet-like layers of bone matrix containing collagen fibers aligned in a near parallel fashion [23]. Lamellae are arranged in three ways. First, in cortical bone they are arranged into Haversian systems, or osteons, whereby a central vascular canal is surrounded by concentric layers of lamellae. Second, they occupy the interstitial spaces between the Haversian systems, and third they are stacked on top of each other [23]. Lamellae run parallel to trabeculae in trabecular bone and to Haversian systems in cortical bone [19]. Lamellar bone fulfills a variety of mechanical functions, and is able to adapt its structure depending on the imposed loads [21]. Lamellar bone is the primary building block of both cortical and trabecular bone. 3  Chapter 2 – Literature Review Together, the structure of cortical and trabecular bone is arranged to withstand the loads applied to the skeleton [19]. Despite similar material properties, the mechanical properties of cortical and trabecular bone are quite different [17]. Cortical bone is a very dense tissue with porosity between 5 and 30% [26]. Its high density relative to trabecular bone is due to tight packing of osteons along the long bone axis [27]. Cortical bone forms the hard outer shell of bones [20,26]. It is thickest at the diaphyses and thins out at metaphyses and epiphyses of long bones [26]. The outer surface of cortical bone is the periosteum while the inner, metabolically active surface is the endosteum [20]. At the endosteal surface, cells are involved in the active process of bone formation and resorption. Cortical bone is an anisotropic material, thus it is stronger in the longitudinal direction along which osteons are aligned, compared with the transverse direction [26]. In contrast to cortical bone, trabecular bone is highly porous and is characterized by a lattice-like support network of lamellar bone [17,20,23]. The porosity of trabecular bone ranges from 30 to over 90%, significantly greater than cortical bone [20]. The spaces between individual trabeculae are occupied by blood vessels, bone marrow and connective tissue [20]. Trabeculae are comprised of lamellar bone; however, the organization is less concentric than in cortical bone [23]. Lamellae run approximately parallel to the direction of trabeculae. Trabeculae are approximately 0.1mm in diameter and 1mm in length prior to joining other trabeculae [16]. The arrangement of trabeculae allows for the transfer of load from place to place through bending moments [16]. Consequently, the mechanical properties of trabecular bone are a result of this lattice-like structure [16,28]. Trabeculae are not randomly organized, but in many cases align themselves along lines of principle stress and often appear to cross each other at right angles when fanning out from cortical bone [16,28]. Trabecular bone is found in the metaphyses and epiphyses of long bones, in short and flattened bones, under bony protuberances for tendon attachment and in the medullary cavity of some long bones [16]. In this thesis I focus on trabecular bone found in the metaphyses and epiphyses of long bones.  2.1.2 Bone Growth Bone growth is the increase in bone length and width that occurs from birth until early adulthood. Growth of bones is necessary not only for the skeleton to withstand loads placed upon it during childhood, but also to prepare the adult skeleton for the lifetime of loads that it will  4  Chapter 2 – Literature Review accrue [29]. Longitudinal growth occurs at the growth plate through ossification of cartilaginous tissue [30,31]. The growth plate is a layer of cartilage located between the epiphysis and diaphysis of long bones (Figure 1) [32]. The growth plate consists of three zones: the resting or reserve zone, proliferative zone, and hypertrophic zone [24,31]. Cartilage is stored in the resting zone next to the bony epiphysis and is multiplied in the proliferative zone. It is removed by chondroclasts from the hypertrophic zone and is replaced with mineralized trabecular bone laid down by osteoblasts [30]. Growth plates are active throughout childhood and adolescence. Importantly, the contribution of growth plate activity to bone length varies between bones, at proximal and distal sites of long bones and at different developmental time points [33,34]. For example, in 7 year old girls and boys, 50% of the increase in tibial length is from activity at the proximal growth plate whereas at 14 years in girls and 16 years in boys the contribution of the proximal tibial growth plate increases to 80% [33]. In contrast, at the radius approximately 80% of growth is at the distal growth plate at 4 and 5 years in girls and boys, respectively, which increases to 90% at 11 years in girls and 12 years in boys [34].  Figure 1. Schematic of a long bone during embryonic development, including the growth plate. The right image shows the proliferation of chondrocytes in the growth plate. Figure from Wallis GA. Curr Biol. 1996;6(12):1577-1580, [35] with permission. 5  Chapter 2 – Literature Review Bone growth and maintenance occur through two processes: modeling and remodelling, which are mediated by bone cells. Osteoblasts produce and lay down bone’s extracellular matrix and regulate its mineralization, whereas osteoclasts resorb or break down bone tissue [19]. The most abundant bone cells are osteocytes, mature osteoblasts within the mineralized matrix [19]. Bone modeling provides the mechanism for increased bone width, whereby bone is deposited onto surfaces by osteoblasts without being preceded by bone resorption [19,29]. Bone modeling contributes primarily to increases in bone size rather than to change in bone shape [29]. Conversely, bone remodelling occurs when bone resorption by osteoclasts precedes bone deposition by osteoblasts. Osteoclastic resorption leaves holes that are then filled in with new bone tissue by osteoblasts [19]. The functions of bone remodelling include repairing microdamage, changing the grain of osteocyte organization in response to altered external strain, removing hypermineralized, brittle bone and maintaining mechanical competence [25]. Remodeling occurs throughout the lifespan and is necessary for bones to adapt in shape, size, and strength in response to external demands [36]. Expansion of bone width occurs through periosteal apposition [24,36,37]. Most often, osteoblasts deposit bone on the periosteal surface while bone is simultaneously resorbed at the endocortical surface [36]. The increase in bone width continues into adulthood, but occurs at a slower rate than during childhood and adolescence [36]. At the periphery of the growth plate in metaphyses during longitudinal growth, trabeculae from endochondral bone formation are enlarged and consolidate to generate new cortical bone [30]. As I discuss in Section 2.1.3, increases in bone width contribute to exponential gains in bone bending strength. Bone growth is modulated by a variety of factors, two of the major factors being genetic and endocrine effects [19], which I address in Sections 2.3.1 and 2.3.2, respectively. Furthermore, I discuss the important role of mechanical loads on bone accrual in Section 2.3.4.1. In addition to bone growth on the periosteal surface, longitudinal accrual of bone mineral contributes to bone growth and is also essential for gains in stature. Bone mass accrual accelerates during puberty and approximately 26% of adult bone mass is laid down during the 2 years surrounding peak height velocity in girls and boys [5]. The rate of bone accrual slows following puberty and ceases in early adulthood [31]. The timing and velocity of linear growth in height for boys and girls differ. Girls reach peak height velocity (PHV) at approximately 11.8 ± 0.9 years and gain approximately 9 cm in height compared with boys who reach PHV at 13.4 ± 1.0 years and gain approximately 10 cm [38,39]. Bone mass accrual follows a similar pattern but 6  Chapter 2 – Literature Review the timing lags behind growth in height by approximately 7 months, on average in boys and 8 months, on average in girls [38]. Boys accrue 407 ± 93 g and girls accrue 325 ± 67 g of total body bone mass at the time of peak bone mass accrual, on average [38].  2.1.3 Mechanical Integrity of Bone Bones are mechanically competent as a result of their material and structural properties [40]. The structural properties of bone are organized in a hierarchical manner from macrostructure (cortical and trabecular bone as a whole) to sub-nano structure, below a few hundred nanometers (including the molecular composition of collagen and mineral) [41] (Figure 2). Macrostructure includes cortical and trabecular bone, whereas microstructure ranges from 10 to 500 μm and includes individual trabeculae, Haversian systems and osteons [41]. On the macro structural scale, bone structural properties include size, shape and cross-sectional area, while microstructural properties include cortical thickness, cortical porosity and trabecular architecture [40]. Separate from bone structure, bone mineral density (BMD, g/cm3) is defined as the mass of bone tissue over the total volume, including pores. Thus, cortical and trabecular densities are highly variable due to the greater porosity of trabecular bone [41]. In addition to structural properties and bone density, the intrinsic properties of cortical and trabecular bone are considered material properties, and include stress, strain, stiffness, strength, mass, toughness, and fatigue [40-43]. Properties across the macrostructure to subnanostructure spectrum contribute to these material properties [41]. Stress is defined as force applied per unit area and can be compressive, tensile, or shear. These types of stresses occur simultaneously, regardless of the simplicity of loading on the bone [42]. Strain is a measure of the deformation of a material [40,42,43]. It is calculated as the change in dimension divided by the original dimension. Stiffness is related to both stress and strain as it is the amount of force needed to deform an object [40]. The slope of the linear segment of the stress-strain curve (Figure 3) is stiffness, and is referred to as Young’s modulus or elastic modulus, and represents bone’s resistance to loading [43].  7  Chapter 2 – Literature Review  Figure 2. Hierarchical structural organization of bone. Figure from Rho J, et al. Med Eng Phys. 1998,20:92-102, [41] with permission.  Figure 3. Stress-strain curve. E = elastic modulus, which is the slope of the stress-strain curve before yield. Figure from Currey JD. Calcif Tissue Int. 2001,68:205-210, [43] with permission.  8  Chapter 2 – Literature Review It is highly desirable for bones to be strong so as to withstand high and unexpected loads and prevent fracture, so I describe the concept of bone strength, below. Strength is defined as bone’s ability to resist fracture [40]. Ultimate strength is the maximum stress a bone can sustain without fracturing. Toughness, or the amount of energy bones can absorb without breaking, is related to strength [42,43]. Type I collagen found in bones allows for elastic deformation during energy absorption [27]. Currey [44] examined cadaveric specimens older than 2 years of age, and found that Young’s modulus increased with age; however, the ability of bones to absorb energy decreased. Therefore, younger bones are able to absorb more energy compared with older bones, but they are less stiff [44]. In general, toughness is inversely related to Young’s modulus and bending strength [43]. With repeated exposure to stresses and strains, bones experience fatigue. Fatigue is manifested as a gradual decline in strength and Young’s modulus over time as a result of repetitive loading [42]. Consequently, a load that would not normally cause a bone to fail, or fracture, applied many times may result in fracture since the repetitive loading leads to bone damage [43]. At the whole-bone level, widened metaphyses help to dissipate loads from contact forces and support the skeleton while the hollow diaphyseal shaft is structured to best respond to torsional and bending loading patterns [40]. In elongated structures such as long bone diaphyses, bending strength is related to bone diameter to the third power and inversely related to bone length to the third power [32]. Longitudinal bone growth increases the lever arm and thus the bending moment and deformation of bone. In order to maintain mechanical integrity, longitudinal bone growth must be coupled with increased bone width [7]. Cross-sectional moment of inertia is a major structural property that influences bone strength. A hollow tube, such as a long bone diaphysis, provides the greatest strength with the least mass in response to torsional or bending loading [40,43]. Maximal cross-sectional moment of inertia in the skeleton is achieved when as little bone mineral as possible is placed as far from the neutral axis as possible (Figure 4). It is calculated as the sum of the products of mineralized pixel area and the square of the distance from the pixel to the bending or torsion axis [45]. This means bone can be strong with minimal weight which also supports efficient bone movement [19]. For example, look at two bones (Figure 5) that have identical mass and cortical (compartmental) BMD. The bone with the greater outer circumference and larger hollow center will be substantially (approximately 3 times!) stronger and stiffer than the bone with a smaller outer diameter and smaller hollow center [46]. Therefore, adding even small amounts of bone 9  Chapter 2 – Literature Review mineral through periosteal rather than endosteal apposition is advantageous, as it increases bone diameter and consequently increases cross-sectional moment of inertia, stiffness, and strength.  DXA: BMC calculated from planar x-ray attenuation data  y A dz dx  pQCT:  x  CSMIx = ∑ (Ai x dx2) CSMIp = ∑ (Ai x dz2)  Neutral axis (z)  Figure 4. Schematic of a long bone and its properties measured by DXA and pQCT. The bending (CSMIx, CSMIy) and torsional (CSMIz) cross-sectional moments of inertia are calculated as the sum of the products of the area of each pixel (Ai) and the squared distance (dx, dy, dz) to the corresponding bending (x,y) or torsion (z) axis. Adapted from Macdonald HM. Thesis, University of British Columbia. 2006, [47] with permission.  Figure 5. The relation between bone geometry and bone strength. Figure adapted from Leonard MB, et al. Bone. 2004;34(6):1044-1052, [46] with permission.  10  Chapter 2 – Literature Review At metaphyseal sites, resistance to bending is not an appropriate index of strength since the metaphysis is predominantly axially loaded in compression [48]. Bone strength index (BSI) incorporates bone density material and geometric properties to estimate bone strength and is calculated as the product of total area (Tt.Ar, mm2) and the square of total density (Tt.BMD, mg HA/cm3) [49]. BSI estimates predict up to 85% of failure load. Therefore, similar to crosssectional moment of inertia, the addition of bone mineral to the periosteal surface at bone metaphyses substantially increases compressive bone strength. Further, greater bone density at the metaphysis is related to enhanced compressive bone strength [48]. Failure load estimation by finite element (FE) analysis is another estimate of compressive bone strength at the distal radius and distal tibia [18]. The average stiffness (E) of the specimen is determined by the reaction force of the finite element models at 1% strain and the average crosssectional area of the specimen [18] (Figure 6). FE determined stiffness correlated highly with bone failure (R2 = 0.972) [18]. Again, greater bone cross-sectional area appears to benefit estimates of compressive bone strength.  Figure 6. Stress-strain curve of destructive loading of cadaveric distal radii to determine linear and elastic failure regions. P = platen force. Figure from MacNeil JA, et al. Bone. 2008;42:12031213, [18] with permission. Material and structural properties of bone interact with each other, and if together they are insufficient to support imposed loads, the bone may fail or fracture [43]. The load-deformation curve describes the relationship between the applied load and the resulting bone deformation 11  Chapter 2 – Literature Review [42]. Permanent bone deformation is called plastic deformation and occurs past the yield point of the stress-strain curve (Figure 3). The elastic deformation region is prior to the yield point, where bone deformation increases with load and returns to its original shape once the load is removed [42]. Applied forces in the elastic region cause only temporary deformation [40]. Therefore, if increasing loads, past the yield point, are placed on bone, it will eventually undergo permanent deformation. Within the whole bone structure, trabecular bone has unique biomechanical properties and contributes to whole bone strength. Trabecular bone is a porous, anisotropic tissue that is highly heterogeneous [50,51]. As a result of this heterogeneity, trabecular bone strength and elastic modulus vary depending on the location within the bone and with the type of stress: tension, compression, shear or a combination [50]. Each individual trabecula has its own stiffness and therefore, when trabeculae are together in a structure, trabecular bone as a material has its own unique structural stiffness [42]. Trabecular bone plays a major role in the energy absorbing capability of bone as a result of its structure [50,51]. Despite being a heterogeneous material, trabecular bone microstructure is generally organized with a ‘grain’ along the direction where stress and strain are greatest [50]. In accordance with Wolff’s law, trabecular bone seems to be placed where it is most needed creating anisotropy, or directional dependence, in response to functional loading [28,50]. Furthermore, the strength of trabecular bone varies with anatomical location due to the changing function of the bone with location. For example, trabecular bone strength is greater at the proximal tibia and distal femur than in vertebral bodies due to differences in trabecular organization [51]. Cortical bone also has unique biomechanical properties. The elastic modulus of cortical bone is dependent on the tissue volume since mineralization, or density, on its own does not directly influence stiffness [17]. Furthermore, cortical bone is an anisotropic tissue and is stronger longitudinally or axially than it is transversely [26]. Cortical bone has greater stiffness and strength compared with trabecular bone as it is less porous [17]. However, the porosity of cortical bone contributes to its mechanical competence, and thus to total bone strength since cortical porosity is inversely related to yield stress [52]. The contribution of cortical bone to total bone strength differs with measurement site. Since bone diaphyses are predominantly cortical bone, it holds that the contribution of cortical bone to total bone strength is much greater at the diaphyses than the epiphyses where the combination of trabecular and cortical bone contribute to total bone strength [26]. 12  Chapter 2 – Literature Review 2.2 Measuring Bone in Children and Adolescents When measuring bone in children and adolescents, researchers and clinicians are interested in determining bone’s ability to resist fracture. While it is impossible to measure bone strength in vivo, a number of imaging tools are currently used to estimate bone strength in the growing skeleton. In this thesis I refer to bone geometry, bone microstructure, bone quality and bone health. I define bone geometry as bone areas (cortical, trabecular and total) and circumferences (periosteal and endosteal). I define bone microstructure as per Manske et al. [53] as bone properties measured on the micrometer scale (i.e. trabecular properties, cortical thickness and porosity). I define bone quality as a combination of features including bone strength, geometry, density and microstructure. In this section I discuss three methods of bone measurement: dual energy X-ray absorptiometry (DXA) (Figure 7A), peripheral quantitative computed tomography (pQCT) (Figure 7B) and high-resolution pQCT (Figure 7 C,D).  A  B  C  D  Figure 7. Sample DXA spine, whole body and hip scan (A); sample pQCT distal tibia scan (B); sample HR-pQCT tibia scan (integral, C) and showing the cortical and trabecular structures (D). 13  Chapter 2 – Literature Review 2.2.1 Dual Energy X-Ray Absorptiometry Dual energy X-ray absorptiometry (DXA) is the most prominent tool used to evaluate children and adolescents’ bone in clinical and research settings [54]. Advantages of DXA include low radiation exposure (1.0 - 9.2 µSv), short scan times, wide availability and ease of use [55]. DXA scans most commonly performed on children and adolescents include total body (and regions), lumbar spine and proximal femur. The X-ray beam travels from the source (beneath the person) to the detector (in the overhead arm above the person). The amount of X-ray beam energy attenuated as it travels through tissue (including bone mineral) along a straight path between the source and the detector is added to generate a pixel by pixel map with each pixel being assigned a value [56]. DXA technology converts X-ray attenuation of two different energies during radiation transmission into relevant bone measures based on known density of different tissues. Bone parameters generated include areal bone mineral density (aBMD, g/cm2), bone mineral content (BMC, g) and bone area (BA, cm2) [55]. Although DXA is the current gold standard in clinical pediatrics and in most pediatric studies, it has a number of limitations. DXA generates two-dimensional (2D) images of a 3D structure and consequently, can only provide an areal measure of BMD (aBMD) rather than true volumetric BMD (g/cm3). In addition, DXA is unable to directly measure bone cross-sectional geometry, and it cannot separate the cortical and trabecular bone compartments [46]. Another major limitation of DXA is that aBMD is highly related to body size, which is of particular concern in growing children. Increases in aBMD during growth are due to increases in bone size rather than true increases in density [54]. However, a number of strategies were proposed to address these limitations. One method corrected values generated from lumbar spine scans. The model assumed vertebral bodies have a fixed geometric shape and it used projected anteroposterior bone area to estimate vertebral volume that was, in turn, used to calculate bone mineral apparent density (BMAD) [57]. A second method adjusted BMC for BA, height, weight and maturity status [58]. A third method developed to correct for the 2D nature of DXA imaging at the proximal femur is hip structure analysis (HSA). HSA derives cross-sectional moment of inertia (CSMI) and section modulus at the proximal femur based on the distribution of pixels from the center of mass [56]. All of these estimates have encountered criticism including HSA, as its derivation is from a 2D (DXA-based, planar) measure, and all should be interpreted with caution [55,56]. Despite these limitations animal and human cadaver research supports a strong correlation between aBMD and failure load or fracture risk. For example, goat humeral and 14  Chapter 2 – Literature Review femoral aBMD [59] and human tibial aBMD [60] were moderately correlated (r = 0.70 – 0.76 and 0.74, respectively) with failure load.  2.2.2 Peripheral Quantitative Computed Tomography Unlike planar DXA technology, peripheral quantitative computed tomography (pQCT) can measure bone cross-sectional geometry and volumetric BMD, and estimate bone strength. As a peripheral measurement system, sites scanned using pQCT include distal and diaphyseal radius and tibia [53]. Advantages of pQCT include short scanning time (< 3 minutes), low radiation exposure (< 1 µSv/scan) and ability to independently assess cortical and trabecular bone [53,55]. In addition, pQCT can measure muscle cross-sectional area (MCSA, mm2), which is often used as a surrogate measure for muscle force (see Section 2.3.4.2) [53]. As for other CT units, the X-ray source and detector rotate around the limb of interest in a coordinated manner and acquires a ~2.5 mm slice to construct 3-dimensional voxel-based images of the region of interest [55]. The voxel value is the average X-ray attenuation of bone and soft tissues within the voxel. The resulting value is converted into mineral density and represents the average amount of mineralized tissue within each voxel [56]. Peripheral QCT outcomes include area, density and mineral content for total, cortical and trabecular regions as well as endosteal and periosteal circumferences and cortical thickness. Data obtained from pQCT images are used to estimate bone strength, including bone strength index (BSI, Equation 1) and polar strength-strain index (SSIp, Equation 2), measures of compressive strength and bending strength, respectively [48,61]. BSI = Tt.Ar  Tt.BMD  (Equation 1),  where Tt.Ar = total area; Tt.BMD = total density and d  SSIp =  Av  Dv  Ct.BMD  (Equation 2),  d ma  where dx = distance from a cortical voxel to the x-axis; Av = area of the voxel; Dv = density of the vo el; and Ct.BMD = estimated physiological “ma imal” cortical bone density (1 00 mg/cm3). Furthermore, BMD measured by pQCT is not influenced by bone size, an improvement upon the aBMD measure generated by DXA systems [54]. Peripheral QCT has a number of limitations. First, voxel size assessed using pQCT ranges from 300-500 μm, greater than adult average trabecular thickness (Tb.Th) of 100-300 μm. 15  Chapter 2 – Literature Review Normative data in children are not available; however, a study from our lab reported Tb.Th of 6777 μm in prepubertal boys and girls and 74-91 μm in postpubertal boys and girls at the distal radius and distal tibia [8]. Consequently, pQCT resolution is inadequate to directly assess trabecular bone microstructure [55]. Second, the voxel size may also contribute to partial volume effects (PVE) when small cortices are assessed, due to some voxels being only partially filled with bone tissue. PVE may underestimate cortical BMD [58]. Third, image analysis protocols and outcome variables reported are not standardized across studies; consequently it is not possible to compare results [53,56]. Fourth, due to its small gantry size, pQCT is unable to assess central and clinically relevant measurement sites such as the lumbar spine and proximal femur [53]. Despite these limitations, pQCT is a useful tool for studies of children and adolescents as its outcomes are not dependent on bone size and it is able to measure bone geometry and BMD, and estimate bone strength.  2.2.3 High-Resolution pQCT (HR-pQCT) High-resolution peripheral quantitative computed tomography (HR-pQCT; also called XtremeCT) is a recent technological advancement in bone imaging. The main advantage of HRpQCT is its high imaging resolution. It has an 82 μm isotropic voxel size that allows direct assessment of trabecular bone microstructure. Adult trabecular thickness is 100-300 μm, on average [53] and is even smaller in children (~60-100 μm reported in our lab) [8]. Thus, researchers are able to determine the unique contribution of trabecular bone microstructure to bone strength, which is not possible with DXA or pQCT. Similar to pQCT, HR-pQCT uses an X-ray and detector, which simultaneously rotate around the limb. The X-ray tube has a 0.08 mm point-focus and projects onto a two-dimensional detector array [62]. The system acquires 110 parallel CT slices, stacked to form a 3D image. HRpQCT images are acquired from the distal radius and distal tibia and outcomes from the standard morphological analysis include total area (Tt.Ar, mm2), total BMD (Tt.BMD, mg HA/cm3), cortical density (Ct.BMD, mg HA/cm3), cortical thickness (Ct.Th, mm), trabecular density (Tb.BMD, mg HA/cm3), trabecular bone volume to total volume fraction (BV/TV), trabecular number (Tb.N, 1/mm), trabecular thickness (Tb.Th, mm) and trabecular separation (Tb.Sp, mm) [62]. Of these measures, Tb.BMD and Tb.N are measured directly whereas BV/TV, Tb.Th, and Tb.Sp are derived from these direct measures [53]. Autosegmentation algorithms can also be applied to HR-pQCT scans to measure Tt.Ar, cortical area (Ct.Ar, mm2), cortical porosity (Ct.Po, 16  Chapter 2 – Literature Review %), Ct.BMD, and Ct.Th [63]. Given the high resolution, finite element analysis (FEA) can be used to estimate bone strength from HR-pQCT images (Section 2.2.3.1). I provide a detailed description of how these variables are measured and derived in Sections 0 and 0. Density and microstructure variables are highly reproducible (CV = 0.7-1.5% and 0.9-4.4%, respectively) in adults [62]; however, reproducibility in children and adolescents is currently unknown. HRpQCT was validated using µCT as the gold standard to assess human cadaveric specimens. Standard analysis of BV/TV, Tb.N, Tb.Th, Tb.Sp and Tb.N by HR-pQCT correlated well (R2 = 0.59 – 0.96) with ex vivo µCT measures of the same variables [64]. HR-pQCT has a number of limitations. In order to achieve a resolution of 82 μm the field of view for HR-pQCT systems is small, thus, only peripheral sites are measured [53,55]. Another consequence of high resolution imaging (and thus slower scan times) is the increased likelihood of motion artifacts, which often require scan acquisition to be repeated [65,66].  2.2.3.1  Finite Element Analysis Finite element analysis (FEA) is a numerical approach that has been applied to estimate  bone strength from HR-pQCT images. FEA takes an object, such as a bone image acquired by HR-pQCT, and represents the object as a series of elements defined by reference points [55]. The HR-pQCT images are converted directly into 3D mathematical models that represent the heterogeneity of mineralized tissue distribution [55]. Computer programs (FE models) have been developed to simulate applied loads including forces experienced during fracture. FE models are able to calculate the stiffness and strength in response to applied forces [18,55]. A uniaxial, compression force is most often applied in a simulation; however, forces experienced by bones in vivo are seldom, if ever, experienced in pure compression. Outcomes derived from FE analysis of HR-pQCT scans include failure load (N; force that causes bone to fail) and ultimate stress (N/cm2; force applied per unit area that causes bone to fail). An HR-pQCT and FEA study of cadaveric radii estimated bone strength using FEA models and compared strength estimates to strength evaluated using destructive loading assessed ex vivo [18]. Correlations between strength measured using FE models versus direct strength measures from destructive loading were high (R2 > 0.93). Thus, authors concluded that FEA models were able to estimate bone strength at the distal radius [18]. Currently, FEA models derived from HR-pQCT scans have only been validated in adult bone. However, this technique has been applied to estimate bone strength in children and adolescents in two recent studies that utilized HR-pQCT [67,68]. 17  Chapter 2 – Literature Review One limitation of FEA is that it requires significant computational effort (time) and resources [18]. There are two typical approaches to FE analysis, linear and non-linear. Linear FE models assume a linear relationship between failure load and apparent elastic properties, but require less computational power than non-linear methods [18]. Estimation of elastic properties correlates highly with failure using this method (r = 0.97). Non-linear analysis may provide a more accurate estimate of bone strength since it can directly predict bone strength properties based on a modulus created from CT attenuation values. The correlation of apparent elastic properties with failure load is higher using non-linear analysis (r = 0.98); however, its computational demands are significantly greater due to the millions of degrees of freedom in each model, thus, linear methods are most commonly used [18].  2.3  Determinants of Bone Strength in Childhood and Adolescence There are a myriad of factors that determine bone strength in childhood and adolescence.  In this section I outline selected non-modifiable and modifiable predictors of bone strength. Nonmodifiable factors include the genetic and endocrine environments. I did not assess the impact of genetics or hormonal factors on bone strength for this thesis but present a brief overview given their contribution to bone structure and function. Specifically, I briefly outline key genetic and endocrine modulators of bone strength. The modifiable determinants of bone strength I focus upon in this thesis are diet and physical activity given the key relation of both to fat, muscle and bone tissue. Finally, I discuss the mechanostat, functional model of bone development and the muscle-bone relationship. As the fat-bone relationship is the focus of my thesis, I focus specifically on fat as a determinant of bone strength in Section 2.4.  2.3.1 Genetics It is well established that genetics account for the majority of variability in bone mass and aBMD. Heritable factors account for between 18 and 80 percent of the variability in BMC and aBMD [69-71]. High heritability is evident in studies that examined family trends in osteoporosis and fracture and found that relatives of those with osteoporosis have lower aBMD or BMC by DXA compared with controls. Further results of twin studies suggest that heritability may account for 50 to 80% of the variability in aBMD [70-73]. Heritability estimates for bone crosssectional geometry as measured by pQCT are slightly lower (27-75%), which may be a result of 18  Chapter 2 – Literature Review greater environmental influences on site specific bone geometry [74]. Further, heritability of trabecular BMD may be greater than cortical BMD (~70% as compared to ~35%). Why the contribution of heredity differs by bone compartment is unknown [75]. The specific polymorphisms affecting bone mass and geometry are still under investigation; however, a myriad of genes are implicated in the regulation of bone health. Genes that may play a role in bone modeling and remodeling include genes for the vitamin D receptor, estrogen receptor, androgen receptor, apolipoprotein E4, interleukins, parathyroid hormone (PTH) receptor, collagen, and calcitonin receptor [70,76-78]. While genetics clearly play a large role in determining an individual’s bone mass, density and geometry, modifiable lifestyle factors account for as much as 40% of the remaining variability and are important to consider [79] (Section 0).  2.3.2 Ethnicity Ethnicity also plays an important role in bone health. In this thesis I account for ethnic differences in bone quality between Asian and Caucasian children, adolescents and young adults. I define ethnic differences rather as the contribution of lifestyle as biological factors to differences in overall bone quality between Asians and Caucasians [80]. In particular, McKay et al. [80] reported that Asian children were 15% less active and consumed 35% less dietary calcium than Caucasian children in the same geographic area. Approximately, 57% of Asian children attended academic lessons outside of school hours compared with 28% of Caucasian children [80]. Clearly there are lifestyle differences between Asian and Caucasian children that influence bone health.I address the contributions of physical activity and diet to bone health in Sections 2.3.4.1 and 2.3.4.2, respectively. A number of studies compared bone outcomes between Asians and Caucasians [80-84]. In adults they reported lower aBMD at the lumbar spine and total hip in Asian compared Caucasian men and women, which was paradoxiacally accompanied with lower fracture risk in Asians [8183]. Pre- and early-pubertal Asians children also had lower femoral neck, proximal femur and total body aBMD compared with Caucasians [80,84]. This ethnic difference in aBMD is often attributed to the smaller bone size of Asians compared with Caucasians [81,85]. As previously discussed, DXA cannot measure bone geometry and microstructure. Thus, early studies comparing Asians and Caucasians did not assess potential differences in bone geometry or microstructure, which may explain the lower fracture risk in Asians compared with Caucasians. 19  Chapter 2 – Literature Review Studies that used pQCT HR-pQCT to assess ethnic differences in bone quality between Asians and Caucasians are limited, particularly in children and adolescents. However, crosssectional studies of pre- [86,87] and post-menopausal [88] women using HR-pQCT found that Asian women had smaller bone size at both the distal radius and distal tibia compared with Caucasian women. Conversely, Asian women were reported to have thicker and denser cortices [86,87], which may compensate for the smaller bone size [88]. Additionally, our lab previously reported smaller Tt.Ar and Ct.BMD at the tibial midshaft as measured by pQCT in pre- and early-pubertal Asian girls compared with Caucasian girls [61]. Similar results were found in postpubertal adolescents in our lab as Asian boys were found to have lower Tt.Ar and Ct.Po but greater Tt.BMD, Ct.BMD, and Ct.Th by HR-pQCT at the distal radius compared with Caucasian boys [89].  2.3.3 Hormones Growth, maintenance and health of bones are under hormonal influence [70]. Calcitropic hormones, gonadal steroids, growth hormone, and leptin all influence bone modeling and remodelling [71]. Three key hormones that regulate serum calcium are parathyroid hormone (PTH), vitamin D, and calcitonin [20,90]. Since bone is a reservoir of calcium, these hormones modulate levels of serum calcium to maintain calcium homeostasis. PTH is the main regulator of serum calcium and stimulates bone resorption to increase plasma calcium concentrations [19,23,91]. As a secondary function, PTH enhances calcium absorption from both the kidneys and the intestines (Figure 8) [90]. However, it is important to note the paradoxical influence of PTH on bone mineralization. As previously noted, PTH causes net bone loss through bone resorption when secreted or administered continuously; however, when PTH is administered intermittently in pulses, it results in net bone formation [92].  20  Chapter 2 – Literature Review Low concentration of calcium in blood Release of parathyroid hormone  Vitamin D  Efflux of calcium from bone  Decreased loss of calcium in urine  Enhanced absorption of calcium from intestine  Increased concentration of calcium in blood  Figure 8. Influence of parathyroid hormone on serum calcium. Figure from Biomedical Hypertextbooks http://www.vivo.colostate.edu/hbooks/pathphys/endocrine/thyroid/pth.html, retrieved July 15, 2012 with permission from author. The thyroid gland produces the hormone calcitonin. The primary function of calcitonin is to inhibit osteoclast metabolism, and in turn, bone resorption [90]. In addition, vitamin D in its active metabolic form (1,25-dihydroxyvitamin D) influences renal calcium absorption, calcium transport in the intestines, and calcium mobilization from bone to aid in the maintenance of serum calcium homeostasis [20,90]. Low levels of vitamin D promote bone turnover and the loss of bone mineral [93]. A 3-year prospective study of peripubertal girls found that girls with the highest vitamin D intake had the greatest change in lumbar spine aBMD (by DXA) and that girls with low vitamin D intake were at risk of not reaching peak bone mass [94]. Furthermore, a randomized controlled trial of calcium and vitamin D supplementation in peripubertal female identical twins found that 6 months of supplementation increased SSIp and trabecular density and area measured by pQCT at distal radial and tibial sites [95]. The gonadal steroids estrogen and testosterone also influence bone metabolism. The primary effect of estrogen on bone is to limit bone resorption by inhibiting osteoclast differentiation [20,96]. Further, estrogens may limit periosteal apposition and endocortical expansion in girls [37]. Testosterone also limits bone resorption, although it may contribute to larger bone size and enhance periosteal apposition in men compared with women [20]. During puberty, sex hormones promote bone cell proliferation and maturation. This results in the 21  Chapter 2 – Literature Review pubertal growth spurt, which ends with fusion of the epiphyseal growth plate [70]. Androgens and estrogens play a critical role in establishing the sexual dimorphism of bone characteristics [37]. In girls, estrogen rises rapidly at the onset of puberty while for boys, testosterone rises steadily throughout puberty [24]. The increase of sex steroids during puberty is clearly linked with a rapid increase in bone mass [37]. A prospective study of the influence of sex steroids on bone geometry at the tibial midshaft in girls found that estrogen positively predicted Tt.BMD and Ct.Th and negatively predicted endosteal circumference while testosterone positively predicted Tt.Ar and endosteal circumference and negatively predicted Tt.BMD [97]. Our lab has reported extensively on the sex specificity in the timing and magnitude of changes in bone parameters as assessed by DXA [98], pQCT [99] and HR-pQCT [100]. Although it has been suggested that in boys, bone mineral is predominantly added through periosteal apposition and in girls through apposition on the endocortical surface [37], we did not observe a difference in endosteal apposition in previous studies conducted in our lab [101]. In addition to calcitropic and gonadal hormones, growth hormone (GH) and insulin-like growth factor-1 (IGF-1) play a critical role in longitudinal bone growth and attainment of peak bone mass [71]. GH and IGF-1 influence bone modeling by stimulating chondrocyte and osteoblast proliferation [71]. IGF-1 levels peak during the adolescent growth spurt, roughly one year following peak height velocity, at 13-14 yrs for girls and 15-16yrs for boys [24]. The timing for peak IGF-I levels corresponds with the timing of peak BMC velocity [5]. After closure of the epiphyseal growth plate, GH and IGF-1 levels decrease, but these hormones remain active in the regulation of bone remodelling [71]. A decrease in GH with age may contribute to reduced bone formation with advancing age [20]. A further hormone that influences bone metabolism is leptin. Leptin is synthesized by adipocytes; consequently obesity is associated with greater levels of serum leptin [91]. Leptin’s action on bone is complex and may differ at central and peripheral sites [102]. Outside the central nervous system, or peripherally, the leptin receptor is expressed on both osteoblasts and chondrocytes; thus, leptin is associated with bone preservation by the inhibition of osteoclastogenesis [103]. In contrast, leptin targets the hypothalamus centrally, and is associated with enhanced bone resorption [103]. The balance of these two pathways may reflect the overall effect of leptin on the skeleton. In a study of 1068 young men aged 18-20 yrs, leptin was a negative predictor of pQCT-derived cortical bone area and thickness (but not cortical BMD) at the 25% site of the radius and tibia (adjusted for fat mass, lean mass, age, height, physical 22  Chapter 2 – Literature Review activity, calcium intake and smoking) [11]. It may be that increased levels of leptin associated with greater adipose tissue may be one mechanism that underpins the fat-bone relationship.  2.3.4 Lifestyle Factors There are a large number of modifiable lifestyle factors that also influence bone across the lifespan. Two major modifiable factors that are key to bone health are physical activity and diet [104]. I briefly discuss these in the following two sections.  2.3.4.1  Physical Activity As highlighted by the functional model of bone development [105] and mechanostat  theory [6] (Section 2.3.4), muscle forces drive bone adaptation. Mechanotransduction is the process by which muscle forces and other imposed demands drive skeletal adaptation; this occurs through physical transduction of load (from mechanical force) to a form that can be interpreted by osteocytes [20]. As such, physical activity has the potential to increase muscular forces exerted on bone and thus, bone should directly adapt to physical activities that increase the muscular load applied to the skeleton [106]. A systematic review of controlled trials assessing the impact of physical activity interventions on DXA-related bone outcomes found that weightbearing physical activity interventions had a positive impact on BMC and aBMD in children and adolescents [107]. However, as highlighted in Section 2.2.1, DXA is unable to capture changes in bone geometry or bone strength. Therefore, in this section I focus on those studies that assessed the relationship between physical activity and 3-dimensional measures of bone (structure) and estimates of bone strength. A recent meta-analysis assessed 10 randomized controlled trials that determined the impact of physical activity interventions of at least 6 months duration on bone strength estimated with QCT, pQCT, MRI or HSA across multiple age groups (childhood, adolescence, and young and older adulthood) [108]. Five of the studies were in children and adolescents. The metaanalysis found a small (effect size = 0.17), but significant, positive effect of weight-bearing physical activity on lower extremity bone strength in boys, but not in girls. One positive feature of this review was that it adopted strict inclusion criteria and reviewed only randomized controlled trials. One limitation was that the authors evaluated different papers that reported outcomes from the same cohort as if they were independent datasets. There is as yet a dearth of  23  Chapter 2 – Literature Review high quality studies that evaluated the role of exercise on bone structure and strength as assessed by HR-pQCT. In fact, no intervention study to date used HR-pQCT to investigate the effects of physical activity on cortical or trabecular bone microstructure. However, one recent cross-sectional study used HR-pQCT to evaluate the relation between weight bearing physical activity and bone outcomes, in adolescent girls and boys. They reported a positive relationship between weightbearing physical activity and Tb.N, Tt.BMD and Tb.BMD at the distal tibia in adolescent girls, only [8]. Interestingly, impact loading physical activity was positively related to estimated bone strength (minimal and maximal moments of inertia; Imin and Imax, respectively) and Tt.Ar at the distal tibia in adolescent boys. Results from the meta-analysis [108] and the relatively few studies that assessed growing bone using HR-pQCT [8] highlighted sex differences in bone adaptation to impact loading. This variability between boys and girls is consistent with sex differences in normal bone growth as discussed in Section 2.1.2, including greater periosteal apposition in boys compared with girls [37].  2.3.4.2  Diet A variety of nutrients modulate skeletal health in children and adolescents. Here, I focus  on two of the main dietary nutrients: calcium and vitamin D [109]. It is well established that calcium plays an essential role to develop and maintain skeletal health [110]. Ninety-nine percent of calcium in the body is stored in bones and teeth [104]. The two key roles of calcium are to maintain the skeleton’s structural integrity and to regulate metabolic function [104]. Vitamin D stimulates bone maturation and bone matrix formation among other functions [104]. In addition, with PTH, vitamin D aids in the absorption of calcium, particularly when dietary calcium is low [109]. Please see Section 2.3.2 for a discussion of the role of vitamin D and PTH on bone metabolism. Much of the research addressing the influence of calcium on bone health in children and adolescents focused on calcium supplementation and/or intake of dairy products. Reviews also focused primarily on DXA measures of bone mass. A review of studies examining dietary calcium intake and bone measures in children and adolescents suggested a positive influence of dietary calcium and calcium supplementation on aBMD and BMC as assessed by DXA [111]. They also reported a greater increase in aBMD and BMC in children and adolescents who 24  Chapter 2 – Literature Review reported low calcium intake prior to supplementation [111]. A meta-analysis of 21 randomized controlled trials of calcium intake in children and adolescents indicated that increases in dietary or supplemental calcium enhanced BMC of the whole body and lumbar spine by DXA in children with low baseline calcium intake [110]. Additionally, study participants who received a combination calcium and vitamin D supplement experienced greater BMC gains at the lumbar spine compared with controls. Conversely, compared with controls, children with normal calcium intake at baseline who received calcium treatment experienced little change in BMC [110]. These reviews suggest that calcium supplementation may only be beneficial for those with low calcium intake and that there may be a ceiling effect in terms of how much dietary or supplemental calcium is required to increase BMC. I briefly highlight the few studies that examined the relationship between calcium and pQCT derived bone outcomes. A 6-month randomized controlled trial of calcium and vitamin-D supplementation in 20 pairs of 9 to 13 year old identical twin girls found 5-7% greater gains in Tb.BMD, Tb.Ar, SSIp, and Ct.Ar at the distal radius and distal tibia in the treatment group compared with the placebo group [95]. Further, dietary calcium was weakly but positively related to Ct.Ar in boys and girls (accounting for < 4% of variance) and to SSIp in boys at the tibial midshaft in a sample of 514 children (257 boys) [61]. Overall, studies suggest that dietary calcium and calcium plus vitamin-D supplementation may have a modest, but positive, effect on bone geometry and strength. However, it is unclear if these gains persist when supplementation stops.  2.3.4 The Mechanostat and the Functional Model of Bone Development Harold Frost’s mechanostat hypothesis proposes that there is a minimum effective strain (MES) setpoint; loads above and below the setpoint stimulate or slow bone mineral accretion, respectively [6]. Frost proposed that bones have more bone mineral than necessary in order to resist deformation from typical peak voluntary mechanical loads. Conversely, if typical peak voluntary mechanical loads decrease as a result of disuse, bone strength is reduced through the removal of trabecular and endocortical bone [112]. Building on the mechanostat hypothesis, the functional model of bone development proposes that physiologic loads from muscle forces cause bones to self-regulate in order to maintain functional structural integrity and strength [7,105,113,114]. By adapting to physiologic loads, the skeleton continually adapts its strength to keep bone deformation within safe limits and 25  Chapter 2 – Literature Review thus reduce fracture risk [105]. Muscles apply the largest loads on bones and these loads can be up to 10 times greater than body weight and other external forces [113,115]. Since this model proposes that muscle force is a driving factor to maintain bone strength, it is important to understand the relationship between muscle force and bone during growth and across the lifespan. Thus, Frost’s model implies that mechanically loading the skeleton is key to structural integrity, whereas factors such as hormones and nutrition serve as modulators of the functional muscle-bone relation [7], as depicted in Figure 9.  Bone architecture Bone strength Bone mass  Challenges Increase in:  Bone length  Muscle force  Regulatory Feedback Loop Tissue strain  Setpoint  Osteocytes  Effector signals  Osteoblasts Osteoclasts  Hormones, nutrition; behavioural, environmental factors  Modulators Figure 9. Functional model of bone development based on mechanostat theory. Adapted from Rauch F, et al. Pediatr Res. 2001;50(3):309-314, [7] with permission. 2.3.4.1  The Muscle-Bone Relationship During growth and development, increases in muscle mass precede increases in bone  mass, on average [105]. Many believe this increase in muscle mass stimulates bone mineral accretion as per the functional model of bone development [32,105,115,116]. A longitudinal study of children developed velocity curves that showed total body peak lean mass (assessed by DXA) preceded total body peak bone mass by 0.51 years in girls and 0.36 years in boys (Figure 10) [105]. This supports the functional model of bone development, whereby muscles apply the greatest load on bones and stimulate bone mineral accretion where these loads are experienced. Similarly, Xu et al. conducted a 7-year longitudinal study of girls (11.2 yrs at baseline) and found 26  Chapter 2 – Literature Review that peak muscle area assessed by pQCT preceded peak BMC and Ct.BMD at the tibial midshaft [117]. However, this group did not find that MCSA preceded peak bone area or bone length. Thus, bone growth in length and circumference may be independent of muscle force whereas the accrual of bone mineral may be more closely related to increased muscle force.  Figure 10. Velocity curves that depict total body lean body mass (LBM) and bone mineral content (BMC) accrual during the pubertal growth spurt. Note the difference in magnitude with boys’ accrual being higher than girls’ at peak and the difference in timing with girls achieving peak values almost 2 years in advance of boys. Figure from Rauch F, et al. Bone. 2004;32:364377, [105] with permission. A number of studies in children and adolescents examined the relationship between muscle force, or surrogates such as lean mass by DXA or MCSA by pQCT, and estimates of bone strength. An early cross sectional study from the University of Cologne reported a strong positive correlation between grip strength and pQCT measures of bone geometry (Tt.Ar, Ct.Ar) and bone strength (BSI) at the distal radius in 14 children aged 6 to 13 yrs [116]. Another cross-sectional study examined the contribution of MCSA to measures of bone geometry and strength (by pQCT) in pre- and early-pubertal boys and girls. They found that after controlling for tibial length, MCSA predicted 10-16% of the variance in tibial bone geometry and bone strength [61]. Further, reduced amounts of lean tissue as a result of muscle disuse resulted in bone loss from 27  Chapter 2 – Literature Review regions that experienced reduced mobility or immobility in patients with spinal cord injury or stroke, for example [6,115]. Together, these studies begin to form a picture that supports a strong link between muscle and bone. Thus, the muscle-bone relationship must be considered when examining the influence of external loads on bone strength.  2.3.4.2  Measuring Muscle Mechanostat theory supports the notion that muscle forces, not muscle mass, applied to  the skeleton result in bone strain, which stimulates bone modelling [6,118]. Therefore, it behooves researchers to more appropriately use measures of muscle force wherever possible to examine the functional muscle-bone unit. However, it is not possible to directly measure muscle force using non-invasive techniques. Thus, methods such as dynamometry and mechanography are used to estimate muscle force [118]. It is possible to measure maximal isometric grip strength (kg) with a hand held dynamometer, an easy, reliable and inexpensive measure in children and adolescents. However, these measures did not correspond well with daily activities of children and adolescents [118]. Mechanography is another option used to evaluate lower limb motor performance through force plate jumps [118]. Outcomes from mechanography include peak jump force (N), peak jump power (W), maximal velocity at takeoff (m/s), jump height (m), and jump energy (J) [118,119]. These measures provide highly reliable outcomes that represent muscle function and performance. Bone mineral free lean tissue mass (lean mass (LM), kg) and MCSA from imaging techniques such as DXA and pQCT, respectively, are commonly used to estimate muscle mass and as surrogates of muscle force. As with measures of bone and fat mass from DXA total body scans, LM is derived based on the attenuation of X-rays through an assumed fixed lean tissue density [120]. LM estimations by DXA were validated against chemical analysis of child-sized phantoms composed of ground beef [121]. Importantly, LM by DXA is highly correlated (r = 0.77) with leg muscle power [122]. MCSA by pQCT is assessed as a single 2.3 mm slice at the radius or tibia. Measures are most often taken at 66% of the limb length as this is where limb circumference and thus muscle circumference is greatest, on average [123]. Thresholding techniques are applied to delineate between muscle, bone, and fat. MCSA is often used as a surrogate for muscle force when functional measures are not available as MCSA is highly correlated (r = 0.86) with functional measures of muscle force in adults [124]. 28  Chapter 2 – Literature Review 2.4  The Fat-Bone Relationship There has been a rapid rise of childhood and adolescent overweight and obesity in recent  years. Although the prevalence increased in Canada from 15% in the late 1970s to 26% in 2004 [125], the influence of adiposity on bone health in children and adolescents remains poorly understood. A paradox exists whereby body weight appears to protect older adults against fracture [126] but overweight children sustain more fractures and report more orthopaedic complications than their healthy weight peers [2]. Goulding et al. [3] reported that overweight children were at 33% greater risk of forearm fracture than healthy weight children. Thus, excess fat mass may be disadvantageous for the growing skeleton. However, studies that examined the fat-bone relationship in pediatric populations are thus far equivocal and conflicting, as I discuss in sections 2.4.3 and 2.4.4. Fat mass is defined as the entire mass of adipose tissue in the body based on its chemical composition, whereas adiposity is an estimate of fat mass based on the tool of measurement (i.e. DXA, skinfolds, hydrodensitometry etc.). For the purposes of this thesis I use the terms fat mass and adiposity interchangeably since adiposity by DXA is most commonly referred to as fat mass.  2.4.1 Cellular Mechanisms Interactions between fat and bone during growth are not well understood. However, it is believed that fat and bone share common regulation at the hypothalamus and bone marrow [127]. The hypothalamus regulates fat by modulating appetite, insulin sensitivity, energy homeostasis, and skeletal remodelling via the sympathetic nervous system [103,127]. Leptin is believed to play a role in hypothalamic fat-bone interactions [103]. Leptin is released from adipocytes and then processed in the hypothalamus. In theory, as leptin levels increase, the hypothalamus sends sympathetic signals that regulate skeletal remodeling by depressing osteoblast activity and enhancing osteoclast activity [103,127]. At the bone marrow level, there is a growing body of literature that suggests osteoblasts and adipocytes are derived from common mesenchymal stem cells [102,128]. The combination of physical, chemical and molecular stimuli in bone marrow determines whether mesenchymal stem cells differentiate into osteoblasts or adipocytes--although these processes are not necessarily mutually exclusive [129]. It is proposed that if more mesenchymal cells differentiate into adipocytes and lead to increased marrow adiposity, there is a concomitant reduction of mesenchymal stem cells that differentiate into osteogenic cells [128]. A longitudinal study (39 29  Chapter 2 – Literature Review healthy women, aged 15-20 yrs) assessed the relationship between bone mineral accrual and marrow adiposity. Results showed a strong inverse association between marrow fat and cortical area of the femoral midshaft measured with computed tomography, even after adjusting for body size and lean mass in [130].  2.4.2 Measuring Adiposity in Children and Adolescents It is not possible to measure adiposity directly in children and adolescents [131]. Thus, we must rely on various indirect methods. Measures used to quantify adiposity range from very simple (e.g. body mass index (BMI, weight/height2)) to very complex (e.g. 4-component models) [132]. In this section I focus on two common methods used to measure adiposity: BMI and body fat (%BF) assessed by DXA. There are a variety of other measures used to assess adiposity in children and adolescents. These methods differ in their prediction of %BF and all have advantages and disadvantages. I briefly summarize them in Table 1 [132,133].  30  Chapter 2 – Literature Review Table 1. Summary of advantages and disadvantages of body composition measurement methods. Method Body mass index  Advantages Quick Easy to obtain Reference data available  Disadvantages Does not measure adiposity May inappropriately categorize muscular children  Dual energy X-ray absorptiometry (DXA)  Low radiation Accessible Easy to use Measures total body fat, bone mineral free lean mass and % body fat  Influenced by anterior-posterior thickness Fat and lean mass measures vary with age, sex, maturation, instrument and software  Sskinfold Thickness  Quick Easy to obtain in most age groups  Difficult to measure in obese children Large inter-measurer variability May be uncomfortable for participants  Waist circumference  Quick Provides measure of central fatness  No reference data Does not distinguish between fat and muscle  Bioelectric impedance analysis  Quick  Unreliable, varies with hydration level  Hydrodensitometry  Distinguishes fat mass and fat-free mass  Requires specialized equipment Requires participants to exhale to residual volume Assumes constant densities of fat and fat-free mass  Air displacement plethysmography  Distinguishes fat mass and fat-free mass  Requires specialized equipment Assumes constant densities of fat and fat-free mass  Isotope dilution  Allows estimation of fat free mass  Requires specialized equipment  Multicomponent models  Criterion method Most accurate approach  Requires specialized equipment Expensive  2.4.2.1  Body Mass Index BMI is the most widely used measure of overall adiposity. It is based on the relation  between two simple anthropometric measures - weight and height (weight/height2). Therefore, it can be easily derived in clinical settings and for epidemiological studies. In a validation study, BMI predicted 85% and 89% of the variance in total body fat mass (measured by DXA) in boys (n = 90) and girls (n = 98; 5-19 yrs), respectively [134]. Variance of percent body fat explained by BMI was 63% for boys and 69% for girls [134]. By comparison, BMI explained 41% to 88% of the variance in percent body fat and total body fat mass measured by hydrodensitometry in boys (n=201; 8 to 18 yrs) [135]. In girls, the variance of total body fat mass and percent body fat (measured using hydrodensitometry) explained by BMI was 14% to 81% [135].  31  Chapter 2 – Literature Review Reference data for BMI are available for children and adolescents in the United States [136], the United Kingdom [137] and other countries [138]. Reference curves use cut points to classify children and adolescents into four weight categories; underweight, healthy weight (HW), overweight (OW) or obese (OB). Cut points are age- and sex-specific and vary depending on the growth chart used. For example, the American Centers for Disease Control in 2000, developed BMI-for-age charts to categorize children and adolescents aged 2 to 20 yrs into HW, OW and OB groups based on their BMI-for-age percentile [136]. In this classification system, HW children have a BMI-for-age < the 85th percentile, OW children fall between the 85th and 95th percentiles, and OB children have a BMI-for-age ≥ the 95th percentile compared with reference data from 1963-1994 National Health Examination Surveys (NHES) and National Health and Nutrition Examination Surveys (NHANES) in a total of 16648 boys and 16005 girls aged 2-20 [136]. Similarly, to develop the UK BMI-for-age growth charts (1990), BMI was converted into percentiles or standard deviation scores [137]. In contrast, the International Obesity Task Force (IOTF) classify children as HW if their BMI-for-age is less than the curve passing through a BMI of 25 kg/m2 at age 18, OW if their BMI-for-age is greater than the curve passing through a BMI of 25 kg/m2 but less than the curve passing through a BMI of 30 kg/m2 at age 18, or OB if their BMI-for-age is greater than the curve passing through a BMI of 30 kg/m2 at age 18 [138]. While BMI is a quick and simple measure, it is a measure of relative weight and is therefore unable to differentiate the contribution of lean and fat mass to total body mass [132]. One limitation is that children and adolescents may be categorized inappropriately. For example, children with a high proportion of lean mass may be categorized as overweight or obese and thus considered ‘over fat’ when this is not actually the case [139]. While BMI has utility for assessing large populations it is not possible to garner information about an individual child’s body composition. Thus, specific measurement tools such as DXA are better suited to assess body fatness.  2.4.2.2  Dual-energy X-ray Absorptiometry Although DXA was developed to measure bone mineral, fat and lean tissue can also be  assessed from whole body scans [132]. Due to its widespread availability, relative ease of use and low radiation exposure, DXA is commonly used to measure whole body and regional fat mass in HW, OW and OB children and adolescents [140,141]. Importantly, %BF by DXA was highly correlated with %BF predicted using the criterion 4-compartment (4C) model in children and 32  Chapter 2 – Literature Review adolescents (R2 = 0.85 - 0.90) [142,143]. A study comparing %BF by DXA and 4C compared the two values in 411 children aged 6 to 18 yrs using univariable regression found that DXA %BF underestimated 4C %BF in those with low 4C %BF while DXA %BF overestimated 4C %BF in those with high 4C %BF [142]. Although some validation studies suggest that DXA predictions of %BF in children are inaccurate (Table 2), the consensus is that these inaccuracies should not prevent the use of DXA to measure body fat [141-144]. Further, the 4C model is difficult to use as it requires specialized equipment to measure body density (by hydrometry or air displacement plethysmography), total body water (by deuterium dilution) and bone mineral (by DXA) [141]. This equipment is not widely available, thus many pediatric studies use DXA as the criterion method to assess body composition [140].  Table 2. Summary of studies that compared percent body fat (%BF) measured by DXA with %BF measured by 4-compartment (4C) models. Primary Author Sopher [142]  N 411  Age (yrs) 6-18  Population Multiethnic  Agreement of %BF (by DXA) with %BF (by 4C models) DXA underestimated 4C %BF in those with low 4C %BF DXA overestimated 4C %BF in those with high 4C %BF  Silva [144]  78  15  Athletes  DXA overestimated 4C %BF  Van der Ploeg [131] Wells [141]  152  18-59  Adults  DXA underestimated 4C %BF of leaner individuals  174  5-22  Obese  DXA overestimated 4C FM by 0.9kg  Wong [143] 141 9-17 Female DXA = dual energy X-ray absorptiometry.  DXA overestimated 4C %BF by 3.9%  It is important to note that despite its substantial advantages, DXA has a number of limitations. Greater anterior-posterior thickness may decrease the accuracy of fat mass measures; thus, DXA may overestimate %BF in obese individuals and underestimate %BF in those with smaller anterior-posterior depth DXA calculates fat mass with instrument-specific algorithms that assume constant density of fat tissue. These assumptions may not always be correct as the depth and density of adipose tissue may vary from person to person [132,142]. Further, DXA body composition measures in children vary with bone maturation, sex, age, DXA model (i.e. Hologic vs. Lunar) and software [143]. Individual body fat measures from DXA must be interpreted with these limitations in mind. However, at the population level, DXA provides good estimates of fat mass and percent body fat that are more accurate than some commonly used field measures and are easier to obtain than more sophisticated approaches [143]. 33  Chapter 2 – Literature Review 2.4.3 Studies of the Fat-Bone Relationship using DXA in Children and Adolescents Many early studies that examined the fat-bone relationship in children and adolescents used measures of BMC and aBMD by DXA and compared them against BMI categories or fat mass (FM) by DXA. Studies were difficult to compare as they varied in sample size, participant age and ethnicity, bone measurement site and statistical approach (Table 3). Most studies were cross-sectional [145-153]; one was longitudinal [154]. Five cross-sectional studies compared BMC and aBMD between obese (OB), overweight (OW) and healthy weight (HW) participants after classifying participants into these weight categories based on BMI-for-age [148-151] or %BF [152]. Few studies adjusted for muscle force or surrogates [150-152] and this may have contributed to inconsistent findings regarding the fat-bone relationship in children and adolescents. The functional model of bone development recommends that muscle or its surrogates be controlled in studies of this nature [7]. Five studies used DXA to examine the relationship between FM (as a continuous variable) and bone mass [145-147,153,154]. Prior to adjusting models for LM, three of these studies reported a positive relationship between total body FM and bone measures that included lumbar spine [145], femoral neck [145] and total body aBMD [145,146], total body BMC [145,147] and bone area [147]. However, after adjusting for total body LM, the relationship between FM and bone mass was no longer statistically significant in most cases [145,147,154]. After adjusting for total body LM, El Hage et al. [146] found that whole body FM negatively predicted whole body and lumbar spine aBMD in a sample of 65 Caucasian boys between 14 and 16 years. Two studies assessed the influence of FM as a continuous variable on total body BMC [147,153] and total body aBMD in children aged 4 – 20 yrs [153]. One reported a negative relationship between FM and bone mass [153] whereas the other reported a weak positive relationship [147]. Weiler et al. recruited a relatively small sample size (n = 60 girls, 10 - 19 yrs) and adjusted regression models for weight, height, and %BF, but not LM [153]. This analysis may have been confounded by multicollinearity due to the relation of weight and height with %BF. It may also have been underpowered to adequately assess this association due to the small sample size; therefore, I interpret the negative results with caution. Conversely, Ackerman et al. fit regression models and used LM, age, height, Tanner stage and ethnicity as covariates in a large sample of children (n = 444 girls, n = 482 boys; 6 - 18 yrs) [147]. They reported a weak positive association between FM and total body BMC in prepubertal boys and girls and pubertal 34  Chapter 2 – Literature Review girls [147]. Associations between fat and bone variables were generally weaker when LM was added to regression models, highlighting the important role that LM plays in relation to FM. Clark et al. [154] assessed a mixed ethnic sample of pre- and peri-pubertal boys (n = 1585) and girls (n = 1918). They reported that two-year change (9.9 to 11.8 yrs) in FM positively predicted change in total body bone area and BMC after adjusting for age, ethnicity, socioeconomic status and change in height. However, once LM was added to the models, the variance explained by FM decreased. In a subset of these participants (108 Tanner stage 3 girls), Clark et al. [154] found that two-year change in fat mass negatively predicted change in total body bone area after adjusting for age, ethnicity, socioeconomic status, height, and change in LM. Results from these DXA studies suggest a positive relationship between total body FM and DXA derived bone variables (BMC, aBMD, BA); however, once LM was accounted for this relationship weakened, disappeared or even became negative. This is consistent with the functional model of bone development which suggests that muscle force exerts the largest loads on bone [7]. Thus, it seems that fat mass has only a weak positive, if any, influence on two dimensional bone properties (BMC, aBMD, BA) measured by DXA. Given the greater variance explained by lean mass when it is added regression models and the diminished role of fat, it holds that lean mass exerts a greater influence on two dimensional bone properties. There are a number of other factors (e.g. age, LM, ethnicity) and outcomes (bone geometry, structure, strength) that must be considered when assessing the fat-bone relationship. However, 2D bone measures by DXA do not permit evaluation of the relationship between fat mass and bone geometry, BMD or estimates of bone strength. Thus, to gain a broader understanding of the scope of the relationship between adiposity and bone, researchers must evaluate estimates of bone strength and measures of bone geometry. Although relatively few studies examined this relationship, I present those that did in Section 2.4.4.  35  Chapter 2 – Literature Review  Table 3. Summary of studies that used DXA to assess the fat-bone relationship. First Author Goulding (2000) [150]  Participants & Design Design: Cross-sectional  Statistical Approach Regression:  Results – Not adjusted for lean Regression:  Participants: 200 girls, 136 boys, by weight category (HW, OW, OB), Caucasian  Model 1: Age, age2, ln(FM)  WB: Model 1: Girls: BMC: NS  Age: 3-19 yrs Bone measurement site: WB  Boys: BMC: OB > HW OW > HW  Other measures: Tanner stage, FM Weiler (2000) [153]  Design: Cross-sectional  Regression:  Regression:  Participants: 60 girls, ethnicity not reported  Covariates: Age, weight, height, %BF  WB: %BF coefficients BMC: aBMD: BMC/BA: -  Design: Cross-sectional  ANOVA:  n/a  Participants: 444 girls, 421 boys, mixed ethnicities  Covariates: sex, ethnicity, age, Lm  Age: 10-19 yrs Bone measurement site: WB Other measures: %BF Ellis (2003) [152]  Age: 7-16 yrs Bone measurement site: WB Other measures: Tanner stage, body fat category (HW, OW, OB)  Results – Adjusted for lean n/a  n/a  ANOVA: Differences between body fat category, not adjusted for fat mass WB Boys: BMC: NS Girls: BMC: OB > HW; OW > HW  Positive relationships are indicated with + or ++ (++ is a stronger positive relationship than +). Negative relationships are indicated with - or -- (-- is a stronger negative relationship than -). Non-significant relationships are indicated with NS. Abbreviations: DXA, Dual energy X-ray absorptiometry; LS, lumbar spine; WB, whole body; FS, femoral shaft; FN, femoral neck; NN, narrow neck; FM, fat mass; LM, lean mass; BMC, bone mineral content; aBMD, areal bone mineral density; BMAD, bone mineral apparent density; HSA, hip structure analysis; OB, obese; OW, overweight; HW, healthy weight; SES, socioeconomic status.  36  Chapter 2 – Literature Review  First Author Leonard (2004) [149]  Participants & Design Design: Cross-sectional  Statistical Approach Regression:  Results – Not adjusted for lean Regression:  Results – Adjusted for lean Regression:  Participants: 103 OB, 132 HW, mixed ethnicities  Covariates: Model 1 LS: height, Tanner stage, ethnicity  Model 1 LS: aBMD: OB > HW BMC: NS Bone Area: Girls: OB < HW Boys: NS  Model 3 WB: BMC: OB > HW Bone area: OB > HW  Age: 4-20 yrs Bone measurement site: WB, LS Other measures: LM, FM, Tanner stage  Petit (2005) [148]  Model 2 WB: height Model 3 WB: height, Tanner stage, ethnicity, LM  Design: Cross-sectional  Regression:  Participants: 40 OW, 94 HW, 45% black  Model 1 covariates: height, maturation, gender, LM, FM  Age: 4-20 yrs  Model 2 WB: Bone area: OB > HW BMC: OB > HW n/a  Model 1: FS: ln(shaft Z): NS NN: ln(neck Z): NS  Bone measurement site: FS, NN with HSA  Wang (2005) [145]  Regression:  Design: Cross-sectional  Regression:  Model 1:  Model 2:  Participants: 921 girls, mixed ethnicities  Model 1 covariates: FM, ethnicity, height  LS: aBMD: + BMAD: NS  LS: aBMD: NS BMAD: NS  FN: aBMD: ++ BMAD: NS  FN: aBMD: + BMAD: NS  WB: aBMD: ++ BMC: ++  WB: aBMD: + BMC: +  Age: 20-25 yrs Bone measurement sites: LS, FN, WB Other measures: FM, LM  Model 2 covariates: LM, FM, ethnicity, height  Positive relationships are indicated with + or ++ (++ is a stronger positive relationship than +). Negative relationships are indicated with - or -- (-- is a stronger negative relationship than -). Non-significant relationships are indicated with NS. Abbreviations: DXA, Dual energy X-ray absorptiometry; LS, lumbar spine; WB, whole body; FS, femoral shaft; FN, femoral neck; NN, narrow neck; FM, fat mass; LM, lean mass; BMC, bone mineral content; aBMD, areal bone mineral density; BMAD, bone mineral apparent density; HSA, hip structure analysis; OB, obese; OW, overweight; HW, healthy weight; SES, socioeconomic status.  37  Chapter 2 – Literature Review  First Author Ackerman (2006) [147]  Participants & Design Design: Cross-sectional  Statistical Approach Regression:  Results – Not adjusted for lean Correlations: With ln(FM)  Results – Adjusted for lean Regression: ln(FM)  Participants: 444 girls, 482 boys, mixed ethnicities  Correlations: ln(LM) and ln(BMC)  Age: 6-18 yrs  Covariates: FM, age, height, LM, Tanner stage, sex, ethnicity  WB: Prepubertal girls: ln(BMC): ++  WB: Prepubertal girls: ln(BMC): +  Prepubertal boys: ln(BMC): ++  Prepubertal boys: Ln(BMC): +  Pubertal girls: ln(BMC): ++  Pubertal Girls: ln(BMC): +  Pubertal Boys: ln(BMC): +  Pubertal Boys: ln(BMC): NS n/a  Bone measurement site: WB Other measures: LM, FM, maturity by Tanner stage  Rocher (2008) [151]  Design: Cross-sectional  ANCOVA:  ANCOVA:  Participants: 20 OB, 23 HW prepubertal children, ethnicity not reported  Model 1 covariates: FM  Model 1: LS: aBMD: NS BMC: NS Bone area: NS  Age: 8-13 yrs Bone measurement site: WB, LS Other measures: FM  WB: aBMD: NS BMC: NS Bone area: NS BMAD: NS  Positive relationships are indicated with + or ++ (++ is a stronger positive relationship than +). Negative relationships are indicated with - or -- (-- is a stronger negative relationship than -). Non-significant relationships are indicated with NS. Abbreviations: DXA, Dual energy X-ray absorptiometry; LS, lumbar spine; WB, whole body; FS, femoral shaft; FN, femoral neck; NN, narrow neck; FM, fat mass; LM, lean mass; BMC, bone mineral content; aBMD, areal bone mineral density; BMAD, bone mineral apparent density; HSA, hip structure analysis; OB, obese; OW, overweight; HW, healthy weight; SES, socioeconomic status.  38  Chapter 2 – Literature Review  First Author El Hage (2009) [146]  Results – Not adjusted for lean Correlations:  Results – Adjusted for lean Regression:  Participants & Design Design: Cross-sectional  Statistical Approach Correlations:  Participants: 65 boys, 35 girls, Caucasian  Pearson correlations Boys: between FM and bone WB: outcomes BMC: NS aBMD: NS Regression: BMAD: Covariates: LM, FM LS: BMC: NS aBMD: NS BMAD: NS  Boys: WB: aBMD: -  Girls: WB: BMC: + aBMD: + BMAD: NS LS: BMC: NS aBMD: + BMAD: NS  Girls: WB: aBMD: +  Age: 14-16 yrs Bone measurement sites: LS, WB Other measures: FM, LM  LS: aBMD: -  Positive relationships are indicated with + or ++ (++ is a stronger positive relationship than +). Negative relationships are indicated with - or -- (-- is a stronger negative relationship than -). Non-significant relationships are indicated with NS. Abbreviations: DXA, Dual energy X-ray absorptiometry; LS, lumbar spine; WB, whole body; FS, femoral shaft; FN, femoral neck; NN, narrow neck; FM, fat mass; LM, lean mass; BMC, bone mineral content; aBMD, areal bone mineral density; BMAD, bone mineral apparent density; HSA, hip structure analysis; OB, obese; OW, overweight; HW, healthy weight; SES, socioeconomic status.  39  Chapter 2 – Literature Review  First author Clark (2006) [154]  Participants & Design Design: Longitudinal – 2 year follow up  Statistical Approach Regression:  Participants: 1585 boys, 1918 girls, ethnicity not Model 1: FM, age, reported ethnicity, SES, height (or ∆ height) Age: 9.9 yrs at baseline, 11.8 yrs at follow up Bone measurement site: WB Other measures: ethnicity, Tanner stage, SES, FM, LM  Model 2: FM, age, ethnicity, SES, height, LM (or ∆ height, ∆ LM) Separate models for boys, and Tanner stage 1, 2 and 3 girls  Results – Not adjusted for lean Regression Model 1:  Results – Adjusted for lean Regression Model 2:  Boys: WB: BA: ++ BMC: ++ ∆ BA: ++ ∆ BMC: ++  Boys: WB: BA: + BMC: + ∆ BA: + ∆ BMC: +  Girls Tanner 1: WB: BA: ++ BMC: ++ ∆ BA: ++ ∆ BMC: ++ Tanner 2: WB: BA: ++ BMC: ++ ∆ BA: NS ∆ BMC: NS Tanner 3: WB: BA: ++ BMC: ++ ∆ BA: ∆ BMC: NS  Girls Tanner 1: WB: BA: + BMC: + ∆ BA: + ∆ BMC: + Tanner 2: WB: BA: + BMC: + ∆ BA: NS ∆ BMC: NS Tanner 3: WB: BA: + BMC: + ∆ BA: ∆ BMC: NS  Positive relationships are indicated with + or ++ (++ is a stronger positive relationship than +). Negative relationships are indicated with - or -- (-- is a stronger negative relationship than -). Non-significant relationships are indicated with NS. Abbreviations: DXA, Dual energy X-ray absorptiometry; LS, lumbar spine; WB, whole body; FS, femoral shaft; FN, femoral neck; NN, narrow neck; FM, fat mass; LM, lean mass; BMC, bone mineral content; aBMD, areal bone mineral density; BMAD, bone mineral apparent density; HSA, hip structure analysis; OB, obese; OW, overweight; HW, healthy weight; SES, socioeconomic status.  40  Chapter 2 – Literature Review 2.4.4 Studies of the Fat-Bone Relationship using pQCT in Children and Adolescents In recent years a number of studies investigated the relationship between adiposity and pQCT-derived measures of bone strength, BMD and bone geometry (Table 4). They varied by sample size, age range, bone measurement site, adiposity measurement and statistical approach. Most were cross-sectional [9-15,155,156] and only a few were longitudinal studies [157,158]. I focus upon bone outcomes acquired at the distal radius or tibia as these sites might be more closely compared with those measured by HR-pQCT, the focus of my thesis. After careful evaluation of studies that used pQCT to assess the fat-bone relationship in children and adolescents, I conclude that adiposity is not beneficial for compressive bone strength. Nor is it a positive determinant of bone strength, geometry or density at the distal radius or tibia. Although many fat-bone studies that used pQCT reported positive bivariate correlations between total body FM and bone variables [11,13-15], after adjusting for total body LM, the positive influence of fat mass on bone parameters disappeared. I reiterate the importance of adjusting for surrogates of muscle force in studies that examine the relationship between fat and bone [7] in order to avoid biased results. Importantly, many studies did not find significant relationships between total body FM (by DXA) and bone geometry [10,12-14,156], density [9-13,156], or strength [10,13,14,156,157] (by pQCT) at metaphyseal sites after adjusting for total body LM or MCSA. To illustrate, in a mixed ethnic sample of adolescent girls (n = 396, 8 - 13 yrs), Farr et al. [14] did not find a significant relationship between total body FM and trabecular BMD (Tb.BMD), periosteal circumference (PC) and BSI at the distal femur and tibia (4% site; adjusted for MCSA, maturity offset, bone length, physical activity and ethnicity). Conversely, there was a negative relationship between fat (FM or fat cross-sectional area (FA) by pQCT) and bone variables after adjusting for surrogates of muscle force [9,10,12,158]. For example, Fricke et al. conducted a cross-sectional study of Caucasian boys (n = 139) and girls (n = 157) (5 - 19 yrs) [9]. They reported significant negative associations between total body FM and Tt.BMD and Tb.BMD at the 4% radius in pubertal boys (adjusted for height and MCSA). Further, a longitudinal study of 138 boys and 232 girls between 8 and 18 years reported predominantly negative associations between total body FM and bone variables (Tt.Ar, Ct.BMC, Tt.BMD) at the 4% radius in boys and girls (adjusted for age, height, total body LM and physical activity) [158].  41  Chapter 2 – Literature Review In contrast, some pQCT studies that investigated the fat-bone relationship reported a positive relationship between measures of adiposity and bone density [9-11]. Fricke et al. [9] found that in pubertal girls, fat cross-sectional area (by pQCT) was a positive predictor of Tt.BMD and Tb.BMD at the 4% radius. In an older (18 - 20 years) all male cohort (n = 1068), total body FM was a positive predictor of Tb.BMD at the 4% tibia after adjusting for age, height, physical activity, calcium and total body LM [11]. These results suggest that adiposity has a positive but weak relationship with BMD and a non-significant relationship with bone geometry and strength at metaphyseal sites. This is illustrated by the much smaller coefficients for FM compared with MCSA or total body LM in the regression models generated in these studies [911]. This stronger relation supports that muscle has a larger influence on specific bone measures as compared with FM. Results from studies that did not adjust for key variables such as LM add confusion to the bone-fat literature. For example, when Sayers et al. [155] did not include both FM and LM in their regression models, whole body FM was positively related to Ct.BMC and PC (boys n = 1851 and girls n = 2154; aged 15.5 years, on average). Similarly, in a 16-month longitudinal study of 302, 9-11 year old OW and HW children [157], ANCOVA was adjusted for sex, ethnicity, baseline Tanner stage and limb length. Consequently, OW children had higher Tt.Ar, Tt.BMD, BSI, Ct.Ar and SSIp compared with HW children. Further, change in these outcomes was greater in OW compared with HW children; however, neither LM nor FM was included as a covariate. Pearson correlations explored the relation between change in body composition and change in bone strength in the same study [157]. Change in FM was not related to change in bone strength (BSI and SSIp); however, change in LM was positively correlated with change in bone strength. In these studies it is likely that LM rather than FM was responsible for the positive relationships between FM or weight category and bone parameters. There are no published studies that examine the fat-bone relationship in children and adolescents using HR-pQCT. Consequently, there is a gap in the literature that this thesis aims to address, regarding the association between adiposity and bone microstructure in children and adolescents.  42  Chapter 2 – Literature Review  Table 4. Summary of studies that used pQCT to assess the fat-bone relationship in children and adolescents. First Author  Participants & Design  Lorentzon (2006) [11]  Design: Cross-sectional  Janicka (2007) [15]  Bone and Fat Measures  Bone measurement site: Non-dominant radius: 4% and Participants: 1068 boys, ethnicity 25% not reported Non-dominant tibia: 4% and 25% Age: 18-20 yrs Bone variables: Other measures: LM; serum 4%: Tb.BMD leptin; serum testosterone; 25%: Ct.BMD, Ct.BMC, physical activity (h/wk); calcium Ct.Ar, Ct.Th, PC, EC intake (mg/day); smoking status Fat measurement: FM Design: Cross-sectional Participants: 150 boys, 150 girls, all Tanner stage 5, all Caucasian  Bone measurement sites: with CT, not pQCT Bilateral femur: 50% Lumbar vertebrae 1-3: 50%  Age: 13-21 yrs Other measures: LM; skeletal age  Bone variables: Femur: Tt.Ar, Ct.Ar L1-3: Tt.Ar, Tb.BMD Fat measurement: FM  Results: Not adjusted for lean or MCSA n/a  Results: Adjusted for lean or MCSA Regression: FM on bone variables, adjusted for age, height, physical activity, calcium, LM 4% Tibia: Tb.BMD: + 25% Tibia: Ct.BMD, Ct.Ar, PC, EC: + Ct.Th: NS 4% Radius: Tb.BMD: NS 25% Radius: All variables: NS  n/a  Regression: FM on bone variables, adjusted for LM, leg length or trunk height Boys: 50% Femur: Tt.Ar: NS Ct.Ar: 50% L1-3: Tt.Ar: NS Tt.BMD: Girls: ALL variables NS  + indicates statistically significant positive relationships, - indicates negative. Abbreviations: LM, total body lean mass by DXA; FM, total body fat mass by DXA; %BF, percent body fat by DXA; FA, fat cross-sectional area by pQCT; MCSA, muscle cross-sectional area by pQCT; BMI, body mass index; Tt.Ar, total area; Tt.BMD, total bone mineral density; Tt.BMC, total bone mineral content; Ct.Ar, cortical area; Ct.BMD, cortical BMD; Ct.Th, cortical thickness; Ct.BMC, cortical BMC; Tb.BMD, trabecular BMD; EC, endosteal circumference; PC, periosteal circumference; SSIp, polar strength-strain index; BSI, bone strength index; PA, physical activity; NS, not significant; n/a, not applicable.  43  Chapter 2 – Literature Review  First Author  Participants & Design  Bone and Fat Measures  Pollock (2007) [12]  Design: Cross-sectional  Bone measurement sites: Non-dominant tibia: 4%, 20% Non-dominant radius: 4%, 20%  Participants: 115 girls, mixed ethnicities Age: 18-19 yrs Other measures: MCSA, 66% tibia and radius; LM  Fricke (2008) [9]  Design: Cross-sectional Participants: 139 boys (69 prepubertal, 70 pubertal), 157 girls (62 prepubertal, 95 pubertal), all Caucasian Age: 5-19 yrs Other measures: maturity (by Tanner stage), forearm length, MCSA 65% radius-  Results: Not adjusted for lean or MCSA n/a  Bone variables 4%: Tt.BMD, Tb.BMD 65%: Ct.BMD, Ct.Ar, EC, PC, SSIp Fat measurement: FA 65% radius  Partial correlations: FM with bone variables, adjusted for MCSA and limb length: 4% Tibia and Radius: All variables: NS  Bone variables: 4%: Tt.BMD, Tb.BMD, Tt.Ar 20%: Ct.BMD, Ct.Ar, Tt.Ar, Ct.BMC, Ct.Th, PC, EC, SSIp Fat measurement: FM, %BF Bone measurement sites: Non-dominant radius: 4% and 65%  Results: Adjusted for lean or MCSA  20% Tibia and Radius: Ct.BMD, Tt.Ar, Ct.Th, PC, EC, SSIp: NS Ct.Ar, Ct.BMC: n/a  Regression: FA on bone variables, adjusted for MCSA Prepubertal boys (Tanner 1): 65% Radius: Ct.BMD, EC: Ct.Ar: SSIp: + Pubertal boys (Tanner 2-5): 4% Radius: Tt.BMD, Tb.BMD: 65% Radius: Ct.BMD, Ct.Ar, SSIp: PC: + Prepubertal girls (Tanner 1): 4% Radius: Tt.BMD, Tb.BMD: + 65% Radius: Ct.BMD, PC, Ct.Ar, SSIp: Pubertal girls (Tanner 2-5): 4% Radius: Tt.BMD, Tb.BMD: + 65% Radius: EC, PC, Ct.Ar, SSIp: + Ct.BMD: -  + indicates statistically significant positive relationships, - indicates negative. Abbreviations: LM, total body lean mass by DXA; FM, total body fat mass by DXA; %BF, percent body fat by DXA; FA, fat cross-sectional area by pQCT; MCSA, muscle cross-sectional area by pQCT; BMI, body mass index; Tt.Ar, total area; Tt.BMD, total bone mineral density; Tt.BMC, total bone mineral content; Ct.Ar, cortical area; Ct.BMD, cortical BMD; Ct.Th, cortical thickness; Ct.BMC, cortical BMC; Tb.BMD, trabecular BMD; EC, endosteal circumference; PC, periosteal circumference; SSIp, polar strength-strain index; BSI, bone strength index; PA, physical activity; NS, not significant; n/a, not applicable.  44  Chapter 2 – Literature Review  First Author  Participants & Design  Bone and Fat Measures  Ducher (2009) [10]  Design: Cross-sectional  Bone measurement sites: Non-dominant radius: 4%, 66% Contralateral tibia: 4%, 66%  Participants: 221 girls, 206 boys, mixed ethnicities  Bone variables: 4%: BMC, Tt.Ar, Tt.BMD, Other measures: MCSA, 66% Tb.BMD, Ct.Th, BSI radius and tibia; BMI 66%: BMC, Tt.Ar, Ct.Ar, classification by IOTF cut points, Ct.Th, Ct.BMD, SSIp estimated fall load, fat-muscle ratio Fat measurement: FA 66% radius and tibia  Results: Not adjusted for lean or MCSA ANCOVA: adjusted for height  Results: Adjusted for lean or MCSA Partial Correlations: FA with bone variables, adjusted for MCSA  4% Radius: BMC, Tb.BMD, BSI: OW > HW Tt.Ar, Ct.Th: NS  4% Radius: All variables: NS  66% Radius: BMC, Tt.Ar, Ct.Ar, SSI: OW > HW Ct.Th, Ct.BMD: NS  66% Radius: BMC, SSIp, Ct.Ar: + Tt.Ar, Ct.Th, Ct.BMD: NS  4% Tibia: BMC, Tt.Ar, Tb.BMD, BSI: OW > HW Ct.Th: NS  4% Tibia: Tt.Ar: + BMC, Tb.BMD, BSI, Ct.Th: NS  Age: 7-10 yrs  66% Tibia: BMC, Tt.Ar, Ct.Ar, Ct.Th, SSI: OW > HW Ct.BMD: NS Farr (2010) [14]  Design: Cross-sectional Participants: 396 girls, mixed ethnicities (87% Caucasian) Age: 8-13 yrs Other Measures: maturity (Tanner stage and maturity offset), physical activity, calcium, LM , MCSA 66%  Bone measurement sites: Non-dominant femur: 4%, 20% Non-dominant tibia: 4%, 66% Bone variables: 4%: Tb.BMD, PC, BSI 20%, 66%: Ct.BMD, PC, SSIp  n/a  66% Tibia: BMC, SSIp, Tt.Ar, Ct.Ar: + Ct.BMD: NS  Regression: FM on bone variables, adjusted for MCSA, maturity offset, bone length, PA, ethnicity 4% Femur and 4% Tibia: All variables: NS 20% Femur and 66% Tibia: All variables: NS  Fat Measurement: FM  + indicates statistically significant positive relationships, - indicates negative. Abbreviations: LM, total body lean mass by DXA; FM, total body fat mass by DXA; %BF, percent body fat by DXA; FA, fat cross-sectional area by pQCT; MCSA, muscle cross-sectional area by pQCT; BMI, body mass index; Tt.Ar, total area; Tt.BMD, total bone mineral density; Tt.BMC, total bone mineral content; Ct.Ar, cortical area; Ct.BMD, cortical BMD; Ct.Th, cortical thickness; Ct.BMC, cortical BMC; Tb.BMD, trabecular BMD; EC, endosteal circumference; PC, periosteal circumference; SSIp, polar strength-strain index; BSI, bone strength index; PA, physical activity; NS, not significant; n/a, not applicable.  45  Chapter 2 – Literature Review  First Author  Participants & Design  Bone and Fat Measures  Sayers (2010) [155]  Design: Cross-sectional  Bone measurement: Tibia (no side specified): 50%  Participants: 1851 boys, 2154 girls, ethnicity not reported  Pollock (2011) [13]  Age: 15.5 yrs  Bone variables: Ct.BMC, Ct.BMD, Ct.Ar, PC, EC, Ct.Th  Other measures: LM; maturity (Tanner stage)  Fat measurement: FM  Design: Cross-sectional  Bone measurement site: Non-dominant radius: 4% and 20% Non-dominant tibia: 4% and 20%  Participants: 48 girls, all black; 33 HW, 15 OW (classified by % BF) Age: 18-22 yrs Other measures: limb lengths, LM, MCSA 66%  Bone variables: 4%: Tt.BMD, Tb.BMD, Tt.Ar, BSI 20%: Tt.Ar, Ct.BMD, Ct.Ar, Ct.BMC, Ct.Th, EC, SSIp Fat measurement: FM, %BF  Results: Not adjusted for lean or MCSA Regression: FM on bone variables Model 1: adjusted for height Boys & Girls: Ct.BMC, PC: + Model 2: adjusted for height, PC Boys & Girls: EC: Model 3: adjusted for height, PC, EC Boys & Girls: Ct.BMD: +  Results: Adjusted for lean or MCSA  n/a  Partial correlations: FM with bone variables, adjust for LM  n/a  4% Radius and Tibia: All variables: NS 20% Radius: Ct.Ar, Tt.Ar, EC, SSIp: Ct.BMD, Ct.BMC, Ct.Th: NS 20% Tibia: Tt.Ar, SSIp: Ct.BMD, Ct.Ar, Ct.BMC, Ct.Th, EC: NS  + indicates statistically significant positive relationships, - indicates negative. Abbreviations: LM, total body lean mass by DXA; FM, total body fat mass by DXA; %BF, percent body fat by DXA; FA, fat cross-sectional area by pQCT; MCSA, muscle cross-sectional area by pQCT; BMI, body mass index; Tt.Ar, total area; Tt.BMD, total bone mineral density; Tt.BMC, total bone mineral content; Ct.Ar, cortical area; Ct.BMD, cortical BMD; Ct.Th, cortical thickness; Ct.BMC, cortical BMC; Tb.BMD, trabecular BMD; EC, endosteal circumference; PC, periosteal circumference; SSIp, polar strength-strain index; BSI, bone strength index; PA, physical activity; NS, not significant; n/a, not applicable.  46  Chapter 2 – Literature Review  First Author  Participants & Design  Bone and Fat Measures  Viljakainen (2011) [156]  Design: Cross-sectional  Bone measurement sites: Non-dominant radius: 4% and 66%  Participants: 113 girls, 73 boys, ethnicity not reported Age: 7-19 yrs  Wetzsteon (2008) [157]  Other measures: MCSA, muscle density (66% site); calcium (mg/day), physical activity (PA score), pubertal development (pre-pubertal, pubertal, postpubertal, by sex steroid concentrations) Design: Longitudinal (baseline and 16-month follow up)  Bone variables: 4%: Tt.Ar, Tt.BMC, Tt.BMD, Tb.Ar, Tb.BMD 66%: Ct.Ar, Ct.BMD, SSIp Fat measurement: FM, %BF  Bone measurement sites: Left tibia: 8%, 50%, 66%  Participants: 302 HW (165 girls), Bone variables: 143 OW (54 girls), mixed 8%: Tt.Ar, Tt.BMD, BSI ethnicities 50% & 66%: Tt.Ar, Ct.Ar, Ct.BMD, SSIp Age: 9-11 yrs Fat measurement: Other measures: maturity by BMI for age groups (Tanner stage, menarche status); (HW/OW); FM physical activity (PA score); calcium (mg/day), tibia length; MCSA, 66%; LM  Results: Not adjusted for lean or MCSA Partial correlations: %BF and bone variables, adjusted for pubertal development 4% Radius: Tt.BMD, Tt.BMD, Tb.Ar, Tb.BMD: NS Tt.Ar: 66% Radius: Ct.Ar: NS ln(SSIp): Ct.BMD: + ANCOVA: HW/OW with bone variables, adjusted for sex, ethnicity, baseline Tanner stage, tibia length  Results: Adjusted for lean or MCSA MANOVA: Tertiles of % fat z-scores (I= low, II = intermediate, III = high) by sex, adjusted for LM, pubertal development, calcium, physical activity 4% Radius: All variables: NS 66% Radius: Ct.Ar, SSIp: NS Ct.BMD: II >I; II > III  n/a  Baseline: 8% Tibia: Tt.Ar, Tt.BMD, BSI: OW > HW 50% & 66% Tibia: Tt.Ar, Ct.Ar, SSIp: OW > HW Ct.BMD: NS Follow up: (change in bone vars) 8% Tibia: Tt.Ar: OW > HW 50% & 66% Tibia: Tt.Ar, SSIp: OW > HW 50% Tibia: Ct.Ar: OW > HW Correlation: ∆ FM vs. ∆ BSI or SSIp: NS  + indicates statistically significant positive relationships, - indicates negative. Abbreviations: LM, total body lean mass by DXA; FM, total body fat mass by DXA; %BF, percent body fat by DXA; FA, fat cross-sectional area by pQCT; MCSA, muscle cross-sectional area by pQCT; BMI, body mass index; Tt.Ar, total area; Tt.BMD, total bone mineral density; Tt.BMC, total bone mineral content; Ct.Ar, cortical area; Ct.BMD, cortical BMD; Ct.Th, cortical thickness; Ct.BMC, cortical BMC; Tb.BMD, trabecular BMD; EC, endosteal circumference; PC, periosteal circumference; SSIp, polar strength-strain index; BSI, bone strength index; PA, physical activity; NS, not significant; n/a, not applicable.  47  Chapter 2 – Literature Review  First Author  Participants & Design  Bone and Fat Measures  Wey (2011) [158]  Design: Longitudinal (baseline, 3 year follow up)  Bone measurement sites: Left radius: 4%, 20%  Participants: 138 boys, 232 girls, Hutterites  Bone variables: 4%: Tt.BMC, Tt.BMD, Tt.Ar 20%: Tt.Ar, Ct.Th, Ct.Ar, Ct.BMC, Ct.BMD, SSIp  Age: 8-18 yrs Other measures: LM; physical activity (7-day recall); menarche status  Fat measurement: FM  Results: Not adjusted for lean or MCSA n/a  Results: Adjusted for lean or MCSA Mixed model regression: FM, on bone variables, adjusted for height, LM, PA Boys: Cross-sectional 4% Radius: Tt.Ar, Ct.BMC: 20% Radius: Tt.Ar, Ct.Ar, Ct.BMC, SSIp: Longitudinal: 4% Radius: Tt.BMD: 20% Radius: Ct.Ar, Ct.Th, Ct.BMC, SSIp: Girls: Cross-sectional 20% Radius: Tt.Ar, Ct.Ar, Ct.BMC, SSIp: 4% Radius: Tt.Ar, Ct.BMC: Longitudinal 20% Radius: Ct.BMC, Ct.BMD, SSIp: + in young girls, - in older girls 4% Radius: Tt.BMC, Tt.Ar: - in premenarcheal  + indicates statistically significant positive relationships, - indicates negative. Abbreviations: LM, total body lean mass by DXA; FM, total body fat mass by DXA; %BF, percent body fat by DXA; FA, fat cross-sectional area by pQCT; MCSA, muscle cross-sectional area by pQCT; BMI, body mass index; Tt.Ar, total area; Tt.BMD, total bone mineral density; Tt.BMC, total bone mineral content; Ct.Ar, cortical area; Ct.BMD, cortical BMD; Ct.Th, cortical thickness; Ct.BMC, cortical BMC; Tb.BMD, trabecular BMD; EC, endosteal circumference; PC, periosteal circumference; SSIp, polar strength-strain index; BSI, bone strength index; PA, physical activity; NS, not significant; n/a, not applicable.  48  Chapter 2 – Literature Review 2.5 Summary of the Literature Bone microstructure, geometry and density all contribute to bone strength [43]. Further, there are both non-modifiable and modifiable determinants of bone strength that include genetics [70], hormones [71], physical activity [107], and diet [110]. Central to bone strength is the muscle-bone relationship. The functional model of bone development proposes that physiologic loads from muscle forces stimulate bones to self-regulate to maintain functional structural integrity and strength [7,105] and a number of excellent studies support this notion [61,105,116]. Conversely, relatively few studies evaluated the fat-bone structure or strength relationship in children, adolescents and young adults. The literature in this field is varied and equivocal given the broad age range of those studied and the various methods used to measure different bone parameters often at different anatomical sites. However, there was a consistent pattern when surrogates of muscle force were included in regression in that the relationship between fat and bone became very weak [9-11] or negative [9,10,12,158] in children and adolescents. Thus, it seems important to evaluate the fat-bone relationship within the context of the mechanostat theory [6] and functional model of bone development [7]. To date no study has examined the fatbone microstructure relationship. Thus, the aim of this thesis is to increase our understanding and advance knowledge regarding the association between fat mass and bone microstructure, geometry, density and strength in children, adolescents and young adults.  49  Chapter 3 - Rationale, Objectives and Hypotheses  Chapter 3 - Rationale, Objectives and Hypotheses Rationale: The prevalence of childhood overweight and obesity has risen dramatically in recent years [1]. Overweight children sustain more fractures than their healthy weight peers [3]; however, studies investigating the relationship between fat and bone are thus far equivocal and conflicting. Discrepancies across studies may reflect the wide variation in medical imaging systems, sites and protocols used to assess bone and fat. Specifically, commonly used planar DXA technology has many well known limitations and imaging results have been interpreted inconsistently with many not accounting for the functional model of bone development [7]. In addition, most studies focused upon bone mass (by DXA) or bone structure (by pQCT). The advent of HR-pQCT technology enabled evaluation of the hierarchical nature of bone including cortical and trabecular bone microstructure, in vivo [159,160]. Furthermore, the application of finite element analysis to HR-pQCT scans permits more accurate estimates of bone strength. Taken together a unique opportunity exists to better understand the relationship between these two important tissues, specifically, how fat mass influences bone strength and microstructure as no studies have as yet examined this. Objectives: The primary objective of this thesis is to determine the relationship between adiposity and bone strength at the distal radius and distal tibia in the context of the functional model of bone development in children, adolescents and young adults. The secondary objective is to determine the relationship between adiposity and components of bone strength including bone density, geometry and microstructure at the distal radius and distal tibia in the context of the functional model of bone development in children, adolescents and young adults. Primary Hypothesis: After adjusting for lean mass (a surrogate of muscle force), adiposity will not be positively associated bone strength at the distal radius and distal tibia in boys or girls. Secondary Hypothesis: After adjusting for lean mass (a surrogate of muscle force), adiposity will not be positively associated with bone density, geometry or microstructure at the distal radius and distal tibia in boys or girls. Contribution: This is the first study to explore the fat-bone relationship in children, adolescents and young adults with a focus on bone microstructure measured using HR-pQCT and bone strength estimated using finite element analysis. The findings of this research may help to explain 50  Chapter 3 - Rationale, Objectives and Hypotheses why overweight children and adolescents sustain more fractures than their overweight peers. It may also serve to support the need for effective intervention studies that endeavour to promote healthy weight during growth and development.  51  Chapter 4 – Methods  Chapter 4 – Methods In this chapter I describe the study design, participant recruitment and all relevant measurements including acquisition and analysis of HR-pQCT scans. I also outline the statistical analyses I used to address my primary and secondary study objectives. My contributions to this thesis include data cleaning and analysis.  4.1 Study Design For this cross-sectional observational study, I used data from the 2009 annual measurement of the longitudinal Healthy Bones III Study (HBS III, described in Section 4.2). I chose this measurement period for two reasons: first, this was the first year HR-pQCT radius scans were included as part of HBS III protocol and second, we recruited a younger cohort of participants to HBS III who had their first measurements in spring 2009 (details in Section 4.2). All measurements took place in the Bone Health Research Lab at the Centre for Hip Health and Mobility, Vancouver Coastal Health Research Institute (VCHRI) in May and June 2009. We obtained ethical approval from the UBC Clinical Research Ethics Board and VCHRI.  4.2 The Healthy Bones III Study Cohort 4.2.1 Cohort Description and Participant Recruitment HBS III is a mixed longitudinal, observational study that includes participants recruited for three intervention studies between 1999 and 2003 and as part of this prospective trial, in 2009. Intervention studies were implemented across 11 months (Healthy Bones II Study, HBS II: 1999 - 2001, n = 383); 8 months (Bounce at the Bell: 2000 - 2001, n = 51) or 16 months (Action Schools! British Columbia, AS! BC: 2003 - 2004, n = 515). Children (10.3 ± 0.6 years of age on average) were recruited from 24 elementary schools in Vancouver and Richmond school districts. At the cessation of each intervention, children were invited to participate in annual follow-up measurements. Details of each school-based intervention are beyond the scope of this thesis but are described in detail elsewhere [98,161-163]. In 2009, a fourth cohort of children (n = 121, aged 10.5 ± 0.6 yrs; new recruits) was recruited from five elementary schools in the Vancouver and Richmond districts. This fourth cohort was recruited to increase the number of pre- and early- pubertal children from whom we obtained measures of bone microstructure using HR-  52  Chapter 4 – Methods pQCT. Together, these four cohorts comprise the HBS III cohort, the dataset that is the focus of this thesis. Recruitment methods were similar for the HBS II and AS! BC studies. Briefly, the recruitment team first made presentations to school principals at district meetings, and interested principals then volunteered their school to participate. The study team then presented an overview of the study to grade 4, 5 and 6 teachers in volunteer schools. Finally, the study team presented the study to grade 4, 5 and 6 students and answered any questions. Following these meetings, information letters and consent forms were given to classroom teachers who then distributed the forms to students to take home to their parents. Consent and assent for participation in the follow-up study was obtained in 2001, 2003, 2006 and 2009, for HBS II and Bounce at the Bell participants and in 2004, 2006, 2007 and 2009 for AS! BC participants. We used a similar approach to recruit the cohort of Grade 4 and 5 students in 2009 from 5 schools that previously participated in HBS II and AS! BC. Information letters and consent forms (Appendix A) were distributed in the classroom to grade 4 and 5 students at each school.  4.2.2 Participant Scheduling and Measurement HBS III study participant volunteers were contacted in two ways to schedule measurement appointments. The research coordinator contacted students attending secondary schools directly by telephone. Students attending the same secondary school had measurement appointments booked at a similar time and travelled to the lab by minivan in groups of up to 6. For participants attending elementary school, the research coordinator contacted their teachers and arranged for participants from each class to be picked up at the school door and travel to the lab in groups of 5. Those for whom we received parental consent were excused from class for approximately 3 hours and transported to the Centre for Hip Health and Mobility (the Centre) for measurement. The driver and one other research assistant served to transport and chaperone participants for their trip from the elementary schools to the Centre. When participants arrived at the lab they rotated through 6 stations: 1. anthropometry (5 mins), 2. questionnaires (30 mins), 3. jumps (15 mins), 4. DXA (20 mins), 5. HR-pQCT (20 mins) and 6. pQCT (10 mins). Participants were supervised by a research team member at all times.  53  Chapter 4 – Methods Prior to measurement the research coordinator conducted a full-day training session that all members of the research team attended. The training session outlined the protocol for all measures and measurement team members were trained to correctly conduct procedures (such as anthropometry) and to administer questionnaires. Team members practiced all measurements during the training session to maintain quality assurance. Measurers were also advised on the ethics of data collection. Medical imaging was conducted by individuals specifically trained in these procedures.  4.3 Measurements 4.3.1 Anthropometry Anthropometric measures included height, weight, tibial length and ulnar length. We measured standing height (stretch stature) twice to the nearest 0.1 cm using a wall-mounted stadiometer (model 242; Seca, Hanover, MD). If the two measures were within 0.4 cm, the average of the two values was used. If measures were greater than 0.4 cm apart, a third measure was taken and the closest two of the three values within 0.4 cm were averaged for the final height measure. We measured weight to the nearest 0.1 kg using an electronic scale (Seca model 840). Similar to height, the average of two measures was used if the two values were within 0.2 kg of each other. We calculated body mass index (BMI) as weight (kg) divided by height (m) squared. For imaging the lower leg, we used an anthropometric tape to measure the tibial length in duplicate as the distance from the distal edge of the medial malleolus to the medial edge of the tibial plafond (nearest millimetre). For imaging the forearm, we measured ulnar length in duplicate as the distance from the olecranon process to the distal and medial edge of the ulnar styloid process (nearest millimetre). We measured ulnar length rather than radial length due to the difficulty of landmarking the distal end of the radius.  4.3.2 Maturity We assessed maturity by self-reported breast and pubic hair stage for girls and boys, respectively, based on the method of Tanner [164]. Participants were asked to select from a set of line drawings that depict the 5 stages of development of secondary sex characteristics the drawing he/she felt was most similar to his/her own physical appearance (Appendix B). The questionnaire has pictures of breast and pubic hair stages for girls and pubic hair stages for boys. The drawings were accompanied by a brief description of the visual appearance at each stage. 54  Chapter 4 – Methods Study participants completed the questionnaire in private after they were provided standard instructions from a research assistant. They returned the completed questionnaire in a sealed envelope. For girls, I used breast stage only to represent stage of maturity due to the known differences in timing between the onset of pubic hair and breast development [165] and the association between estrogen levels and breast development [166]. Participants who had reached maturity (Tanner stage 5) as per a previous data collection were not required to complete the questionnaire. For girls, we also determined menarcheal status. A research assistant asked female participants if they had experienced their first menstrual period. If they replied “yes”, they were asked to recall the approximate date. Measurers did not ask this question if girls reported reaching menarche during a previous year’s data collection.  4.3.3 Health History Questionnaire At baseline (HBS II: 1999, Bounce at the Bell: 2000, AS! BC: 2003, new recruits: 2009) participants’ parents or guardians completed a health history questionnaire that addressed medical history, medication use and ethnicity (Appendix B). This questionnaire was used to identify participants with medical conditions and/or taking medications known to influence bone metabolism or that would prevent participation in regular physical activity. Ethnicity was based on parents’ or grandparents’ place of birth. articipants were classified as ‘Asian’ if both parents or all 4 grandparents were born in Hong Kong, China, India, hilippines, Vietnam, Korea or Taiwan. We classified participants as ‘Caucasian’ if both parents or all 4 grandparents were born in North America or Europe. Participants of other ethnic origins (i.e. Black or Hispanic) or of mixed ethnicities were classified as ‘Other’. Parents also selfreported their child’s ethnicity as, for e ample, Asian-Canadian or Caucasian-Canadian. To screen for changes in medical history during the follow-up, participants completed a shorter version of the health history questionnaire at all subsequent measurement periods (Appendix B).  4.3.4 Body Composition Total body fat mass (FM, kg) and bone mineral-free lean mass (LM, kg) were measured from DXA total body scans (QDR 4500W; Hologic, Waltham, MA, USA). The effective radiation dose of each total body scan for persons 10 yrs or older is 4.2 - 4.8 µSv [167], an 55  Chapter 4 – Methods extremely low and very safe dose. One trained technician scanned all participants using standard acquisition procedures [168]. Prior to positioning, participants were asked to remove all removable metal objects from their body. Metal-free clothing was provided to participants, if necessary. Participants were asked to lay down on the scanning table with their arms by their sides, palms on the table with a space between the hands and the thighs. The technician rotated the feet so the toes were touching and heels apart. The technician wrapped a Velcro strap around the feet to ensure the participants remained stationary for the duration of the 6 minute scan. The technician ensured proper alignment of each participant within the scan limit border lines. The same technician analyzed the total body scans according to standard procedures [168] (Figure 11). A spine phantom and anthropomorphic phantom were scanned daily and weekly, respectively, to maintain quality assurance of the DXA system. In adults, the %CV with repositioning in our lab was 1.9% for fat mass and 0.33% for lean mass. Due to unnecessary exposure to ionizing radiation we did not conduct a study in our lab to determine the precision of total body fat and lean mass measures in children and adolescents.  Figure 11. Sample image of an analyzed total body DXA scan of a 12-year old girl. 56  Chapter 4 – Methods 4.3.5 Bone Strength, Geometry, Density and Microstructure Bone strength, geometry, density and microstructure at the distal radius and distal tibia were measured using HR-pQCT (XtremeCT; Scanco Medical, Brüttisellen, Switzerland).  4.3.5.1 HR-pQCT Scan Acquisition For the radius and tibia, we used a standard region of interest (ROI) to assess the same relative site from year to year [159,160]. The ROIs included both cortical and trabecular bone and excluded the tibial and radial growth plates. Technicians scanned the non-dominant radius and tibia of each participant, unless the non-dominant limb had a history of fracture, in which case the contralateral limb was measured. Prior to each scan, the HR-pQCT technician immobilized the limb in a carbon fibre cast shaped for the forearm or leg (Figure 12A-B). For the forearm scan, the technician placed additional padding around the hand and wrist to minimize movement; no additional padding was needed for tibia scans. The technician then placed the limb in the system gantry and adjusted the chair to ensure the participant was as comfortable as possible (Figure 12C-D). The technician explained to the participants the importance of being still during the scans and dimmed the lights and played soft music to create a relaxing environment. The same technician performed all scans for all participants. A density and volume phantom was scanned daily to ensure quality control.  Figure 12. Set up for HR-pQCT radius (A, C) and tibia (B, D) scans. 57  Chapter 4 – Methods Step one during scan acquisition involved placing the reference line. The technician acquired an anterior-posterior scout view to identify the relative ROI for each scan. For the radius scan, the reference line was placed at the medial edge of the distal radius and the ROI was defined as the region distal to 7% of ulnar length from the reference line (Figure 13A). For the tibia scan, the reference line was placed at the tibial plafond and the ROI was defined as the region distal to 8% of tibial length from this reference line (Figure 13B). Following the scout view scan, 110 parallel CT slices with an isotropic voxel size of 82 µm were obtained from distal to proximal covering approximately 9.02 mm. The following settings were used for both radius and tibia scans: effective energy of 60 kVp, X-ray tube current of 900 µA, integration time of 100 ms, and matrix size of 1536 × 1536. The effective radiation dose for each scan was < 3 µSv. Each scan took approximately 3 minutes to complete and the scan was repeated once if there were significant motion artifacts.  Figure 13. Scout view images of the distal radius (A) and distal tibia (B) including the relative regions of interest (ROI). 4.3.5.2 Scan Analysis The same technician who acquired the images contoured all 110 slices and analyzed HRpQCT scans. As per standard procedure, the 3 most proximal and 3 most distal slices were omitted from analysis, thus final values are based on 104 slices. Three analyses were performed. First, the standard manufacturer’s analysis was used to obtain measures of trabecular microstructure (Tb.N, Tb.Th, Tb.Sp, BV/TV) and Tt.BMD [62]. Second, a customized segmentation algorithm [63] was applied to obtain measures of cortical microstructure and BMD (Ct.Po, Ct.Th, Ct.BMD) and total bone area (Tt.Ar). Third, finite element analysis was applied to the HR-pQCT scans to estimate bone strength (FLoad, N; UStress, MPa). I provide details of 58  Chapter 4 – Methods each method below. During scan analysis, the technician graded all scans for motion artifacts according to the manufacturer’s motion grading scale [169]. Scans with a motion grade of 4 or 5 were excluded from analysis as they had either had discontinuities in the cortex or large amounts of streaking. Scans with a motion grade of 3 or lower were included as they had no breaks in the cortex (Figure 14).  Figure 14. Sample images of a motion grade 1 scan (left) and a motion grade 4 scan (right). Standard Analysis All images were segmented using the system software as per the manufacturer’s standard in vivo protocol for morphological analysis [170]. Attenuation data were converted by the system into equivalent hydroxyapatite (HA) data. Total density (Tt.BMD, mgHA/cm3) was directly measured as the mass of bone tissue in the volume of interest. Cortical and trabecular regions were identified using a threshold-based algorithm. The threshold was set to one third of the apparent volumetric density of cortical bone (Ct.BMD, mg HA/cm3) in order to differentiate trabecular from cortical bone. Trabecular outcomes are either directly measured or derived from other measures. Trabecular density (Tb.BMD, mg HA/cm3) is directly measured and calculated as the mean trabecular density in the volume of interest [62]. Trabecular bone volume (BV) fraction (BV/trabecular volume (TV)) is derived from Tb.BMD as: BV TV  =  Tb.BMD mgHA cm mgHA cm  (Equation 3)  assuming the density of fully mineralized bone is 1200 mg HA/cm3. Individual trabeculae are too thin (100-300 µm) for properties such as Tb.Th and Tb.Sp to be accurately measured with HRpQCT, thus the standard manufacturer thickness-independent algorithm is used to assess trabecular structure [171]. Trabecular number is a direct 3-dimensional measure calculated as the 59  Chapter 4 – Methods mean of the inverse spacing between trabecular mid-axes. This measure is not dependent on underlying assumptions of the plate- or rod-like nature of the trabecular structure [171]. Trabecular thickness is derived from BV/TV and Tb.N [62] as: Tb.Th (mm)  =  BV TV Tb.  1 mm  (Equation 4)  Trabecular separation is also derived from the same measures as: Tb.Sp (mm) =  1 - BV TV Tb.  1 mm  (Equation 5)  These trabecular variables were highly correlated (r2 = 0.81 – 0.96) with the same variables in an assessment ex vivo, using micro-CT (voxel size 28 µm) [171]. Reproducibility for HR-pQCT standard analysis acquired outcomes was 0.5-4.5% at the distal radius and distal tibia based on 20 scans each of 15 male and 15 female participants, 10 with repositioning and 10 without [172].  Auto-segmentation Analysis Although it is the current gold standard, the standard HR-pQCT morphological analysis is limited by its semi-automated nature and reliance on hand contouring by the operator [63]. In addition, problems with segmentation may arise when analyzing scans with a thin cortical shell. The standard analysis assumes that the cortex is thick compared with individual trabeculae; however, if this assumption is not met, the cortex may be segmented incorrectly. Finally, the standard analysis does not permit measurements of cortical porosity, which is an important determinant of bone strength [52]. Recently, Buie and colleagues developed a fully-automated segmentation algorithm for HR-pQCT scan analysis that addresses these limitations [63]. We applied this algorithm to our HR-pQCT scans. This algorithm used two thresholds: one to identify the periosteal surface and one to identify the endosteal surface (Figure 15). The threshold used to identify the periosteal surface extracts the non-bone region with the assumption that the cortical shell encapsulates the trabecular region [63]. Bone tissue is then extracted from the image. The second threshold is masked with the non-bone region from the first threshold, which leaves only marrow cavities. The spaces in the mask are then used to extract the trabecular region.  60  Chapter 4 – Methods  Figure 15. Summary of the filters and parameters implemented in the dual threshold algorithm for analysis of a 3D dataset. Figure from Buie HR, et al. Bone. 2007;41(4):505-515, [63] with permission.  61  Chapter 4 – Methods Together, the two thresholding techniques are combined to create a mask of the trabecular, cortical and non-bone regions [63]. This process occurs for each of the 104 slices in the HRpQCT image. Outcome variables included cortical porosity (Ct.Po, %), cortical thickness (Ct.Th, mm), cortical density (Ct.BMD, mg HA/cm3) and total area (Tt.Ar, mm2). Cortical porosity is calculated as the number of void voxels within the cortex and Ct.Th using a distance transform [173]. Cortical BMD is determined by converting attenuation data into hydroxyapatite densities using the cortical masks generated from the automatic segmentation method. Tt.Ar is determined based on the automatic segmentation algorithms as the average area within the periosteal surface of the bone segment [67].  Finite Element Analysis To estimate bone strength, finite element (FE) analysis was applied to HR-pQCT scans. FE meshes were generated from the three-dimensional HR-pQCT images using the voxel conversion approach [174,175]. To simulate uniaxial compression on each radius and tibia, we used a Young’s modulus of 689 M a and a oisson’s ratio of 0. [172]. We used a customized FE solver (FAIM, Version 4.0, Numerics88 Solutions, Calgary, Canada) on a desktop workstation (MacPro, OSX, Version 10.5.6, Apple Inc., Cupertino, CA, USA; 2 × 2.8 GHz Quad-Core Intel Xenon) to solve the models [67]. For the radius, the meshes generated approximately 1 million elements compared with 2.5 million elements for the tibia [176]. Outcome variables included failure load (N) and ultimate stress (MPa). Additionally, we estimated the risk of forearm fracture using the load-to-strength ratio (Ф) [177] as the ratio of the estimated load applied to the outstretched hand during a fall from standing height [178] to failure load derived from FE analysis. This is based on participant-specific fall force estimated from a forward dynamics model that incorporates participant height [178,179].  6 0 Fall Load (N) =  9.81  ht (Equation 6),  10  670 is the damping coefficient (Ns/m), ht = participant height (cm) and 9.81 = gravitational constant (m/s2). Load-to-strength ratio (Ф) =  Fall Load ( ) Failure Load ( )  (Equation 7).  62  Chapter 4 – Methods In theory, the biomechanical fracture threshold occurs at Ф = 1, thus values greater than 1 indicate increased fracture risk and values below 1 indicate lowered fracture risk [177].  4.4  Statistical Analysis All statistical analyses were completed using StataCorp (Stata Statistical Software,  version 10.1, College Station, TX). Primary outcomes for my study are the following bone strength variables: FLoad, UStress and load-to-strength ratio (radius only). Secondary outcomes are bone geometry (Tt.Ar), bone density (Tt.BMD, Ct.BMD) and microstructure (Ct.Po, Ct.Th, BV/TV, Tb.N, Tb.Th and Tb.Sp) variables. The primary explanatory variable was FM and secondary explanatory variable was LM. Control variables included age (age for boys, centered age and centered age squared for girls), ethnicity (Caucasian vs. Asian), menarche status (pre- vs. post-menarche for Tt.BMD, Ct.BMD and Ct.Po only in girls) and limb length (ulna or tibia).  4.4.1 Inclusion and Exclusion Criteria For analysis I included participants who had HR-pQCT radius and/or tibia scans in 2009. I excluded participants if their total body DXA, maturity, and/or ethnicity data were missing. I also e cluded participants with ethnicity classified as ‘Other’ or ‘Mi ed’ since few participants were in these ethnic categories and due to the variability in ethnicity represented by participants of mixed ethnicity. As there are known bone differences between Asians and Caucasians, I entered ethnicity (Caucasian compared with Asian) into regression models [89,180]. I also excluded participants who reported using oral corticosteroids for three months or longer in the previous year and participants with medical conditions known to affect bone metabolism, such as Type I diabetes [181,182]. Finally, I excluded participants whose HR-pQCT scans were given an ‘unacceptable’ motion grade (4 or 5) at either measured site. I analyzed boys and girls separately due to known sex differences in bone structure [99] and in the tempo and timing of growth and maturation [183].  4.4.2 Data Exploration To visually represent and inspect the distribution of descriptive variables I used histograms for continuous variables and dotplots for categorical variables. I also used scatter plots (for continuous variables) and box plots (for categorical variables) to visually assess relationships between the descriptive variables and each bone variable. I created scatter plots for 63  Chapter 4 – Methods fat mass, lean mass, age and limb length versus all bone variables. I used a smoother function (lowess) to visually assess functional form. I also created box plots to compare each bone variable between Asian and Caucasian participants. After exploratory data analysis I concluded that the relationship between adiposity and bone quality was similar in the younger and older cohorts; therefore, I analyzed all participants in the same multivariable linear regression models.  4.4.3 Descriptive Analyses I report the mean, standard deviation, minimum and maximum for age (yrs), height (cm), weight (kg), BMI (kg/m2), ulna length (mm), tibia length (mm), fat mass (kg), lean mass (kg), body fat (%), age at menarche (yrs) and bone variables. For categorical variables I calculated the proportion of participants in each group (i.e. maturity, ethnicity). Bone variables (dependent variables) included FLoad (N), UStress (MPa), load-to-strength ratio (radius only), Tt.Ar (mm2), Tt.BMD (mg HA/cm3), Ct.BMD (mg HA/cm3), Ct.Po (%), Ct.Th (mm), BV/TV, Tb.N (1/mm), Tb.Th (mm) and Tb.Sp (mm).  4.4.4 Regression Analyses I fit hierarchical multivariable linear regression models to determine the independent effect of predictor variables FM and LM on all dependent bone outcomes and entered variables in three blocks. For boys, variables included in the first block were age, limb length (ulna or tibia) and ethnicity (Asian vs. Caucasian). The second and third blocks were LM and FM, respectively. For girls, variables included in the first block were centered age (Age – 15), centered age squared, limb length (ulna or tibia) and ethnicity (Asian vs. Caucasian). The second and third blocks were LM and FM, respectively. For girls, menarcheal status was added with the first block of covariates for Tt.BMD, Ct.BMD, and Ct.Po. In order to determine the independent influence of FM on bone variables before adjusting for LM, I also created hierarchical regression models without LM. Model 1 evaluates the influence of FM on bone outcomes after adjusting for age (or centered age and centered age squared for girls), limb length and ethnicity. Model 2 evaluates the influence of FM on bone outcomes after adjusting for age (or centered age and centered age squared for girls), limb length, ethnicity and LM. Model 3 evaluates the invluence of LM on bone outcomes after adjusting for age (or centered age and centered age squared for girls), limb length, ethnicity and FM. For girls, menarche status was added to all models for Tt.BMD, Ct.BMD and Ct.Po. 64  Chapter 4 – Methods I adjusted for age to account for age-related gains in bone quality [5]. Ideally, I would have used a measure of maturity rather than age since maturity is more closely related to gains in bone mass across the growing years; however, I did not have an even distribution of boys in all 5 Tanner stages. For girls, I adjusted for both centered age and centered age squared because I observed a plateau effect in exploratory scatter plots of age vs. bone variables. As in previous studies [61,99], I included limb length as an estimate of moment arm. I included ethnicity in the models due to known ethnic differences in bone outcomes. Specifically Asian children have lower bone mineral content compared with Caucasian children [80,89,180]. For girls, I added menarcheal status into the bone density (Tt.BMD, Ct.BMD) and Ct.Po models based on the well known gains in bone mineral density that coincide with the onset of menarche [184]. I added menarcheal status into models for all other bone variables in girls; however, this did not change the inference so I do not report these outcomes in this thesis. I added lean mass as a surrogate of muscle force so as to interpret results in the context of the functional model of bone development [7]. Finally, I added fat mass to assess its independent relationship with bone quality. I used QQ plots, residual vs. fit plots and residual vs. covariate plots to check the assumptions of regression models. To assess the functional form of regression models I plotted the residuals of two regression models; model 1. bone variable adjusted for all model covariates except FM, second; model 2. FM adjusted for all model covariates except bone variables. I fit a lowess smoother and a linear line to the scatter plots of these residuals to visually inspect the functional form of the overall regression models.  65  Chapter 5 – Results  Chapter 5 – Results In this chapter I describe the participants included in my analysis. I also present descriptive statistics for all measured outcomes, univariate relationships between descriptive outcomes (FM, LM, height, weight, age, limb length, ethnicity) and bone variables (FLoad, UStress, load-to-strength ratio, Tt.Ar, Tt.BMD, Ct.BMD, Ct.Po, Ct.Th, BV/TV, Tb.N, Tb.Th and Tb.Sp) and results of hierarchical multivariable regression analysis.  5.1 Cohort Characteristics 5.1.1 Participants I provide a flowchart outlining participant exclusions (Figure 16). Briefly, 340 participants (159 boys, 181 girls) visited the lab for measurement in 2009. Twenty-two boys and 24 girls were excluded from analysis for the following reasons. First, I excluded 32 participants (14 boys, 18 girls) of ‘Other’ or ‘Mi ed’ ethnicities. Second, I e cluded 10 participants (6 boys, 4 girls) who did not have whole body DXA scans due to contraindications including chest X-rays in the previous 6 months (n = 4), multiple head and neck X-rays (n = 1) or other undefined Xrays (n = 4). Third, I excluded one boy because his head was outside of the DXA scanning area. Fourth, I excluded one boy and one girl who had medical conditions known to affect bone metabolism (the boy had childhood leukemia; the girl had Type I diabetes) and one girl with Crohn’s disease who reported taking oral corticosteroids for longer than 3 months. Finally, I excluded one boy because he chose not to complete the self-assessment of maturity status (Tanner staging). Thus, the present analysis includes 137 boys and 157 girls. After these initial exclusions, I also excluded some participant’s HR-pQCT scans due to missing consent, motion artefacts or measurement error. In all cases, scan issues affected only one site of the two HR-pQCT sites measured. Therefore, I did not exclude these participants from my overall analysis. Four boys did not have consent for the radius scan and I excluded four radius scans due to a motion grade of 4 or 5. Three girls did not have consent for the radius scan and I excluded eight radius scans due to a motion grade of 4 or 5 and one tibia scan due to improper positioning. Thus, the present analysis includes 129 radius scans and 137 tibia scans for boys and 146 radius scans and 156 tibia scans for girls.  66  Chapter 5 – Results  Total number of participants measured in spring 2009  Boys  n = 159  Exclusions and missing data: ‘Other’ or ‘mi ed’ n = 14 ethnicity No DXA scans n=5 Childhood leukemia n=1 Excluded DXA scans n=1 No maturity rating n=1 n = 22 Total Excluded  Boys included for analysis  n = 137  Exclusions for radius scans: No consent n=4 Motion artifacts n=4 n=8 Total Excluded  n = 340  Girls  n = 181  Exclusions and missing data: ‘Other’ or ‘mi ed’ n = 18 ethnicity No DXA scans n=4 Corticosteroid usage n=1 Type I Diabetes n=1 Total Excluded  Girls included for analysis  n = 24  n = 157  Exclusions for radius scans: No consent n=3 Motion artifacts n=8 n = 11 Total Excluded  Exclusions for tibia scans: Measurement error n=1 n=1 Total Excluded  Included for radius analysis Included for tibia analysis  n = 129 n = 137  Included for radius analysis Included for tibia analysis  n = 146 n = 156  Figure 16. Flow chart that describes participant exclusions from data analysis.  67  Chapter 5 – Results 5.1.2 Descriptive Characteristics of the Sample Participant descriptive characteristics including age, height, weight, BMI, maturity, ethnicity and body composition are provided (Table 5). Although a comparison between sexes was not an aim of my thesis, I comment briefly on differences in boys’ and girls’ descriptive characteristics. As expected, boys were taller and weighed more than girls. In addition, boys had lower percent body fat and higher values for lean mass, compared with girls. Descriptive characteristics for all bone variables are provided (Table 6). Boys had higher values for FLoad, UStress, Tt.Ar, Tt.BMD, Ct.Po, Ct.Th, BV/TV, Tb.N, and Tb.Th and lower load-to-strength ratio, Ct.BMD, and Tb.Sp, compared with girls. As it was not a focus of my thesis, I did not perform any statistical tests to compare bone variables between boys and girls.  Table 5. Demographic, maturity and anthropometric outcomes for boys and girls. Values are mean (SD), minimum and maximum unless otherwise indicated. Boys  Girls  Mean (SD)  Min.  Max.  Mean (SD)  Min.  Max.  n  137  ---  ---  157  ---  ---  Age (yrs)  15.6 (3.3)  9.5  20.8  14.5 (3.9)  9.5  21.1  # Asian/Caucasian  67/70  ---  ---  82/75  ---  ---  Tanner stage (# 1/2/3/4/5)  16/14/5/39/63  ---  ---  29/33/25/28/42  ---  ---  Menarche (# pre/post)  ---  ---  ---  68/89  ---  ---  Age at menarche  ---  ---  ---  12.4 (1.3)  9.9  15.7  Height (cm)  167.0 (15.6)  129.7  196.4  154.0 (11.8)  130.7  181.6  Weight (kg)  60.4 (15.6)  27.8  133.0  48.9 (15.0)  22.2  101.2  BMI (kg/m )  21.2 (4.2)  14.8  42.6  20.2 (4.1)  12.6  38.2  Ulna length (mm)  271 (29)  204  322  244 (20)  196  290  Tibia length (mm)  404 (40)  306  485  367 (30)  300  444  Fat mass (kg)  10.9 (7.1)  3.6  56.7  13.4 (7.0)  2.8  41.3  Lean mass (kg)  45.7 (13.4)  20.8  74.2  32.8 (8.2)  17.9  57.0  % Body fat  18.4 (8.2)  7.8  43.9  26.8 (6.5)  12.5  45.8  2  SD = standard deviation.  68  Chapter 5 – Results Table 6. Bone outcomes (assessed by HR-pQCT) for boys and girls at the distal radius and distal tibia. Values are mean (SD), minimum and maximum. Boys Mean (SD)  Min.  Girls Max.  Mean (SD)  Min.  Max.  Distal Radius n  129  FLoad (N)  2265 (872)  629  4576  1537 (558)  477  2711  UStress (MPa)  34.4 (13.2)  4.8  66.5  28.9 (11.4)  6.3  61.3  Fall Load (N)  2711 (131)  2390  2941  2606 (100)  2399  2829  Load-to-strength ratio  1.42 (0.64)  0.63  3.80  1.95 (0.75)  0.96  5.11  Tt.Ar (mm2)  261.3 (59.6)  139.0  428.3  202.9 (36.7)  126.2  321.4  318.8 (79.6)  155.7  524.6  298.0 (76.2)  167.1  498.9  Ct.BMD (mg HA/cm )  743.3 (98.2)  544.0  937.5  753.3 (122.6)  540.1  974.2  Ct.Po (%)  4.0 (2.3)  0.6  14.2  2.9 (2.2)  0.1  10.5  Ct.Th (mm)  1.00 (0.31)  0.43  1.68  0.86 (0.26)  0.44  1.44  BV/TV  0.160 (0.031)  0.082  0.246  0.141 (0.026)  0.075  0.237  Tb.N (1/mm)  2.00 (0.26)  1.28  2.74  1.96 (0.26)  1.37  2.48  Tb.Th (mm)  0.080 (0.015)  0.050  0.136  0.072 (0.010)  0.053  0.120  Tb.Sp (mm)  0.428 (0.070)  0.281  0.689  0.448 (0.074)  0.315  0.673  Tt.BMD (mg HA/cm3) 3  146  Distal Tibia n  137  FLoad (N)  6324 (1853)  2186  10720  4796 (1342)  2434  7977  35.1 (10.2)  10.9  58.9  31.1 (9.8)  13.3  59.6  742.1 (132.8)  450.3  1052.7  619.5 (83.9)  421.2  904.7  Tt.BMD (mg HA/cm )  301.1 (61.0)  167.8  448.8  278.3 (61.2)  165.1  467.9  Ct.BMD (mg HA/cm3)  779.1 (84.7)  605.5  904.9  781.6 (112.8)  601.9  959.6  Ct.Po (%)  5.8 (2.2)  1.8  13.4  4.5 (2.3)  1.2  10.9  Ct.Th (mm)  1.23 (0.36)  0.58  2.30  1.02 (0.30)  0.45  1.86  BV/TV  0.168 (0.026)  0.112  0.234  0.154 (0.025)  0.099  0.219  Tb.N (1/mm)  1.93 (0.29)  1.16  2.63  1.84 (0.25)  1.33  2.55  Tb.Th (mm)  0.088 (0.014)  0.055  0.119  0.085 (0.014)  0.056  0.131  Tb.Sp (mm)  0.442 (0.077)  0.291  0.760  0.489 (0.072)  0.306  0.660  UStress (MPa) 2  Tt.Ar (mm ) 3  156  FLoad = failure load; UStress = ultimate stress; Tt.Ar = total area; Tt.BMD = total bone mineral density; Ct.BMD = cortical bone mineral density; Ct.Po = cortical porosity; Ct.Th = cortical thickness; BV/TV = trabecular bone volume fraction; Tb.N = trabecular number; Tb.Th = trabecular thickness; Tb.Sp = trabecular separation; SD = standard deviation.  69  Chapter 5 – Results 5.2 Relationship Between Body Fat and Bone Quality 5.2.1 Unadjusted I report univariate regression model R values for fat mass, lean mass, height, weight, limb length (ulna or tibia), ethnicity and age with all bone variables in Appendix C (Table 10 -boys and Table 11-girls). Here I focus on the univariate relationships between FM and bone outcomes and LM and bone outcomes. Scatter plots of FM and LM with a selection of bone variables (FLoad (bone strength), Tt.Ar (bone geometry), Tt.BMD (bone density), BV/TV (bone microstructure)) for boys and girls are presented in Figure 17 and Figure 18, respectively. For boys, FM was weakly but positively associated with Tt.BMD, Ct.BMD and Ct.Th (R = 0.18 - 0.21, p < 0.05 for all) at the distal radius and to FLoad, Tt.Ar and Tb.N (R = 0.19 - 0.29, p < 0.05 for all) at the distal tibia. In addition, FM was negatively associated with Tb.Sp at the distal tibia in boys (R = -0.21, p < 0.05). For girls, FM was positively associated with a number of bone outcomes including FLoad, UStress, Tt.Ar, Tt.BMD and Ct.Th at both the distal radius and distal tibia (R = 0.35 - 0.64, p < 0.05 for all). Conversely, FM was negatively associated with girls’ Ct. o at the distal radius and distal tibia (R = -0.35 to -0.46, p < 0.01). At the radius in girls I also noted negative correlations between FM and load-to-strength radio and Tb.N (R = -0.20 to -0.49, p < 0.05) and a positive correlation between FM and Tb.Sp (R = 0.19, p < 0.05). Finally, at the distal tibia only, FM correlated positively with BV/TV in girls (R = 0.24, p < 0.01). I observed similar relationships between bone outcomes and LM. Importantly the strength of these associations was stronger (more statistically significant) than for FM. For boys and girls at the distal radius and distal tibia, LM correlated positively with FLoad (R = 0.79 - 0.88, p < 0.01), UStress (R = 0.52 - 0.69, p < 0.01), Tt.Ar (R = 0.53 - 0.75, p < 0.01), Tt.BMD (0.55 - 0.69, p < 0.01), Ct.BMD (R = 0.65 - 0.76, p < 0.01), Ct.Th (R = 0.66 - 0.77, p < 0.01) and Tb.Th (R = 0.19 - 0.48, p < 0.05). Thus, it appears that LM correlated most strongly with FLoad and most weakly with Tb.Th at both sites in boys and girls. Furthermore, LM correlated negatively with Ct.Po (R = -0.21 to -0.56, p < 0.05 for all) at both the radius and tibia and with load-to-strength ratio and Tb.N (-0.23 to -0.81, p < 0.05 for all) at the radius only, in both boys and girls. LM was positively related to BV/TV at both sites for boys, but only at the distal tibia for girls (R = 0.34 – 0.38, p < 0.01). Additionally, LM positively correlated with Tb.Sp at the distal radius in girls (R = 0.23, p < 0.01). In all cases, the correlations between LM and bone outcomes were noticeably stronger than the relationships between FM and bone outcomes.  70  Chapter 5 – Results 5000  5000 R = 0.05  FLoad (N)  4000 3000  3000  2000  2000  1000  1000  0  0 0  20  40  60  R = -0.07  400  Tt.Ar (mm2)  0  40  60  80  40  60  80  40  60  80  40 60 Lean Mass (kg)  80  R = 0.67  400  300  300  200  200  100  100  0  0 0  20  40  60  600  0  20  600 R = 0.21  500  R = 0.62  500  400  400  300  300  200  200  100  100  0  0 0  20  40  60  0.3  0  20  0.3 R = 0.34  R = 0.04 BV/TV  20  500  500  Tt.BMD (mg HA/cm3)  R = 0.84  4000  0.2  0.2  0.1  0.1  0  0 0  20 40 Fat Mass (kg)  60  0  20  Figure 17. Scatter plots of fat mass and lean mass vs. failure load (FLoad), total area (Tt.Ar), total density (Tt.BMD) and trabecular bone volume fraction (BV/TV) at the radius for BOYS. 71  Fload (N)  Chapter 5 – Results 3000  3000  2500  2500  2000  2000  1500  1500  1000  1000  R = 0.54  500  500  0  0  Tt.Ar (mm2)  0  20  40  60 350  300  300  250  250  200  200  150  150  100  R = 0.49  0  20  40  60  20  40  60  20  40  60  R = 0.75  100 50 0  0  Tt.BMD (mg HA/cm3)  0  350  50 20  40  60  0  600  600  500  500  400  400  300  300  200  R = 0.55  200 R = 0.40  100  100  0  0 0  BV/TV  R = 0.79  20  40  60  0  0.3  0.3  0.2  0.2  0.1  0.1  R = -0.02  R = -0.10 0  0 0  20 40 Fat Mass (kg)  60  0  20 40 Lean Mass (kg)  60  Figure 18. Scatter plots of fat mass and lean mass vs. failure load (FLoad), total area (Tt.Ar), total density (Tt.BMD) and trabecular bone volume fraction (BV/TV) at the radius for GIRLS. 72  Chapter 5 – Results 5.2.2 Regression Models I provide unstandardized multivariable regression coefficients in Appendix C (Table 12; boys radius, Table 13; boys tibia, Table 14; girls radius and Table 15; girls tibia). I provide R2 values from hierarchical regression models and change in R2 for boys and girls in Table 7 (Model 1: FM, adjusted for age, limb length, ethnicity, not LM), Table 8 (Model 2: FM, adjusted for age, limb length, ethnicity, LM) and Table 9 (Model 3: LM, adjusted for age, limb length, ethnicity, FM).  5.2.2.1  Bone Strength For boys, without LM in the regression models (Model 1), FM positively predicted FLoad  at the distal tibia accounting for 1% of the variance. However, after LM was added to the model (Model 2), FM was no longer a significant predictor of FLoad at the distal tibia. At the distal radius in boys, before LM was added to the model (Model 1), FM did not significantly predict FLoad. However, in Model 2, once LM was added to the model, FM negatively predicted FLoad at the radius, and explained 3% of the total variance in FLoad. When all covariates are held equal, Model 2 predicted that, in boys, for every 1 kg increase in fat mass, we would expect a 23 N (0.9%) decrease in FLoad. Prior to the addition of LM to the models, FM did not significantly predict load-to-strength ratio at the distal radius in boys (Model 1). However, once LM was added in Model 2, FM weakly but positively predicted load-to-strength ratio at the distal radius, accounting for 2% of the variance explained by the model. Thus, a boy with 1 kg increase in fat mass but identical values for all covariates in Model 2 would experience a 0.01 (0.7%) increase in load-to-strength ratio and thus forearm fracture risk, on average. FM was not a significant predictor of UStress at the radius or tibia in either Model 1 or Model 2 in boys. In Model 3, LM positively predicted FLoad and UStress at both the radius and the tibia in boys, accounting for 415% of the variance in bone strength. In addition, LM was a negative predictor of load-tostrength ratio at the distal radius, accounting for 4% of the variance. For girls in Model 1, before adjusting for LM, FM positively predicted FLoad at the distal tibia and accounted for 5% of the variance. FM did not predict any other strength variable at either the radius or the tibia. Once LM was added to Model 2, FM was not a significant predictor of any bone strength variable at the distal radius or the distal tibia. FM did not explain any variance in any bone variable at the distal radius even before adjusting for LM (Model 1).  73  Chapter 5 – Results In girls for Model 3, LM positively predicted FLoad at both sites accounting for 2% of the variance at the distal radius and 15% of the variance at the distal tibia. This suggests that with a 1 kg increase in LM and with all other covariates held equal, FLoad increased 26 N (1.7%) and 148 N (3.1%) at the radius and tibia, respectively. Further, LM was a positive predictor of UStress at the distal tibia accounting for 3% of the total variance. LM was a negative predictor of load-tostrength ratio at the distal radius accounting for 2% of the total variance.  5.2.2.2  Bone Geometry For boys, similar to the models for bone strength, in Model 1 (no LM) FM positively  predicted Tt.Ar at the distal radius and distal tibia accounting for 2% and 4% of the variance, respectively. Conversely, once LM was adjusted for in Model 2, FM negatively predicted Tt.Ar at the distal radius accounting for 7% of the variance. However, FM was not a predictor of Tt.Ar at the distal tibia in Model 2. With all other covariates equal, a 1 kg increase in FM predicted 2.4 mm2 (0.9%) less Tt.Ar at the distal radius for boys. Conversely, LM (Model 3) positively predicted Tt.Ar at both the distal radius and distal tibia, accounting for 3% and 12% of the variance, respectively. This reflects 3.1 mm2 (1.2%) and 7.0 mm2 (0.9%) increases in total area at the distal radius and distal tibia, respectively with a 1 kg increase LM with all other covariates equal. For girls, in Model 1 before LM was adjusted for, FM positively predicted Tt.Ar at the distal tibia but not the distal radius, which accounted for 10% of the variance. However, after adjusting for LM in Model 2, FM negatively predicted Tt.Ar at the distal radius and accounted for 1% of the variance. All other variables being equal for girls, a 1 kg increase in FM predicted 0.9 mm2 (0.4%) smaller total area. FM did not predict Tt.Ar at the distal tibia (Model 2). Conversely, in Model 3 LM positively predicted Tt.Ar at both the distal radius and distal tibia, and accounted for 8% and 20% of the variance, respectively. If all covariates were equal, this would predict a 3.7 mm2 (1.8%) and 9.2 mm2 (1.5%) increase in Tt.Ar at the distal radius and distal tibia, respectively, with a 1 kg increase in LM.  5.2.2.3  Bone Density For boys, at the distal radius but not the distal tibia, in Model 1 (no LM) FM positively  predicted Tt.BMD and Ct.BMD (4% and 3% of the variance, respectively). After adjusting for LM in Model 2, FM was not a significant predictor of Tt.BMD or Ct.BMD at the radius or tibia 74  Chapter 5 – Results (Table 12 and Table 13, respectively Appendix C). In contrast, Model 3 LM positively predicted Tt.BMD and Ct.BMD at both the radius and tibia and accounted for 7-9% of the variance in Tt.BMD and 1-3% of the variance in Ct.BMD. For bone density outcomes in girls (Tt.BMD and Ct.BMD), I added menarcheal status to the regression models as a covariate to account for the rapid rise of estrogen at the onset of menarche. Estrogen stimulates increases in bone density for reproductive needs [184]. In Model 1, prior to adjusting for LM, FM did not significantly predict any bone density outcome at either the radius or the tibia in girls. After adjusting for LM in Model 2, FM did not predict girls’ Tt.BMD at either the distal radius or the distal tibia. However, in Model 2 FM positively predicted Ct.BMD at the distal radius and accounted for 0.3% of the variance. With all other variables equal in girls, a 1 kg increase in FM predicted a 1.6 mg HA/cm3 (0.2%) increase in Ct.BMD at the distal radius. FM did not significantly predict Ct.BMD at the distal tibia. Furthermore, in Model 3 LM positively predicted Tt.BMD at the distal tibia, and accounted for 2% of the total variance explained by the model, but did not explain any other bone density variable at either site.  5.2.2.4  Bone Microstructure  Cortical Bone Microstructure For boys in Model 1, before LM was added to the regression models, FM positively predicted Ct.Th at the distal radius; however this relationship was no longer significant once LM was adjusted for in Model 2. In Model 1, FM did not significantly predict Ct.Th at the distal tibia in boys; however, in Model 2 FM was a weak negative predictor of Ct.Th at the distal tibia, and accounted for only 1% of the variance. If all model covariates were equal, this predicts a 0.01 mm (0.8%) decrease in Ct.Th with a 1 kg increase FM for boys. LM (Model 3) positively predicted Ct.Th at both the distal radius and the distal tibia in boys and accounted for 9% and 7% of the variance, respectively. Before adjusting for LM (Model 1), FM negatively predicted Ct.Po at the distal radius but not the distal tibia and accounted for 2% of the variance at the distal radius in boys. After adjusting for LM in Model 2, FM did not significantly predict Ct.Po at the distal radius or tibia. LM (Model 3) did not predict Ct.Po at either site in boys. In girls, before LM was added to the model, (Model 1) FM positively predicted Ct.Th at the distal tibia and explained 1% of the variance, but did not predict Ct.Th at the distal tibia. Once 75  Chapter 5 – Results LM was adjusted for, FM did not significantly predict Ct.Th at either site; however, LM positively predicted Ct.Th at the distal tibia and explained 4% of the variance, but did not significantly predict Ct.Th at the distal radius. Similar to bone density outcomes in girls, I added menarcheal status to the regression models as a covariate for Ct.Po to account for the rapid rise of estrogen at the onset of menarche. In Model 1, prior to adjusting for LM, FM did not significantly predict Ct.Po at either the distal radius or distal tibia in girls. However, in Model 2, after adjusting for LM, FM negatively predicted Ct.Po at the distal radius and accounted for 2% of the variance. With all other variables equal in girls, a 1 kg increase in FM predicted a 0.1% (3.5%) decrease in Ct.Po at the distal radius. In addition, in Model 2 FM negatively predicted Ct.Po at the distal tibia and accounted for 3% of the variance. With a 1 kg increase in FM we would observe a 0.1% (2.2%) decrease in Ct.Po at the distal tibia. In Model 3, LM positively predicted Ct.Po at the distal radius and distal tibia in girls, and accounted for 0.2% and 4% of the variance, respectively. With all covariates held equal, this model predicted that a 1 kg increase in LM would be associated with 0.1% (3.5%) and 0.2% (0.9%) greater Ct.Po at the distal radius and distal tibia, respectively.  Trabecular Bone Microstructure For boys in Model 1, before LM was added to the regression models, FM positively predicted Tb.N but negatively predicted Tb.Th and Tb.Sp at the distal tibia and did not predict any trabecular bone microstructure variable at the distal radius. However, once LM was adjusted for (Model 2), FM did not significantly predict any trabecular bone microstructure variable (BV/TV, Tb.N, Tb.Th or Tb.Sp) at the tibia. Conversely, in Model 3 LM positively predicted BV/TV, Tb.N and Tb.Th and negatively predicted Tb.Sp at the distal radius, accounting for 13%, 8%, 3% and 11% of the variance, respectively. Furthermore, LM positively predicted BV/TV and Tb.N and negatively predicted Tb.Sp at the distal radius, accounting for 12%, 19% and 24% of the variance, respectively. For girls, before LM was added to the models, in Model 1 FM positively predicted Tb.N at the distal tibia in girls and explained 5% of the variance. FM did not significantly predict any bone microstructure variable at the distal radius in either Model 1 or Model 2. At the distal tibia, after adjusting for LM in Model 2, FM no longer predicted Tb.N and was a weak negative predictor of Tb.Th and accounted for 3% of the variance. With all covariates in the model equal, we would observe that a 1 kg increase in FM would predict 0.001 mm (1.2%) thinner trabeculae. 76  Chapter 5 – Results In Model 3, LM did not predict any bone microstructure outcomes at the distal radius in girls. Conversely, LM positively predicted BV/TV and Tb.Th at the distal tibia and accounted for 11% and 2% of the variance, respectively.  77  Chapter 5 – Results  Table 7. Model 1: Hierarchical multivariable linear regression models to demonstrate the independent contribution of fat mass to bone variables at the distal radius and distal tibia in boys (controlled for age, limb length and ethnicity) and girls (controlled for centered age, centered age squared, limb length and ethnicity: menarcheal status added to models for Tt.BMD, Ct.BMD, Ct.Po).  Boys  Radius  FLoad  UStress  Load:Str  Tt.Ar  Tt.BMD  Ct.BMD  Ct.Po  Ct.Th  BV/TV Tb.N  Tb.Th  Tb.Sp  R2 FM in model 0.65  0.44  0.69  0.49  0.48  0.69  0.36  0.58  0.07  0.23  0.28  0.14  R2 no FM in model R2 difference  0.65  0.43  0.69  0.47  0.45  0.66  0.34  0.55  0.07  0.21  0.28  0.12  0.00  0.01  0.00  0.02 *  0.04 **  0.03 **  0.02 *  0.03 ** 0.00  0.02  0.00  0.02  R FM in model 0.65  0.48  ---  0.39  0.44  0.75  0.19  0.52  0.15  0.09  0.30  0.07  R2 no FM in  0.65  0.47  ---  0.35  0.44  0.74  0.19  0.52  0.14  0.04  0.28  0.02  0.01 *  0.00  ---  0.04 **  0.00  0.00  0.00  0.00  0.00  0.05 ** 0.02 *  0.05 *  0.51  0.67  0.52  0.60  0.90  0.68  0.72  0.02  0.21  0.11  0.18  0.74  0.51  0.67  0.52  0.60  0.90  0.68  0.72  0.01  0.21  0.10  0.18  0.00  0.00  0.00  0.01  0.00  0.00  0.01  0.01  0.01  0.00  0.01  0.00  R FM in model 0.67  0.63  ---  0.36  0.66  0.93  0.51  0.71  0.16  0.23  0.36  0.19  0.62  0.63  ---  0.27  0.66  0.93  0.51  0.70  0.15  0.19  0.36  0.15 **  0.05 **  0.00  ---  0.10 **  0.00  0.00  0.00  0.01 *  0.01  0.05 ** 0.00  Tibia  2  model R2 difference 2  Radius  R FM in model 0.74 2  R no FM in model  Girls  R2 difference  Tibia  2 2  R no FM in model R2 difference  0.04  * p < 0.05; ** p < 0.01 FLoad = failure load; UStress = ultimate stress; Load:Str = load-to-strength ratio; Tt.Ar = total area; Tt.BMD = total bone mineral density; Ct.BMD = cortical bone mineral density; Ct.Po = cortical porosity; Ct.Th = cortical thickness; BV/TV = trabecular bone volume fraction; Tb.N = trabecular number; Tb.Th = trabecular thickness; Tb.Sp = trabecular separation; FM = total body fat mass by DXA.  78  Chapter 5 – Results  Table 8. Model 2: Hierarchical multivariable linear regression models to demonstrate the independent contribution of fat mass to bone variables prior to adjusting for lean mass at the distal radius and distal tibia in boys (controlled for age, limb length, ethnicity and lean mass) and girls (controlled for centered age, centered age squared, limb length, ethnicity and lean mass: menarcheal status added to models for Tt.BMD, Ct.BMD, Ct.Po). FLoad  UStress  Load:Str  Tt.Ar  Tt.BMD  Ct.BMD  Ct.Po  Ct.Th  BV/TV  Tb.N  Tb.Th  Tb.Sp  0.50  0.75  0.56  0.54  0.70  0.36  0.64  0.22  0.28  0.33  0.24  0.75  0.50  0.73  0.50  0.54  0.70  0.34  0.64  0.20  0.28  0.31  0.24  0.03 **  0.00  0.02 **  0.07 **  0.01  0.01  0.02  0.00  0.02  0.00  0.02  0.00  0.52  ---  0.48  0.52  0.76  0.20  0.60  0.27  0.22  0.30  0.26  0.79  0.51  ---  0.47  0.51  0.76  0.20  0.59  0.26  0.22  0.28  0.26  0.00  0.01  ---  0.00  0.01  0.00  0.00  0.01 *  0.01  0.00  0.02  0.00  R2 FM in model 0.76  0.51  0.69  0.61  0.60  0.90  0.70  0.72  0.03  0.22  0.11  0.18  R2 no FM in  0.76  0.51  0.69  0.60  0.60  0.90  0.68  0.72  0.01  0.21  0.10  0.18  0.00  0.00  0.00  0.01 *  0.00  0.003*  0.02 **  0.01  0.01  0.00  0.01  0.00  R2 FM in model 0.78  0.67  ---  0.47  0.68  0.93  0.59  0.74  0.28  0.24  0.42  0.22  R2 no FM in  0.77  0.66  ---  0.47  0.68  0.93  0.56  0.74  0.26  0.23  0.39  0.21  0.00  0.01  ---  0.00  0.00  0.00  0.03 **  0.00  0.02  0.01  0.03 **  0.00  Boys  Radius  R2 FM in model 0.77 2  R no FM in model R2 difference  Tibia  R2 FM in model 0.79 2  R no FM in model  Radius  R2 difference  model  Tibia  Girls  R2 difference  model R2 difference  * p < 0.05; ** p < 0.01 FLoad = failure load; UStress = ultimate stress; Load:Str = load-to-strength ratio; Tt.Ar = total area; Tt.BMD = total bone mineral density; Ct.BMD = cortical bone mineral density; Ct.Po = cortical porosity; Ct.Th = cortical thickness; BV/TV = trabecular bone volume fraction; Tb.N = trabecular number; Tb.Th = trabecular thickness; Tb.Sp = trabecular separation; FM = total body fat mass by DXA.  79  Chapter 5 – Results  Table 9. Model 3: Hierarchical multivariable linear regression models to demonstrate the independent contribution of lean mass to bone variables at the distal radius and distal tibia in boys (controlled for age, limb length, ethnicity and fat mass) and girls (controlled for centered age, centered age squared, limb length, ethnicity and fat mass: menarcheal status added to models for Tt.BMD, Ct.BMD, Ct.Po).  Radius  FLoad  UStress  Load:Str Tt.Ar  Tt.BMD  Ct.BMD  Ct.Po  Ct.Th  BV/TV  Tb.N  Tb.Th  Tb.Sp  R2 LM in model 0.75  0.50  0.73  0.50  0.54  0.70  0.34  0.64  0.20  0.28  0.31  0.24  R2 no LM in  0.65  0.43  0.69  0.47  0.45  0.66  0.34  0.55  0.07  0.21  0.28  0.12  0.10 **  0.07 **  0.04 **  0.03 **  0.09 **  0.03 **  0.00  0.09 **  0.13 **  0.08**  0.03 *  0.11 **  0.51  ---  0.47  0.51  0.76  0.20  0.59  0.26  0.22  0.28  0.26  0.64  0.47  ---  0.35  0.44  0.74  0.19  0.52  0.14  0.04  0.28  0.02  0.15 **  0.04 **  ---  0.12 **  0.07 **  0.01 *  0.00  0.07 **  0.12 **  0.19 **  0.01  0.24 **  R2 LM in model 0.76  0.51  0.69  0.60  0.60  0.90  0.68  0.72  0.01  0.21  0.10  0.18  R2 no LM in  0.74  0.51  0.67  0.52  0.60  0.90  0.68  0.72  0.01  0.21  0.10  0.18  0.02 **  0.00  0.02 **  0.08 **  0.00  0.00  0.00  0.00  0.00  0.00  0.00  0.00  R LM in model 0.77  0.66  ---  0.47  0.68  0.93  0.56  0.74  0.26  0.23  0.39  0.21  R2 no LM in  0.62  0.63  ---  0.27  0.66  0.93  0.52  0.70  0.15  0.19  0.36  0.15  0.15 **  0.03 **  ---  0.20 **  0.02 **  0.00  0.04 **  0.04 **  0.11 **  0.04 **  0.02 *  0.07 **  model  Boys  R2 difference 2  Tibia  R LM in model 0.79 2  R no LM in model  Radius  R2 difference  model  Girls  R2 difference  Tibia  2  model R2 difference  * p < 0.05; ** p < 0.01 FLoad = failure load; UStress = ultimate stress; Load:Str = load-to-strength ratio; Tt.Ar = total area; Tt.BMD = total bone mineral density; Ct.BMD = cortical bone mineral density; Ct.Po = cortical porosity; Ct.Th = cortical thickness; BV/TV = trabecular bone volume fraction; Tb.N = trabecular number; Tb.Th = trabecular thickness; Tb.Sp = trabecular separation; FM = total body fat mass by DXA.  80  Chapter 6 – Discussion  Chapter 6 – Discussion In this chapter I provide an overview of my results and discuss my findings in the context of the current literature related to the fat-bone relationship. I then discuss the implications of these findings with particular emphasis on the use of HR-pQCT to assess bone quality in studies of children and youth. I then comment on the methodological limitations of this study. I follow this with a discussion of the unique contributions and strengths of this study. I conclude this chapter with my thoughts on future directions for research in this field.  6.1 Overview of Findings In this thesis I explored the association between body fat and aspects of bone quality in children, adolescents and young adults. These data add to the current body of knowledge regarding fat-bone interactions in the growing years and highlight the need to interpret this relationship in the context of the functional model of bone development [7]. In addition, this study is the first to provide the following evidence related to bone microstructure and strength (assessed by HR-pQCT and FEA, respectively): 1) bone strength was lower for greater fat mass at the non-weight bearing distal radius in boys after adjusting for lean mass and other covariates, 2) lean mass was a mediator of the fat-bone relationship and 3) fat mass minimally impacted bone microstructure at both weight-bearing and non weight-bearing sites in girls and boys. Due to the cross-sectional design of this study these findings can only be considered hypothesis generating. Thus, there is a need to further explore these relationships in a longitudinal study or intervention study designed to examine how change in fat mass influences change in bone quality.  6.2 Adiposity and Bone Quality 6.2.1 Bone Strength I observed sex- and bone outcome-specific linear associations among tissues (fat, lean and bone). In boys, bone strength negatively adapted to fat mass at the distal radius after adjusting for lean mass and other covariates (age, ulnar length and ethnicity). I did not observe a negative relationship between fat mass and ultimate stress (another indicator of strength) at the distal radius in boys. This is likely because ultimate stress is calculated as the force applied per unit area, and fat mass was a negative predictor of both failure load and total area. In addition, boys’ 81  Chapter 6 – Discussion bone strength at the distal tibia was NOT associated with fat mass. Simply stated, these results indicate that higher amounts of fat mass may be detrimental to bone strength at the non-load bearing distal radius in boys. It has been postulated that given its contribution to total body mass, fat mass might contribute to increased bone strength at load bearing sites [11]. However, my results did not support this. That is, bone strength was not enhanced in those with higher fat mass after adjusting for lean mass at the distal tibia. Importantly - and as I describe below associations with bone strength were very closely linked with lean mass. Specifically, in keeping with the functional model of bone development [7], all bone strength variables were positively associated with lean mass in boys. Lean mass appeared to mediate the relationship between fat mass and bone strength at the distal tibia in boys. The indication for this mediating effect was, when lean mass was added to any model, the positive association between fat mass and failure load disappeared. This suggests that bone strength at the distal tibia adjusted appropriately to the amount of lean mass present, not to the amount of fat mass. I discuss the implications of this apparent ‘non-adaptation’ below. Unlike boys, fat mass had no influence on bone strength at either the distal radius or distal tibia in girls. Similar to boys, and to underscore the central role that lean mass plays, lean mass positively predicted bone strength at both sites. Further, fat mass positively predicted failure load at the load bearing distal tibia until lean mass was added to the model -- the relationship was then no longer statistically significant. These results once again provide evidence for a key role for lean mass and its mediating influence on the fat-bone strength relationship. Of note, lean mass accounted for less of the variance at the distal radius in girls compared with boys. The finding that fat mass was no longer significantly related to compressive bone strength in girls at distal bone sites after adjusting for lean mass is consistent with studies that used pQCT to assess the fat-bone relationship [10,13,14,156,157]. In particular, two studies examined this relationship in girls and found no significant relationship between fat mass and an estimate of bone strength (BSI) at the distal radius [13] or distal tibia [13,14]. No previous studies assessed the fat-bone strength relationship in boys at distal bone sites using the more advanced HR-pQCT system. Therefore it is not possible to directly compare my results with those of other studies. However, earlier studies reported negative associations between fat cross-sectional area by pQCT [9] or fat mass by DXA [158] and SSIp (an indicator of bone strength) at midshaft sites in boys [9,158]. The factor that seems to underpin the lower bone strength with greater fat mass association in boys in this study and other studies [9,158] is the 82  Chapter 6 – Discussion negative relationship between fat mass and bone area. At metaphyseal sites, estimates of bone strength are proportional to bone area (as in BSI [48] and FLoad [18]). At midshaft sites bone strength (CSMI and SSI) are proportional to the square of bone area, so even small increases in bone area have a substantial influence on bone strength [48]. Thus, as both compressive [48] and bending strength [46] are both related to bone area, even slightly smaller values for bone area would be associated with lower bone strength. Others have speculated and in some cases provided evidence, regarding the implications of overweight as it relates to children’s bone health. Overweight boys are at greater risk of fracture compared with their healthy weight peers [3]. As load-to-strength ratio is an estimate of forearm fracture risk [178], the positive relationship between higher levels of fat mass and loadto-strength in boys that I report, supports this finding. It is important to note that the calculation for load-to-strength ratio does not incorporate fat mass or body mass. 6 0  Fall Load (N) =  ∙  ∙9.81  ht  (Equation 6)  10  Load-to-strength ratio (Ф) =  Fall Load ( ) Failure Load ( )  (Equation 7)  Fall load (especially on an outstretched hand) is theoretically higher in overweight and obese youth as a result of their higher body mass. However, as bone strength appears NOT to adjust to a higher proportion of total body mass from fat mass -- load-to-strength ratio may actually underestimate the risk of forearm fracture in young individuals. There is very limited evidence regarding the tissue-specific relation with fracture in boys and girls. However, we do know that the proportion of all fractures is greater in boys (63%) compared with girls (37%), based on Finnish data collected in 2005 [4]. It is possible that the negative fat-bone strength relationship I reported in boys but not girls may at least partially explain this disparity. Mäyränpää et al. [4] reported a 31% increase in incidence of forearm fractures in Finland from 1983 to 2005. They postulated that escalating levels of child and adolescent obesity and sedentarism might explain this alarming rise in the incidence of forearm fracture. Another study compared bone strength by HR-pQCT and FEA between boys with and without any kind of fracture. They found that failure load was 5.8% lower at the distal tibia in boys with fracture compared with boys without fracture, regardless of fracture site [185]. Although fat mass was not measured in the Chevalley et al. study [185], a negative relationship  83  Chapter 6 – Discussion between excess fat mass and failure load at the distal radius (as per my results) might explain the greater incidence of fracture at the distal radius in boys.  6.2.2 Bone Geometry I have already noted the exponential and positive relationship between bone geometry and bone strength (Section 2.1.3). In this part I delve more deeply into the relation between specific bone geometry variables and fat and lean mass. I noted a site-specific relationship between FM and total bone area that varied based on whether the site was weight bearing (tibia – FM non-significant predictor) or non-weight bearing (radius - FM negative predictor). Again, although I am unable to make direct comparisons between my results (from HR-pQCT) and other studies I present findings from studies that used other imaging systems (DXA and pQCT) to assess bone geometry and fat mass. Previous studies of the fat-bone area relationship used planar DXA to assess total body fat and bone area. They found that obese children had greater bone area than healthy weight children after adjusting for lean mass [149,154]. As bone area by DXA is a two-dimensional (2D) measure, these findings should be interpreted with caution [46]. Cross-sectional measures of 3D bone area as assessed with pQCT and HR-pQCT are not limited by biases related to 2D measures. The outcomes I report using HR-pQCT are consistent with studies (below) that used pQCT to assess bone [12-14,158]. At the distal tibia, three studies found no significant relationships between fat mass (by DXA) and measures of bone geometry (e.g., Tt.Ar, periosteal circumference, and endosteal circumference) in girls aged 8 to 22 yrs [12-14]. Conversely, at the distal radius, two studies reported a negative relationship between body fat (fat cross-sectional area by pQCT [10] and fat mass by DXA [158]) and total bone area in both boys and girls after adjusting for lean mass. However, no studies examined this relationship in boys only. I noted that lean mass was consistently a positive predictor of bone geometry (Tt.Ar) at both distal radius and distal tibia in boys and girls. This relationship has been reported previously in a number of studies that used pQCT [116,123,186] to assess bone geometry and DXA [148,187] to assess bone mass, including a number from our laboratory [61,99]. However, in my study the influence of lean mass on total bone area appears to be greater at the tibia compared with the radius. Further, the fat-bone geometry relationship at the distal radius changed from  84  Chapter 6 – Discussion positive to negative once lean mass was added to the models. This suggests that lean mass mediated the relationship between fat mass and bone geometry. A possible explanation for the disparity in the fat-bone geometry relationship between measured sites may be related to the heterogeneous distribution of soft tissue. I used measures of lean and fat mass from total body DXA scans and the regional distribution of these tissues are different at the forearm compared with the leg [188]. To illustrate, Ducher et al. [10] reported a greater ratio of fat cross-sectional area to muscle cross-sectional area (assessed by pQCT) at the forearm compared with the lower leg. This supports the notion of regional variability in the distribution of fat and lean tissues. Further, Ducher et al. [10] also reported that the difference in the fat-muscle ratio between limbs was greater in overweight children compared with healthy weight children. Importantly, a higher proportion of fat mass relative to lean mass in overweight children at the forearm did not benefit bone area or strength [10]. I did not measure regional distribution of lean and fat tissues in my study. However, given the Ducher et al. findings, a higher proportion of fat compared with muscle at the forearm could explain the negative association between fat mass and total area at the distal radius but not the tibia in both boys and girls, which I observed. What this once again suggests is that bone adapts regionally to the amount of muscle present, not the amount of fat mass. The implications of this, should it hold true, are ultimately localized fracture. There are at least two other potential explanations for the negative relationship between adiposity and bone geometry in girls at the distal radius. First, although I did not assess endocrine levels in my study, it is well known that, in girls and boys, estrogen inhibits periosteal apposition [96,184]. Further, elevated adiposity is linked with increased estrogen levels [189]. Thus, in both girls and boys who have higher fat mass periosteal apposition could potentially be compromised. Lower periosteal apposition (smaller bones) would, in turn, result in lower bone strength. In contrast, strength at the weight-bearing tibia would benefit from loads mediated by muscle mass. The positive influence of mechanical loading at this site could potentially counter any negative influence of estrogen on periosteal apposition. Second, in both boys and girls, obesity is associated with increased levels of proinflammatory cytokines such as tumor necrosis factor-α, interleukin-1 and interleukin-6 [102]. Although I also did not measure these inflammatory factors in my study, proinflammatory cytokines are known to increase bone resorption and may hinder bone formation [102]. It is possible that the participants in my study who had higher body fat also had higher levels of 85  Chapter 6 – Discussion proinflammatory cytokines. This would, in turn, increase bone resorption and limit bone formation, particularly at the endocortical surface, and may explain the negative relationship between excess fat mass and bone geometry I reported in boys and girls at the distal radius. Once again, as I did not observe the same relationship at the distal tibia, the positive influence of mechanical loading at this site could potentially compensate for any deleterious influences of the endocrine environment on bone area.  6.2.3 Bone Density I observed sex-specific relationships between bone density, lean and fat mass whereby lean mass seemed a more important explanatory variable in boys, while fat mass had a small positive influence on bone density in girls. A few earlier studies used pQCT to assess the fatBMD relationship in children and adolescents [9-14]. At the distal tibia, the relationship between fat mass and Tt.BMD after adjusting for lean mass in prepubertal boys and girls [10] and girls 822 yrs [12-14], was not significant. Similarly, at the distal radius fat mass was not significantly associated with Tt.BMD in pre- [9] and post-pubertal boys [11] and post-pubertal girls [12,13]. Fricke et al. [9] reported positive associations between fat (cross-sectional area) and total and trabecular BMD at the distal radius in pre- and post-pubertal girls. However, contrary to my results, Wey et al. [158] reported that fat mass was a negative predictor of total density at the distal radius in boys and premenarcheal girls. There are a few mechanisms that might explain these findings. First, sex differences reported regarding the influence of fat mass on bone density are likely a function of gonadal steroids. In Section 2.3.2, I discuss how androgens stimulate the apposition of bone mineral on the periosteal surface. Estrogens on the other hand, primarily stimulate bone apposition on the endosteal surface [184]. In girls, it has been postulated that the rapid increase in estrogen production at the onset of menarche stimulates increased bone density to consolidate bone mineral for later reproductive needs [184]. As adipocytes secrete estrogens [103], higher circulating estrogen levels in girls with greater fat mass, may partially explain higher bone density values. Second, greater cortical density in girls with more fat mass may indicate decreased intra-cortical remodeling [190]. The ultimate consequence of bone fragility in adults is fracture [191]. We know that compressive bone strength is proportional to the square of its density [49]. However, in cadaveric studies true bone density contributed less to failure load than geometric measures such as total 86  Chapter 6 – Discussion area [192]. In humans, results as to the ability of areal (a)BMD assessed by DXA to predict fracture in individuals are mixed [60]. The picture is further complicated in children. A number of studies reported a negative association between aBMD and fracture in children and adolescents [3,185,193,194]. Specifically, Goulding et al. reported that children who sustained forearm fractures were more likely to be overweight and have low aBMD at the ultra distal radius [3]. How body fat plays into the fracture equation is not known. However given the known relationship between density and compressive bone strength [49], enhanced density at the distal tibia in girls with high fat mass should provide them a bone strength benefit. However, I did not observe a positive relationship between high fat mass and bone strength in girls. Given the substantial contribution of geometry to bone strength, the greater cortical bone density at the distal radius and its contribution to bone strength in girls may have been offset by the negative relationship between fat mass and total area. In boys I did not observe a significant relationship between fat mass and density, yet I noted negative relationships between fat mass and total area, which may explain the negative relationship I observed between fat mass and bone strength.  6.2.4 Bone Microstructure To my knowledge, this is the first study to examine the relationship between fat and bone microstructure in any population. The microstructural variables I discuss are trabecular bone volume fraction, trabecular number, thickness and separation and cortical thickness and porosity. As per other bone outcomes, results were sex- and site-specific. Generally, fat mass did not influence trabecular bone microstructure at the distal radius in either boys or girls. However, at the distal tibia, fat mass appeared to have a modest negative effect on trabecular thickness in girls and cortical thickness in boys. In addition, fat mass negatively predicted cortical porosity at both sites in girls. Given the absence of studies on bone microstructure and fat in children, I turn to the animal literature. A number of animal studies examined the influence of obesity-inducing highfat diets on bone microstructure in mice [195-198]. Results varied due to differences in length of diet intervention and statistical approach. For example, a 14-week obesity causing high-fat diet [195] and 21-week high-fat [198] diet had opposite effects on bone microstructure. The 14-week intervention [195] reported that obese animals on the high-fat diet had lower bone volume and trabecular number with higher trabecular separation (measured with micro-computed tomography (μCT) at the distal femur). In contrast, in the longer ( 1 week) study, obese mice (fed the high fat 87  Chapter 6 – Discussion diet) had greater bone volume, trabecular number and thickness and smaller trabecular separation at the distal femur compared with mice fed a regular diet [198]. In their statistical analyses, the authors only compared group differences and did not adjust for important covariates such as lean mass. In my study, lean mass provided benefits to bone microstructure at the distal radius and distal tibia in boys. This was demonstrated by the strong positive association between lean mass and all bone microstructural parameters at the distal radius and tibia (Ct.Th, BV/TV, Tb.N, Tb.Th and Tb.Sp) - with the exception of trabecular thickness at the distal tibia and cortical porosity at both sites. Lean mass explained a greater proportion of the variance in bone microstructure at the distal tibia, compared with the radius. Further, the mediating influence of lean mass on the fatbone microstructure relationship in boys was also evident. Positive associations between fat mass and bone microstructure variables disappeared when lean mass was added to regression models. Cortical porosity is inversely related to bone strength [199]. In girls, I found that fat mass negatively predicted cortical porosity at both weight bearing and non-weight bearing sites. This would lead me to believe that lower porosity with higher fat mass should provide a strength benefit to girls. However, in girls, lean mass positively predicted cortical porosity, which may negatively influence bone strength. Further, the lowered cortical porosity with greater fat mass in girls may be indicative of decreased intra-cortical remodeling in girls, which may negatively affect bone strength [190]. In girls, lean and fat mass appear to be working against each other as lean mass positively predicts cortical porosity, which suggests that mechanical stimulation from lean mass stimulates bone turnover. In girls at the distal tibia, lean mass was positively associated with Ct.Th, BV/TV and Tb.Th and negatively associated with Tb.Sp. However, there was no association between bone microstructure parameters and lean mass at the distal radius. Thus, the influence of lean mass on bone microstructure in girls and boys is stronger at the weight bearing distal tibia compared with the non-weight bearing distal radius. As discussed previously, variation in the proportion of soft tissues in lower versus the upper limbs may provide some explanation for this apparent sitespecificity [10]. Augat et al. noted a strong relationship between cortical thickness and bone strength [200]. In boys, I observed a negative association between fat mass and cortical thickness at the distal tibia. In girls, fat mass was a negative predictor of Tb.Th at the distal tibia. How the small negative associations between fat mass and some bone microstructure variables relates to 88  Chapter 6 – Discussion increased risk of fracture in overweight adolescents is not clear. Given the cross-sectional nature of my study I accept it as primarily hypothesis generating. Longitudinal research with a larger cohort and targeted intervention studies are needed to clarify the site- and sex-specific relationships between fat and bone microstructure in children, adolescents and young adults and how the relationship contributes to fracture risk.  6.3 Implications of Findings I provide evidence that total body weight from fat mass does not provide an overall benefit to bone in boys or girls. A mechanism that might underpin this finding is the negative association between body fat and bone area. Elevated levels of proinflammatory cytokines likely increase bone resorption [102] at the endocortical surface resulting in a smaller bone area and decreased bone strength. Other as yet unknown aspects within the cellular environment of adipose tissue may also compromise bone quality. In addition, the apparent negative relationship between higher body fat and bone strength at the distal radius in boys may help explain why overweight boys sustain more fractures than their healthy weight peers. Although the evolution of imaging toward HR-pQCT lends new insight into the pediatric fat-bone relationship many questions remain unanswered. We do not know how body fat and bone microstructure influence fracture in children and youth as no study has clearly addressed this. We also do not know how change in fatness might influence change in bone quality over time. Further, I cannot generalize results of my study beyond the study sample and my methods so we also do not know the specific role of maturity, ethnicity and the endocrine environment on fat-bone microstructure relationships in children and youth.  6.3.1 Utility of HR-pQCT in Paediatric Bone Research In this thesis I utilized a novel imaging tool, HR-pQCT, to explore the relationship between body fat and aspects of bone quality in children, adolescents and young adults. Another unique part of my study was that HR- pQCT allowed me to apply finite element analysis to derive valid estimates of bone strength. Thus, I believe my results extend the pediatric bone literature and provide a foundation for further bone microstructure research using HR-pQCT. Although truly a technological advance, I acknowledge a number of issues and limitations related to HR-pQCT that must be addressed in order to improve the efficacy of this imaging tool.  89  Chapter 6 – Discussion First, image acquisition must be standardized in pediatric populations to allow researchers to compare results across studies. Currently there are two methods for reference line placement in pediatric research. One method places the reference line at the proximal edge of the distal epiphyseal growth plate [68]. Another method, the one used in our laboratory, places the reference line at a relative distance from either the radius notch or the tibial plafond [159,160]. Both of these methods are suitable for cross-sectional analysis; however, if the same reference line placement was used across all studies, results would be much easier to compare. Further, in longitudinal studies this reference line needs to be reproduced in order to ensure the same relative area of bone is scanned from measurement to measurement. Over time the epiphyseal growth plate may drift proximally or distally [33,34] and will eventually fuse at approximately 16 yrs for girls, 18 yrs for boys [24]. Thus, using the growth plate to set the reference line may not result in the same relative region of bone being measured from year to year. In contrast, a relative region of interest, as used in my study, ensures the same relative region of bone is measured from time point to time point. Second, we do not know the reliability and accuracy of HR-pQCT acquisition in pediatric populations. Our lab acquired these data in an adult population (reliability) and with cadaveric specimens of older adults (accuracy); however, it is difficult to collect these data in healthy children given the unnecessary exposure to ionizing radiation (albeit small) and lack of availability of pediatric cadaveric specimens. Third, HR-pQCT reference data for children and adolescents are not yet available. Thus, it is not possible to compare clinical populations to normative datasets. Large population-based studies of healthy children across the growing years are needed to create a reference data set. Fourth, due to the gantry size (126mm wide, 150mm deep) researchers are limited to measuring the distal radii and tibiae, only. As a result, measures of muscle cross-sectional area and fat cross-sectional area cannot be obtained at the site where the muscle belly is the largest (e.g., 66% site) [123]. Further, HR-pQCT cannot assess clinically relevant sites such as the proximal femur. Thus, researchers must rely on other tools such as DXA and pQCT to comprehensively assess the relationship between bone quality, fat and lean tissues and at clinically relevant sites such as the proximal femur (DXA). HR-pQCT (8 μm resolution) allows researchers to apply finite element models to estimate bone strength [18]. FE-estimates of bone strength are an improvement over 2-D estimates of bone strength by hip structural analysis and 3-D estimates by pQCT. However, 90  Chapter 6 – Discussion FLoad and UStress calculations are based on uniaxial compressive forces whereas the distal radius and distal tibia experience forces in compression, torsion and shear [18]. These may be altered depending on the context of the applied force. For example, during running the distal tibia experiences compressive forces (primarily from internal muscle force) and posterior shear (primarily from joint reaction force) forces throughout most of the stance phase [201]. Thus, FE is unable to capture the contribution of all forces to bone strength. HR-pQCT remains a valuable tool to investigate bone strength, geometry, volumetric density and microstructure of the distal forearm and leg. Studies that have used HR-pQCT provide insight into specific characteristics of cortical and trabecular bone that contribute to overall bone strength. We are unable to assess many of these parameters with either DXA or pQCT. Although the issues and limitations discussed must be addressed in order to improve utility of HR-pQCT in pediatric research, it should not preclude the use of HR-pQCT in studies of bone quality through the growing years.  6.4 Limitations I acknowledge the limitations of my study. First, based on the cross-sectional study design I cannot infer any causal relationships between fat mass or lean mass and bone quality. Longitudinal studies are required to determine how change in fat mass influences change in bone quality. Additionally, my study sample was based on volunteers and may not represent the larger population. Furthermore, participants were Asian or Caucasian; therefore, results cannot be generalized to youth of other ethnic backgrounds. Additionally, since I used total body fat and lean masses measured by DXA, I did not account for regional variation in the distribution of these tissues [10]. Due to the small number of participants with excess fat mass (13% overweight, 6% obese), I may have been underpowered to detect significant relationships between fat mass and bone outcomes. The few participants with excess fat mass may have disproportionately influenced the relationship between adiposity and bone quality. My results do not capture the relationship between fat and bone quality in overweight or obese youth. More overweight and obese participants would have provided me a better opportunity to assess the fat-bone relationship in children, adolescents and young adults. I explored this source of potential bias further by running the analyses with and without the boy with the greatest fat mass (56.7 kg fat  91  Chapter 6 – Discussion mass, Figure 17). Inference with respect to the fat-bone relationship did not change (data not shown). The age range of study participants was bimodal and the sample did not include any adolescents between the ages of 12 and 15 years. After exploratory data analysis I concluded that the relationship between adiposity and bone quality was similar in the younger and older cohorts; therefore, I analyzed all participants in the same multivariable linear regression models. Consequently, I assumed the relationships I discuss are similar for the missing age group. I do not know if this is true; however, my data exploration leads me to believe that this is a reasonable assumption. In addition, due to the wide age range I used chronological age (boys) or centered age and centered age squared (girls) to account for differences in age and maturation. Calfee et al. [202] found that in American adolescents skeletal age exceeded chronological age by 0.8 yrs, on average. Thus, they concluded that age is an adequate estimate of maturation. Biological age (skeletal age measured using a hand-wrist radiograph [24] or age at peak height velocity (from longitudinal data)) better represents maturity. Thus, prospective studies of the fat-bone relationship across the growing years would benefit from this measure, in future. Finally, I did not measure any endocrine markers, such as leptin or estrogen, or proinflammatory cytokines that are strongly linked with adiposity and bone health [11,184]. Therefore, I cannot report their contribution to my findings and can only speculate on the influence of adiposity-related hormones on bone quality. For instance, estrogen may be responsible for the positive relationship I observed between excess fat mass and bone density in girls. Previously, leptin was found to negatively predict cortical area and thickness [11], and may be a mechanism driving the negative fat-bone strength relationship in boys.  6.5 Unique Contributions and Strengths Despite these limitations this study brings a number of unique contributions and strengths to the pediatric fat-bone literature which I discuss in this section. It is the first to describe the relationship between fat mass and HR-pQCT derived outcomes of bone quality in a relatively large cohort of children, adolescents and young adults. The imaging system used to collect these data (HR-pQCT) is one of relatively few worldwide. This system’s resolution is considerably higher than that of pQCT (82 μm vs. 0.4 mm). This advance in technology allows for the independent assessment of cortical and trabecular bone microstructure, which both contribute to total bone strength [203]. In addition, the application of customized auto-segmentation algorithms 92  Chapter 6 – Discussion and finite element models enables the assessment of cortical porosity and bone strength, respectively. Importantly, I also investigated the relative contributions of fat and lean mass (a surrogate of muscle force) to bone quality. This allowed me to describe how lean mass mediated the relationship between fat mass and bone quality. This study will lay the foundation for future studies that examine the fat-bone quality relationship across the growing years and into adulthood. Finally, my study provides evidence that high levels of fat mass may have a negative influence on bone strength in boys. This adds to a growing body of research regarding the dire consequences of child and adolescent overweight and obesity and supports the need for effective interventions to promote healthy weight across the growing years.  6.6 Future Directions Since the endocrine environment most likely plays an important role in the fat-bone relationship, future studies would benefit from additional measures that include proinflammatory cytokines and endocrine markers such as leptin and gonadal hormones. This may help to clarify potential mechanisms that underpin the fat-bone relationship. The cross-sectional design of this study limits my ability to infer causal relationships between fat mass and bone quality. Therefore, future research would benefit from longitudinal studies that determine how change in fat mass influences change in bone quality in relation to changes in muscle mass. We also need to clarify the fat-bone relationship in the context of the functional model of bone development and the added influence of mechanical loading through exercise. Furthermore, a larger sample size with more participants who were overweight or obese would serve to enhance the statistical power to detect a relationship between fat mass and bone quality.  93  Chapter 7 – Summary and Conclusions  Chapter 7 - Summary and Conclusions In this chapter I summarize the findings of the primary and secondary objectives of this thesis and provide conclusions.  7.1 Primary Objective: Bone Strength I hypothesized that after adjusting for lean mass (a surrogate of muscle force), adiposity would not be positively associated with bone strength at the distal radius and distal tibia in boys or girls. My findings support this hypothesis. In boys:   Greater fat mass after adjusting for lean mass was negatively associated with bone strength at the distal radius.    Fat mass was not beneficial for bone strength at the distal tibia.    Lean mass mediated the relationship between fat mass and bone strength at the distal tibia.  In girls:   Greater fat mass after adjusting for lean mass did not influence bone strength at the distal radius or distal tibia.    Lean mass mediated the relationship between adiposity and bone strength at the distal tibia.  In boys and girls:   The influence of lean mass on bone strength was greater at the distal tibia than at the distal radius.  7.2 Secondary Objective: Bone Geometry, Density and Microstructure I hypothesized that after adjusting for lean mass (a surrogate of muscle force), adiposity would not be positively associated with bone geometry, density or microstructure at the distal radius and distal tibia in boys or girls. In general, my results support this hypothesis. In boys:   High levels of fat mass did not benefit bone density or microstructure at either the distal radius or the distal tibia.  94  Chapter 7 – Summary and Conclusions   Lean mass positively predicted bone density and microstructure at both the distal radius and distal tibia.  In girls:   High levels of fat mass had a modest positive relationship with cortical bone density at the distal radius and negative relationship with cortical porosity at the distal radius and distal tibia.    Fat mass was not an independent predictor of bone microstructure at the distal radius.    Fat mass was an independent negative predictor of trabecular thickness but no other microstructural variable at the distal tibia.    Lean mass did not predict bone microstructure at the distal radius.  In boys and girls:   Fat mass was an independent negative predictor of total area at the distal radius.    Lean mass mediated the relationship between fat mass and total bone area, particularly at the distal tibia.  7.3 Conclusions In conclusion, my results indicate that the fat-bone relationship in children, adolescents and young adults is sex- and site-specific. As in other studies, I highlight the importance of lean mass and other surrogates of muscle force to bone strength. This substantial relationship supports the importance of adjusting for muscle when examining the fat-bone relationship. Future studies with larger cohorts and a greater number of overweight and obese participants would help clarify the sex- and site-specific fat-bone relationships. Longitudinal studies are essential to further describe the fat-bone relationship across the growing years and into adulthood. The potentially hazardous influence of high levels of fat mass on children and youth bone health can be added to the list of dire consequences of overweight and obesity. Targeted and strategic interventions are needed to both evaluate the influence of physical activity and other lifestyle factors on the fatbone relationship and offset the escalating trend of unhealthy weight during the growing years.  95  References 1.  Bell J, Rogers VW, Dietz WH, Ogden CL, Schuler C, Popovic T. CDC Grand Rounds: Childhood obesity in the United States. MMWR: Morb Mortal Wkly Rep. 2011;60(2):4246.  2.  Taylor ED, Theim KR, Mirch MC, Ghorbani S, Tanofsky-Kraff M, Adler-Wailes D, et al. Orthopedic complications of overweight in children and adolescents. Pediatrics. 2006;117(6):2167-2174.  3.  Goulding A, Grant AM, Williams SM. Bone and body composition of children and adolescents with repeated forearm fractures. J Bone Miner Res. 2005;20(12):2090-2096.  4.  Mäyränpää MK, Mäkitie O, Kallio PE. Decreasing incidence and changing pattern of childhood fractures: A population-based study. J Bone Miner Res. 2010;25(12):2752-2759.  5.  Bailey DA, McKay HA, Mirwald RL, Crocker PR, Faulkner RA. A six-year longitudinal study of the relationship of physical activity to bone mineral accrual in growing children: The University of Saskatchewan Bone Mineral Accrual Study. J Bone Miner Res. 1999;14(10):1672-1679.  6.  Frost HM. Bone "mass" and the "mechanostat": A proposal. Anat Rec. 1987;219(1):1-9.  7.  Rauch F, Schoenau E. The developing bone: Slave or master of its cells and molecules? Pediatr Res. 2001;50(3):309-314.  8.  McKay H, Liu D, Egeli D, Boyd S, Burrows M. Physical activity positively predicts bone architecture and bone strength in adolescent males and females. Acta Paediatr. 2011;100(1):97-101.  9.  Fricke O, Sumnik Z, Tutlewski B, Stabrey A, Remer T, Schoenau E. Local body composition is associated with gender differences of bone development at the forearm in puberty. Horm Res. 2008;70(2):105-111.  10. Ducher G, Bass SL, Naughton GA, Eser P, Telford RD, Daly RM. Overweight children have a greater proportion of fat mass relative to muscle mass in the upper limbs than in the lower limbs: implications for bone strength at the distal forearm. Am J Clin Nutr. 2009;90(4):1104-1111. 11. Lorentzon M, Landin K, Mellström D, Ohlsson C. Leptin is a negative independent predictor of areal BMD and cortical bone size in young adult Swedish men. J Bone Miner Res. 2006;21(12):1871-1878. 12. Pollock NK, Laing EM, Baile CA, Hamrick MW, Hall DB, Lewis RD. Is adiposity advantageous for bone strength? A peripheral quantitative computed tomography study in late adolescent females. Am J Clin Nutr. 2007;86(5):1530-1538.  96  13. Pollock NK, Laing EM, Hamrick MW, Baile CA, Hall DB, Lewis RD. Bone and fat relationships in postadolescent black females: A pQCT study. Osteoporos Int. 2011;22(2):655-665. 14. Farr JN, Chen Z, Lisse JR, Lohman TG, Going SB. Relationship of total body fat mass to weight-bearing bone volumetric density, geometry, and strength in young girls. Bone. 2010;46(4):977-984. 15. Janicka A, Wren TAL, Sanchez MM, Dorey F, Kim PS, Mittelman SD, et al. Fat mass is not beneficial to bone in adolescents and young adults. J Clin Endocrinol Metab. 2007;92(1):143-147. 16. Currey JD. Bones: Structure and mechanics. 2nd ed. Princeton, New Jersey: Princeton University Press; 2002. 17. Schaffler MB, Burr DB. Stiffness of compact bone: effects of porosity and density. J Biomech. 1988;21(1):13-16. 18. MacNeil JA, Boyd SK. Bone strength at the distal radius can be estimated from highresolution peripheral quantitative computed tomography and the finite element method. Bone. 2008;42(6):1203-1213. 19. Sommerfeldt DW, Rubin CT. Biology of bone and how it orchestrates the form and function of the skeleton. 2001;10:S86-S95. 20. Forwood M. Physiology. In: Khan K, McKay H, Kannus P, Bailey D, Wark J, Bennell K, editors. Physical activity and bone health Champaign, Ill.; United States: Human Kinetics; 2001. 21. Weiner S, Traub W, Wagner HD. Lamellar bone: Structure-function relations. J Struct Biol. 1999;126(3):241-255. 22. Faulkner RA, Bailey DA. Osteoporosis: A pediatric concern? Med Sport Sci. 2007;51:1-12. 23. Hancox NM. Biology of bone. Great Britain, University Printing House, Cambridge: Cambridge University Press; 1972. 24. Malina RM, Bouchard C, Bar-Or O. Growth, maturation, and physical activity. 2nd ed. Champaign, Ill.; United States: Human Kinetics; 2004. 25. Currey JD. The many adaptations of bone. J Biomech. 2003;36(10):1487. 26. Carter DR, Spengler DM. Mechanical properties and composition of cortical bone. Clin Orthop Relat Res. 1978(135):192-217. 27. Zebaze RMD, Jones AC, Pandy MG, Knackstedt MA, Seeman E. Differences in the degree of bone tissue mineralization account for little of the differences in tissue elastic properties. Bone. 2011;48(6):1246-1251. 97  28. Smit TH, Odgaard A, Schneider E. Structure and function of vertebral trabecular bone. Spine (Phila Pa 1976). 1997;22(24):2823-2833. 29. Parfitt AM, Travers R, Rauch F, Glorieux FH. Structural and cellular changes during bone growth in healthy children. Bone. 2000;27(4):487-494. 30. Cadet ER, Gafni RI, McCarthy EF, McCray DR, Bacher JD, Barnes KM, et al. Mechanisms responsible for longitudinal growth of the cortex: Coalescence of trabecular bone into cortical bone. J Bone Joint Surg Am. 2003;85(9):1739. 31. Nilsson O, Baron J. Fundamental limits on longitudinal bone growth: Growth plate senescence and epiphyseal fusion. Trends Endocrinol Metab. 2004;15(8):370-374. 32. Rauch F. Bone growth in length and width: The Yin and Yang of bone stability. J Musculoskelet Neuronal Interact. 2005;5(3):194-201. 33. Pritchett JW. Longitudinal growth and growth-plate activity in the lower extremity. Clin Orthop Relat Res. 1992(275):274-279. 34. Pritchett JW. Growth plate activity in the upper extremity. Clin Orthop Relat Res. 1991(268):235-242. 35. Wallis GA. Bone growth: Coordinating chondrocyte differentiation. Curr Biol. 1996;6(12):1577-1580. 36. Wang Q, Cheng S, Alén M, Seeman E. Bone's structural diversity in adult females is established before puberty. J Clin Endocrinol Metab. 2009;94(5):1555-1561. 37. Venken K, Callewaert F, Boonen S, Vanderschueren D. Sex hormones, their receptors and bone health. Osteoporos Int. 2008;19(11):1517-1525. 38. Iuliano-Burns S, Mirwald RL, Bailey DA. Timing and magnitude of peak height velocity and peak tissue velocities for early, average, and late maturing boys and girls. Am J Hum Biol. 2001;13(1):1-8. 39. Mirwald RL, Baxter-Jones A, Bailey DA. An assessment of maturity from anthropometric measurements. Med Sci Sports Exerc. 2002;34(4):689-694. 40. Heinonen A. Biomechanics. In: Khan K, McKay H, Kannus P, Bailey D, Wark J, Bennell K, et al, editors. In Physical activity and bone health, Champaign, Ill., Human Kinetics, c2001, p.23-34. United States; 2001. 41. Rho J, Kuhn-Spearing L, Zioupos P. Mechanical properties and the hierarchical structure of bone. Med Eng Phys. 1998;20(2):92-102. 42. Turner CH, Burr DB. Basic biomechanical measurements of bone: A tutorial. Bone. 1993;14(4):595-608.  98  43. Currey JD. Bone strength: What are we trying to measure? Calcif Tissue Int. 2001;68(4):205-210. 44. Currey JD, Butler G. The mechanical properties of bone tissue in children. J Bone Joint Surg Am. 1975;57(6):810-814. 45. Ferretti JL, Capozza RF, Zanchetta JR. Mechanical validation of a tomographic (pQCT) index for noninvasive estimation of rat femur bending strength. Bone. 1996;18(2):97-102. 46. Leonard MB, Shults J, Elliott DM, Stallings VA, Zemel BS. Interpretation of whole body dual energy X-ray absorptiometry measures in children: Comparison with peripheral quantitative computed tomography. Bone. 2004;34(6):1044-1052. 47. Macdonald HM. Effectiveness of a school-based physical activity intervention for increasing bone strength in children: Action Schools! BC. Vancouver: The University of British Columbia; 2006. Retrieved from circle.ubc.ca/handle/2429/18591 August 28, 2012. 48. Kontulainen SA, Johnston JD, Liu D, Leung C, Oxland TR, McKay HA. Strength indices from pQCT imaging predict up to 85% of variance in bone failure properties at tibial epiphysis and diaphysis. J Musculoskelet Neuronal Interact. 2008;8(4):401-409. 49. Carter DR, Hayes WC. Bone compressive strength: the influence of density and strain rate. Science. 1976;194:1174-1176. 50. Keaveny TM, Morgan EF, Niebur GL, Yeh OC. Biomechanics of trabecular bone. Annu Rev Biomed Eng. 2001;3:307-333. 51. Goldstein SA. The mechanical properties of trabecular bone: dependence on anatomic location and function. J Biomech. 1987;20(11-12):1055-1061. 52. Augat P, Schorlemmer S. The role of cortical bone and its microstructure in bone strength. Age Ageing. 2006;35:ii27-ii31. 53. Manske S, Macdonald H, Nishiyama K, Boyd S, McKay H. Clinical tools to evaluate bone strength. Clin Rev Bone Miner Metab. 2010;8(3):122-134. 54. Wren TAL, Liu X, Pitukcheewanont P, Gilsanz V. Bone acquisition in healthy children and adolescents: Comparisons of dual-energy x-ray absorptiometry and computed tomography measures. J Clin Endocrinol Metab. 2005;90(4):1925-1928. 55. Bouxsein ML, Seeman E. Quantifying the material and structural determinants of bone strength. Best Pract Res Clin Rheumatol. 2009;23(6):741-753. 56. Petit MA, Beck TJ, Kontulainen SA. Examining the developing bone: What do we measure and how do we do it? J Musculoskelet Neuronal Interact. 2005;5(3):213-224. 57. Carter DR, Bouxsein ML, Marcus R. New approaches for interpreting projected bone densitometry data. J Bone Miner Res. 1992;7(2):137-145. 99  58. Schoenau E, Saggese G, Peter F, Baroncelli GI, Shaw NJ, Crabtree NJ, et al. From bone biology to bone analysis. Horm Res. 2004;61(6):257-269. 59. Siu WS, Qin L, Leung KS. pQCT bone strength index may serve as a better predictor than bone mineral density for long bone breaking strength. J Bone Miner Metab. 2003;21(5):316322. 60. Liu D, Manske SL, Kontulainen SA, Tang C, Guy P, Oxland TR, et al. Tibial geometry is associated with failure load ex vivo: A MRI, pQCT and DXA study. Osteoporosis Int. 2007;18(7):991-997. 61. Macdonald H, Kontulainen S, Petit M, Janssen P, McKay H. Bone strength and its determinants in pre- and early pubertal boys and girls. Bone. 2006;39(3):598-608. 62. Boutroy S, Bouxsein ML, Munoz F, Delmas PD. In vivo assessment of trabecular bone microarchitecture by high-resolution peripheral quantitative computed tomography. J Clin Endocrinol Metab. 2005;90(12):6508-6515. 63. Buie HR, Campbell GM, Klinck RJ, MacNeil JA, Boyd SK. Automatic segmentation of cortical and trabecular compartments based on a dual threshold technique for in vivo microCT bone analysis. Bone. 2007;41(4):505-515. 64. MacNeil JA, Boyd SK. Accuracy of high-resolution peripheral quantitative computed tomography for measurement of bone quality. Med Eng Phys. 2007;29(10):1096-1105. 65. Pauchard Y, Ayres F, Boyd SK. Automated quantification of three-dimensional subject motion to monitor image quality in high-resolution peripheral quantitative computed tomography. Phys Med Biol. 2011;56(20):6523-6543. 66. Pauchard Y, Liphardt A, Macdonald HM, Hanley DA, Boyd SK. Quality control for bone quality parameters affected by subject motion in high-resolution peripheral quantitative computed tomography. Bone. 2012;50(6):1304-1310. 67. Nishiyama KK, Macdonald HM, Moore SA, Fung T, Boyd SK, McKay HA. Cortical porosity is higher in boys compared with girls at the distal radius and distal tibia during pubertal growth: An HR-pQCT study. J Bone Miner Res. 2012;27(2):273-282. 68. Kirmani S, Christen D, van Lenthe GH, Fischer PR, Bouxsein ML, McCready LK, et al. Bone structure at the distal radius during adolescent growth. J Bone Miner Res. 2009;24(6):1033-1042. 69. Duren DL, Sherwood RJ, Choh AC, Czerwinski SA, Chumlea WC, Lee M, et al. Quantitative genetics of cortical bone mass in healthy 10-year-old children from the Fels Longitudinal Study. Bone. 2007;40(2):464-470. 70. Prentice A. The relative contribution of diet and genotype to bone development. Proc Nutr Soc. 2001;60(1):45-52.  100  71. Bex M, Bouillon R. Growth hormone and bone health. Horm Res. 2003;60 Suppl 3:80-86. 72. Deng H, Liu Y. Current Topics in Bone Biology. River Edge, NJ, USA: World Scientific Publishing Co.; 2005. 73. Bounds W, Skinner J, Carruth BR, Ziegler P. The relationship of dietary and lifestyle factors to bone mineral indexes in children. J Am Diet Assoc. 2005;105(5):735-741. 74. Havill LM, Mahaney MC, Binkley TL, Specker BL. Effects of genes, sex, age, and activity on BMC, bone size, and areal and volumetric BMD. J Bone Miner Res. 2007;22(5):737746. 75. Wang X, Kammerer CM, Wheeler VW, Patrick AL, Bunker CH, Zmuda JM. Genetic and environmental determinants of volumetric and areal BMD in multi-generational families of African ancestry: The Tobago Family Health Study. J Bone Miner Res. 2007;22(4):527-536. 76. Khan K, McKay H, Kannus P, Bailey D, Wark J, Bennell K. Age, sex, genetics, and race. In Physical activity and bone health, Champaign, Ill., Human Kinetics, c2001, p.59-70. United States; 2001. 77. Sowers M, Willing M, Burns T, Deschenes S, Hollis B, Crutchfield M, et al. Genetic markers, bone mineral density, and serum osteocalcin levels. J Bone Miner Res. 1999;14(8):1411-1419. 78. Cashman KD, Seamans K. Bone health, genetics, and personalised nutrition. Genes Nutr. 2007;2(1):47-51. 79. Kelly PJ, Eisman JA, Sambrook PN. Interaction of genetic and environmental influences on peak bone density. Osteoporos Int. 1990;1(1):56-60. 80. McKay HA, Petit MA, Khan KM, Schutz RW. Lifestyle determinants of bone mineral: A comparison between prepubertal Asian- and Caucasian-Canadian boys and girls. Calcif Tissue Int. 2000;66(5):320-324. 81. Barrett-Connor E, Siris ES, Wehren LE, Miller PD, Abbott TA, Berger ML, et al. Osteoporosis and fracture risk in women of different ethnic groups. J Bone Miner Res. 2005;20(2):185-194. 82. Walker MD, Babbar R, Opotowsky AR, Rohira A, Nabizadeh F, Badia MD, et al. A referent bone mineral density database for Chinese American women. Osteoporos Int. 2006;17(6):878-887. 83. Nam H, Shin M, Zmuda JM, Leung PC, Barrett-Connor E, Orwoll ES, et al. Race/ethnic differences in bone mineral densities in older men. Osteoporos Int. 2010;21(12):2115-2123. 84. MacKelvie KJ, McKay HA, Khan KM, Crocker PRE. Lifestyle risk factors for osteoporosis in Asian and Caucasian girls. Med Sci Sports Exerc. 2001;33(11):1818-1824.  101  85. Bhudhikanok GS, Wang MC, Eckert K, Matkin C, Marcus R, Bachrach LK. Differences in bone mineral in young Asian and Caucasian Americans may reflect differences in bone size. J Bone Miner Res. 1996;11(10):1545-1556. 86. Wang X, Wang Q, Ghasem-Zadeh A, Evans A, McLeod C, Iuliano-Burns S, et al. Differences in macro- and microarchitecture of the appendicular skeleton in young Chinese and white women. J Bone Miner Res. 2009;24(12):1946-1952. 87. Walker MD, McMahon DJ, Udesky J, Liu G, Bilezikian JP. Application of high-resolution skeletal imaging to measurements of volumetric BMD and skeletal microarchitecture in Chinese-American and white women: Explanation of a paradox. J Bone Miner Res. 2009;24(12):1953-1959. 88. Walker MD, Liu XS, Stein E, Zhou B, Bezati E, McMahon DJ, et al. Differences in bone microarchitecture between postmenopausal Chinese-American and white women. J Bone Miner Res. 2011;26(7):1392-1398. 89. Kim S, Nettlefold L, Hoy C, Moore S, Donaldson M, Macdonald H, et al. A comparison of trabecular bone microstructure at the distal radius between postpubertal Asian and Caucasian adolescents: An HR-pQCT study. J Bone Miner Res. 2011;26, Suppl. 1:S248. 90. Bronner F. The calcitropic hormones: Parathyroid hormone, calcitonin, and vitamin D. In: Bronner F, Worrell RV, editors. Orthopaedics: Pirnciples of basic and clinical science. Boca Raton: CRC Press; 1999. 91. Gordon CM. Normal bone accretion and effects of nutritional disorders in childhood. J Womens Health (Larchmt). 2003;12(2):137. 92. Kroll MH. Parathyroid hormone temporal effects on bone formation and resorption. Bull Math Biol. 2000;62(1):163-188. 93. Hill TR, Cotter AA, Mitchell S, Boreham CA, Dubitzky W, Murray L, et al. Vitamin D status and parathyroid hormone relationship in adolescents and its association with bone health parameters: Analysis of the Northern Ireland Young Heart's Project. Osteoporos Int. 2010;21(4):695-700. 94. Lehtonen-Veromaa M, Möttönen T, Nuotio IO, Irjala K, Leino AE, Viikari J. Vitamin D and attainment of peak bone mass among peripubertal Finnish girls: A 3-y prospective study. Am J Clin Nutr. 2002;76(6):1446-1453. 95. Greene DA, Naughton GA. Calcium and vitamin-D supplementation on bone structural properties in peripubertal female identical twins: A randomised controlled trial. Osteoporos Int. 2011;22(2):489-498. 96. Bronner F. Systemic and local modifiers of bone other than the calcitropic hormones. In: Bronner F, Worrell RV, editors. Orthopaedics: Pirnciples of basic and clinical science. Boca Raton: CRC Press; 1999.  102  97. Wang Q, Alén M, Nicholson PHF, Halleen JM, Alatalo SL, Ohlsson C, et al. Differential effects of sex hormones on peri- and endocortical bone surfaces in pubertal girls. J Clin Endocrinol Metab. 2006;91(1):277-282. 98. MacKelvie KJ, McKay HA, Khan KM, Crocker P. A school-based exercise intervention augments bone mineral accrual in early pubertal girls. J Pediatr. 2001;139(4):501-508. 99. Macdonald HM, Kontulainen SA, MacKelvie-O'Brien K,J., Petit MA, Janssen P, Khan KM, et al. Maturity- and sex-related changes in tibial bone geometry, strength and bone-muscle strength indices during growth: A 20-month pQCT study. Bone. 2005;36(6):1003-1011. 100. Burrows M, Liu D, Moore S, McKay H. Bone microstructure at the distal tibia provides a strength advantage to males in late puberty: An HR-pQCT study. J Bone Miner Res. 2010;25(6):1423-1432. 101. Kontulainen SA, Macdonald HM, Khan KM, McKay HA. Examining bone surfaces across puberty: A 20-month pQCT trial. J Bone Miner Res. 2005;20(7):1202-1207. 102. Cao JJ. Effects of obesity on bone metabolism. J Orthop Surg Res. 2011;6(1):30-36. 103. Reid IR. Relationships between fat and bone. Osteoporos Int. 2008;19(5):595-606. 104. Lanham-New S, Thompson RL, More J, Brooke-Wavell K, Hunking P, Medici E. Importance of vitamin D, calcium and exercise to bone health with specific reference to children and adolescents. Nutr Bull. 2007;32(4):364-377. 105. Rauch F, Bailey DA, Baxter-Jones A, Mirwald R, Faulkner R. The 'muscle-bone unit' during the pubertal growth spurt. Bone. 2004;34(5):771-775. 106. Daly RM. The effect of exercise on bone mass and structural geometry during growth. Med Sport Sci. 2007;51:33-49. 107. Hind K, Burrows M. Weight-bearing exercise and bone mineral accrual in children and adolescents: A review of controlled trials. Bone. 2007;40(1):14-27. 108. Nikander R, Sievänen H, Heinonen A, Daly RM, Uusi-Rasi K, Kannus P. Targeted exercise against osteoporosis: A systematic review and meta-analysis for optimising bone strength throughout life. BMC Med. 2010;8:47-47. 109. Weaver CM. Vitamin D, calcium homeostasis, and skeleton accretion in children. J Bone Miner Res. 2007;22 Suppl 2:V45-V49. 110. Huncharek M, Muscat J, Kupelnick B. Impact of dairy products and dietary calcium on bone-mineral content in children: Results of a meta-analysis. Bone. 2008;43(2):312-321. 111. Ondrak KS, Morgan DW. Physical activity, calcium intake and bone health in children and adolescents. Sports Med. 2007;37(7):587-601.  103  112. Frost HM. Bone's mechanostat: A 2003 update. Anat Rec A Discov Mol Cell Evol Biol. 2003;275(2):1081-1101. 113. Schoenau E. From mechanostat theory to development of the "Functional Muscle-BoneUnit". J Musculoskelet Neuronal Interact. 2005;5(3):232-238. 114. Schoenau E, Fricke O. Mechanical influences on bone development in children. Eur J Endocrinol. 2008;159:S27-S31. 115. Rittweger J. Ten years muscle-bone hypothesis: What have we learned so far?--almost a festschrift--. J Musculoskelet Neuronal Interact. 2008;8(2):174-178. 116. Schönau E. The development of the skeletal system in children and the influence of muscular strength. Horm Res. 1998;49(1):27-31. 117. Xu L, Nicholson P, Wang Q, Alén M, Cheng S. Bone and muscle development during puberty in girls: A seven-year longitudinal study. J Bone Miner Res. 2009;24(10):16931698. 118. Fricke O, Beccard R, Semler O, Schoenau E. Analyses of muscular mass and function: The impact on bone mineral density and peak muscle mass. Pediatr Nephrol. 2010;25(12):23932400. 119. Fricke O, Weidler J, Tutlewski B, Schoenau E. Mechanography--a new device for the assessment of muscle function in pediatrics. Pediatr Res. 2006;59(1):46-49. 120. Heymsfield SB, Smith R, Aulet M, Bensen B, Lichtman S, Wang J, et al. Appendicular skeletal muscle mass: Measurement by dual-photon absorptiometry. Am J Clin Nutr. 1990;52(2):214-218. 121. Humphries IR, Hua V, Ban L, Gaskin KJ, Howman-Giles R. Validation of estimates of lean tissue mass made by dual-energy X-ray absorptiometry. Ann N Y Acad Sci. 2000;904:104106. 122. Witzke KA, Snow CM. Lean body mass and leg power best predict bone mineral density in adolescent girls. Med Sci Sports Exerc. 1999;31(11):1558-1563. 123. Schoenau E, Neu CM, Beck B, Manz F, Rauch F. Bone mineral content per muscle crosssectional area as an index of the functional muscle-bone unit. J Bone Miner Res. 2002;17(6):1095-1101. 124. Lorbergs AL, Farthing JP, Baxter-Jones A, Kontulainen SA. Forearm muscle size, strength, force, and power in relation to pQCT-derived bone strength at the radius in adults. Appl Physiol Nutr Metab. 2011;36(5):618-625. 125. Shields M. Overweight and obesity among children and youth.Health Rep. 2006;17(3):2742.  104  126. Margolis KL, Ensrud KE, Schreiner PJ, Tabor HK. Body size and risk for clinical fractures in older women. Ann Intern Med. 2000;133(2):123-127. 127. Rosen CJ, Klibanski A. Bone, fat, and body composition: Evolving concepts in the pathogenesis of osteoporosis. Am J Med. 2009;122(5):409-414. 128. Rosen CJ, Ackert-Bicknell C, Rodriguez JP, Pino AM. Marrow fat and the bone microenvironment: Developmental, functional, and pathological implications. Crit Rev Eukaryot Gene Expr. 2009;19(2):109-124. 129. Bredella M. Perspective: The bone-fat connection. Skeletal Radiol.2010;39(8):729-731. 130. Di Iorgi N, Mo AO, Grimm K, Wren TAL, Dorey F, Gilsanz V. Bone acquisition in healthy young females is reciprocally related to marrow adiposity. J Clin Endocrinol Metab. 2010;95(6):2977-2982. 131. Van Der Ploeg G,E., Withers RT, Laforgia J. Percent body fat via DEXA: Comparison with a four-compartment model. J Appl Physiol. 2003;94(2):499-506. 132. Wells JCK, Fewfrell MS. Measuring body composition. Arch Dis Child. 2006;91(7):612617. 133. Treuth MS, Butte NF, Wong WW, Ellis KJ. Body composition in prepubertal girls: comparison of six methods. Int J Obes Relat Metab Disord. 2001;25(9):1352. 134. Pietrobelli A, Faith MS, Allison DB, Gallagher D, Chiumello G, Heymsfield SB. Body mass index as a measure of adiposity among children and adolescents: A validation study. J Pediatr. 1998;132(2):204-210. 135. Maynard LM, Wisemandle W, Roche AF, Chumlea WC, Guo SS, Siervogel RM. Childhood body composition in relation to body mass index. Pediatrics. 2001;107(2):344350. 136. Kuczmarski RJ, Ogden CL, Grummer-Strawn L, Flegal KM, Guo SS, Wei R, et al. CDC growth charts: United States. Adv Data. 2000(314):1-27. 137. Cole TJ, Freeman JV, Preece MA. Body mass index reference curves for the UK, 1990. Arch Dis Child. 1995;73(1):25-29. 138. Cole TJ. Establishing a standard definition for child overweight and obesity worldwide: International survey. BMJ. 2000;320(7244):1240. 139. Power C, Lake JK, Cole TJ. Review: Measurement and long-term health risks of child and adolescent fatness. Int J Obes Relat Metab Disord. 1997;21(7):507. 140. Shypailo RJ, Butte NF, Ellis KJ. DXA: Can it be used as a criterion reference for body fat measurements in children? Obesity. 2008;16(2):457-462.  105  141. Wells JCK, Haroun D, Williams JE, Wilson C, Darch T, Viner RM, et al. Evaluation of DXA against the four-component model of body composition in obese children and adolescents aged 5–21 years. Int J Obes (Lond). 2010;34(4):649-655. 142. Sopher AB, Thornton JC, Wang J, Pierson RN, Heymsfield SB, Horlick M. Measurement of percentage of body fat in 411 children and adolescents: A comparison of dual-energy Xray absorptiometry with a four-compartment model. Pediatrics. 2004;113(5):1285-1290. 143. Wong WW, Hergenroeder AC, Stuff JE, Butte NF, Smith EO, Ellis KJ. Evaluating body fat in girls and female adolescents: Advantages and disadvantages of dual-energy X-ray absorptiometry. Am J Clin Nutr. 2002;76(2):384-389. 144. Silva AM, Minderico CS, Teixeira PJ, Pietrobelli A, Sardinha LB. Body fat measurement in adolescent athletes: Multicompartment molecular model comparison. Eur J Clin Nutr. 2006;60(8):955-964. 145. Wang MC, Bachrach LK, Van Loan M, Hudes M, Flegal KM, Crawford PB. The relative contributions of lean tissue mass and fat mass to bone density in young women. Bone. 2005;37(4):474-481. 146. El Hage RP, Courteix D, Benhamou C, Jacob C, Jaffré C. Relative importance of lean and fat mass on bone mineral density in a group of adolescent girls and boys. Eur J Appl Physiol. 2009;105(5):759-764. 147. Ackerman A, Thornton JC, Wang J, Pierson RN, Horlick M. Sex difference in the effect of puberty on the relationship between fat mass and bone mass in 926 healthy subjects, 6 to 18 years old. Obesity (Silver Spring). 2006;14(5):819-825. 148. Petit MA, Beck TJ, Shults J, Zemel BS, Foster BJ, Leonard MB. Proximal femur bone geometry is appropriately adapted to lean mass in overweight children and adolescents. Bone. 2005;36(3):568-576. 149. Leonard MB, Shults J, Wilson BA. Obesity during childhood and adolescence augments bone mass and bone dimensions. Am J Clin Nutr. 2004;80(2):514-523. 150. Goulding A, Taylor RW, Jones IE, McAuley KA, Manning PJ, Williams SM. Overweight and obese children have low bone mass and area for their weight. Int J Obes Relat Metab Disord. 2000;24(5):627. 151. Rocher E, Chappard C, Jaffre C, Benhamou C, Courteix D. Bone mineral density in prepubertal obese and control children: relation to body weight, lean mass, and fat mass. J Bone Miner Metab. 2008;26(1):73-78. 152. Ellis KJ, Shypailo RJ, Wong WW, Abrams SA. Bone mineral mass in overweight and obese children: Diminished or enhanced? Acta Diabetol. 2003;40 Suppl 1:S274-S277. 153. Weiler HA, Janzen L, Green K, Grabowski J, Seshia MM, Yuen KC. Percent body fat and bone mass in healthy Canadian females 10 to 19 years of age. Bone. 2000;27(2):203-207. 106  154. Clark EM, Ness AR, Tobias JH. Adipose tissue stimulates bone growth in prepubertal children. J Clin Endocrinol Metab. 2006;91(7):2534-2541. 155. Sayers A, Tobias JH. Fat mass exerts a greater effect on cortical bone mass in girls than boys. J Clin Endocrinol Metab. 2010;95(2):699-706. 156. Viljakainen HT, Pekkinen M, Saarnio E, Karp H, Lamberg-Allardt C, Mäkitie O. Dual effect of adipose tissue on bone health during growth. Bone. 2011;48(2):212-217. 157. Wetzsteon RJ, Petit MA, Macdonald HM, Hughes JM, Beck TJ, McKay HA. Bone structure and volumetric BMD in overweight children: A longitudinal study. J Bone Miner Res. 2008;23(12):1946-1953. 158. Wey HE, Binkley TL, Beare TM, Wey CL, Specker BL. Cross-sectional versus longitudinal associations of lean and fat mass with pQCT bone outcomes in children. J Clin Endocrinol Metab. 2011;96(1):106-114. 159. Burrows M, Liu D, McKay H. High-resolution peripheral QCT imaging of bone microstructure in adolescents. Osteoporosis Int. 2010;21(3):515-520. 160. Burrows M, Liu D, Perdios A, Moore S, Mulpuri K, McKay H. Assessing bone microstructure at the distal radius in children and adolescents using HR-pQCT: a methodological pilot study. J Clin Densitom. 2010;13(4):451-455. 161. McKay HA, MacLean L, Petit M, MacKelvie-O'Brien K, Janssen P, Beck T, et al. "Bounce at the Bell": A novel program of short bouts of exercise improves proximal femur bone mass in early pubertal children. Br J Sports Med. 2005;39:521-526. 162. Macdonald HM, Kontulainen SA, Khan KM, McKay HA. Is a school-based physical activity intervention effective for increasing tibial bone strength in boys and girls? J Bone Miner Res. 2007;22(3):434-446. 163. MacKelvie KJ, Petit MA, Khan KM, Beck TJ, McKay HA. Bone mass and structure are enhanced following a 2-year randomized controlled trial of exercise in prepubertal boys. Bone. 2004;34(4):755-764. 164. Tanner JM. Foetus into man: Physical growth from conception to maturity. Cambridge, Mass.: Harvard University Press; 1978. 165. Sherar LB, Baxter-Jones A, Mirwald RL. Limitations to the use of secondary sex characteristics for gender comparisons. Ann Hum Biol. 2004;31(5):586-593. 166. Nottelmann ED, Susman EJ, Dorn LD, Inoff-Germain G, Loriaux DL, Cutler GB, J., et al. Developmental processes in early adolescence. Relations among chronologic age, pubertal stage, height, weight, and serum levels of gonadotropins, sex steroids, and adrenal androgens. J Adolesc Health Care. 1987;8(3):246-260.  107  167. Blake GM, Naeem M, Boutros M. Comparison of effective dose to children and adults from dual X-ray absorptiometry examinations. Bone. 2006;38(6):935-942. 168. Hologic Inc. QDR 4500 Fan Beam X-Ray Bone Densitometer User's Guide. Waltham MA: Hologic Inc.; 2002. 169. Scanco Medical. SOP XtremeCT quality grading suggestions. Basserdorf Switzerland: Scanco Medical; 2009. 170. Scanco Medical. XtremeCT User's Guide. 5.05th ed. Basserdorf Switzerland: Scanco Medical; 2005. 171. Laib A, Rüegsegger P. Calibration of trabecular bone structure measurements of in vivo three-dimensional peripheral quantitative computed tomography with 28-microm-resolution microcomputed tomography. Bone. 1999;24(1):35-39. 172. MacNeil JA, Boyd SK. Improved reproducibility of high-resolution peripheral quantitative computed tomography for measurement of bone quality. Med Eng Phys. 2008;30(6):792799. 173. Nishiyama KK, Macdonald HM, Buie HR, Hanley DA, Boyd SK. Postmenopausal women with osteopenia have higher cortical porosity and thinner cortices at the distal radius and tibia than women with normal aBMD: An in vivo HR-pQCT study. J Bone Miner Res. 2010;25(4):882-890. 174. Müller R, Rüegsegger P. Three-dimensional finite element modelling of non-invasively assessed trabecular bone structures. Med Eng Phys. 1995;17(2):126-133. 175. van Rietbergen B, Weinans H, Huiskes R, Odgaard A. A new method to determine trabecular bone elastic properties and loading using micromechanical finite-element models. J Biomech. 1995;28(1):69-81. 176. Macdonald HM, Nishiyama KK, Kang J, Hanley DA, Boyd SK. Age-related patterns of trabecular and cortical bone loss differ between sexes and skeletal sites: A population-based HR-pQCT study. J Bone Miner Res. 2011;26(1):50-62. 177. Keaveny TM, Bouxsein ML. Theoretical implications of the biomechanical fracture threshold. J Bone Miner Res. 2008;23(10):1541-1547. 178. Chiu J, Robinovitch SN. Prediction of upper extremity impact forces during falls on the outstretched hand. J Biomech. 1998;31(12):1169-1176. 179. Melton LJ, Riggs BL, van Lenthe GH, Achenbach SJ, Müller R, Bouxsein ML, et al. Contribution of in vivo structural measurements and load/strength ratios to the determination of forearm fracture risk in postmenopausal women. J Bone Miner Res. 2007;22(9):1442-1448.  108  180. Burrows M, Baxter-Jones A, Mirwald R, Macdonald H, McKay H. Bone mineral accrual across growth in a mixed-ethnic group of children: Are Asian children disadvantaged from an early age? Calcif Tissue Int. 2009;84(5):366-378. 181. Lettgen B, Hauffa B, Möhlmann C, Jeken C, Reiners C. Bone mineral density in children and adolescents with juvenile diabetes: selective measurement of bone mineral density of trabecular and cortical bone using peripheral quantitative computed tomography. Horm Res. 1995;43(5):173-175. 182. Brandao FR, Vicente EJ, Daltro CH, Sacramento M, Moreira A, Adan L. Bone metabolism is linked to disease duration and metabolic control in type 1 diabetes mellitus. Diabetes Res Clin Pract. 2007;78(3):334-339. 183. Baxter-Jones A, Mirwald RL, McKay HA, Bailey DA. A longitudinal analysis of sex differences in bone mineral accrual in healthy 8-19-year-old boys and girls. Ann Hum Biol. 2003;30(2):160-175. 184. Clarke BL, Khosla S. Female reproductive system and bone. Arch Biochem Biophys. 2010;503(1):118-128. 185. Chevalley T, Bonjour JP, van Rietbergen B, Ferrari S, Rizzoli R. Fractures during childhood and adolescence in healthy boys: Relation with bone mass, microstructure, and strength. J Clin Endocrinol Metab. 2011;96(10):3134-3142. 186. Schoenau E, Neu CM, Mokov E, Wassmer G, Manz F. Influence of puberty on muscle area and cortical bone area of the forearm in boys and girls. J Clin Endocrinol Metab. 2000;85(3):1095-1098. 187. Ferretti JL, Capozza RF, Cointry GR, García SL, Plotkin H, Alvarez Filgueira ML, et al. Gender-related differences in the relationship between densitometric values of whole-body bone mineral content and lean body mass in humans between 2 and 87 years of age. Bone. 1998;22(6):683-690. 188. Ducher G, Daly RM, Hill B, Eser P, Naughton GA, Gravenmaker KJ, et al. Relationship between indices of adiposity obtained by peripheral quantitative computed tomography and dual-energy X-ray absorptiometry in pre-pubertal children. Ann Hum Biol. 2009;36(6):705716. 189. Garnett SP, Högler W, Blades B, Baur LA, Peat J, Lee J, et al. Relation between hormones and body composition, including bone, in prepubertal children. Am J Clin Nutr. 2004;80(4):966-972. 190. Kontulainen SA, Macdonald HM, McKay HA. Change in cortical bone density and its distribution differs between boys and girls during puberty. J Clin Endocrinol Metab. 2006;91(7):2555-2561. 191. Bouxsein ML. Determinants of skeletal fragility. Best Pract Res Clin Rheumatol. 2005;19(6):897-911. 109  192. Lochmüller EM, Miller P, Bürklein D, Wehr U, Rambeck W, Eckstein F. In situ femoral dual-energy X-ray absorptiometry related to ash weight, bone size and density, and its relationship with mechanical failure loads of the proximal femur. Osteoporos Int. 2000;11(4):361-367. 193. Goulding A, Jones IE, Taylor RW, Manning PJ, Williams SM. More broken bones: A 4year double cohort study of young girls with and without distal forearm fractures. J Bone Miner Res. 2000;15(10):2011-2018. 194. Clark EM, Tobias JH, Ness AR. Association between bone density and fractures in children: A systematic review and meta-analysis. Pediatrics. 2006;117(2):e291-e297. 195. Cao JJ, Sun L, Gao H. Diet-induced obesity alters bone remodeling leading to decreased femoral trabecular bone mass in mice. Ann N Y Acad Sci. 2010;1192:292-297. 196. Cao JJ, Gregoire BR, Gao H. High-fat diet decreases cancellous bone mass but has no effect on cortical bone mass in the tibia in mice. Bone. 2009;44(6):1097-1104. 197. Patsch JM, Kiefer FW, Varga P, Pail P, Rauner M, Stupphann D, et al. Increased bone resorption and impaired bone microarchitecture in short-term and extended high-fat diet– induced obesity. Metabolism. 2011;60(2):243-249. 198. Ma H, Turpeinen T, Silvennoinen M, Torvinen S, Rinnankoski-Tuikka R, Kainulainen H, et al. Effects of diet-induced obesity and voluntary wheel running on the microstructure of the murine distal femur. Nutr Metab (Lond). 2011;8(1):1-11. 199. Wachter NJ, Augat P, Krischak GD, Mentzel M, Kinzl L, Claes L. Prediction of cortical bone porosity in vitro by microcomputed tomography. Calcif Tissue Int. 2001;68(1):38-42. 200. Augat P, Schorlemmer S. The role of cortical bone and its microstructure in bone strength. Age Ageing. 2006;35:ii27-ii31. 201. Sasimontonkul S, Bay BK, Pavol MJ. Bone contact forces on the distal tibia during the stance phase of running. J Biomech. 2007;40(15):3503-3509. 202. Calfee RP, Sutter M, Steffen JA, Goldfarb CA. Skeletal and chronological ages in American adolescents: Current findings in skeletal maturation. J Child Orthop. 2010;4(5):467-470. 203. Holzer G, von Skrbensky G, Holzer LA, Pichl W. Hip fractures and the contribution of cortical versus trabecular bone to femoral neck strength. J Bone Miner Res. 2009;24(3):468474.  110  Appendices  Appendix A: Consent Forms Information to families, re-consent and assent forms for participants in the Action Schools! BC cohort.  111  112  113  114  115  116  Information to families, re-consent and assent forms for participants in the Healthy Bones II cohort.  117  118  119  120  Information to families, consent and assent forms for 2009 new recruit participants.  121  122  123  124  125  Appendix B: Questionnaires Girls Tanner Stage  126  127  Boys Tanner Stage  128  Parental Health History Questionnaire (for baseline of all studies)  129  130  131  132  Participant Health History Questionnaire  133  134  135  136  Appendix C: Additional Data Table 10. Univariate regression R values of fat mass, lean mass, height, weight, limb length, age and ethnicity with bone variables at the distal radius and distal tibia for BOYS. Ethnicity is Caucasian compared with Asian. Boys Fat Mass  Lean Mass  Height  Weight  Limb Length  Age  Ethnicity  Failure Load (N)  0.05  0.84 **  0.76 **  0.70 **  0.73 **  0.78 **  0.00  Ultimate Stress (MPa)  0.15  0.64 **  0.51 **  0.57 **  0.52 **  0.64 **  -0.14  Load-to-strength ratio  -0.04  -0.81 **  -0.78 **  -0.67 **  -0.76 **  -0.81 ** 0.00  Tt.Ar (mm2)  Distal Radius  -0.07  0.67 **  0.72 **  0.51 **  0.65 **  0.57 **  0.20 *  3  0.21 *  0.62 **  0.47 **  0.59 **  0.48 **  0.66 **  -0.18 *  3  Ct.BMD (mg HA/cm )  0.18 *  0.65 **  0.56 **  0.60 **  0.55 **  0.79 **  -0.20 *  Ct.Po (%)  -0.14  -0.38 **  -0.33 **  -0.36 **  -0.32 **  -0.53 ** 0.21 *  Ct.Th (mm)  0.20 *  0.69 **  0.56 **  0.64 **  0.56 **  0.71 **  -0.22 *  BV/TV (%)  0.04  0.34 **  0.18 *  0.29 **  0.18 *  0.23 **  0.12  Tb.N (1/mm)  0.12  -0.23 **  -0.37 **  -0.13  -0.35 **  -0.42 ** 0.18 *  Tb.Th (mm)  -0.03  0.48 **  0.41 **  0.37 **  0.40 **  0.53 **  0.01  Tb.Sp (mm)  -0.12  0.09  0.24 **  0.02  0.22 *  0.30 **  -0.19 *  FLoad (N)  0.19 *  0.88 **  0.78 **  0.79 **  0.69 **  0.76 **  -0.06  UStress (MPa)  0.03  0.66 **  0.60 **  0.55 **  0.50 **  0.68 **  -0.14  Tt.BMD (mg HA/cm )  Distal Tibia  Tt.Ar (mm2)  0.29 **  0.66 **  0.61 **  0.65 **  0.59 **  0.45 **  0.10  3  0.05  0.65 **  0.55 **  0.54 **  0.44 **  0.66 **  -0.14  3  Ct.BMD (mg HA/cm )  0.06  0.72 **  0.69 **  0.61 **  0.55 **  0.85 **  -0.18 *  Ct.Po (%)  0.02  -0.21 *  -0.23 **  -0.16  -0.11  -0.37 ** 0.16  Ct.Th (mm)  0.07  0.73 **  0.65 **  0.62 **  0.56 **  0.70 **  -0.14  BV/TV (%)  0.07  0.46 **  0.33 **  0.40 **  0.25 **  0.37 **  -0.06  Tb.N (1/mm)  0.22 **  0.11  -0.08  0.17 *  -0.03  -0.14  0.12  Tb.Th (mm)  -0.13  0.34 **  0.38 **  0.22 **  0.27 **  0.50 **  -0.17 *  Tb.Sp (mm)  -0.21 *  -0.16  0.04  -0.21 *  0.02  0.10  -0.10  Tt.BMD (mg HA/cm )  * p < 0.05; ** p < 0.01 Tt.Ar = total area; Tt.BMD = total bone mineral density; Ct.BMD = cortical bone mineral density; Ct.Po = cortical porosity; Ct.Th = cortical thickness; BV/TV = trabecular bone volume fraction; Tb.N = trabecular number; Tb.Th = trabecular thickness; Tb.Sp = trabecular separation.  137  Table 11. Univariate regression R values of fat mass, lean mass, height, weight, limb length, age and ethnicity with bone variables at the distal radius and distal tibia for GIRLS. Ethnicity is Caucasian compared with Asian. Girls Fat Mass  Lean Mass  Height  Weight  Limb Length  Age  Ethnicity  Failure Load (N)  0.54 **  0.79 **  0.76 **  0.73 *  0.73 **  0.82 **  0.18 *  Ultimate Stress (MPa)  0.35 **  0.52 **  0.50 **  0.48 **  0.49 **  0.66 **  0.00  Load-to-strength ratio  -0.49 **  -0.74 **  -0.71 **  -0.68 **  -0.68 **  -0.77 ** -0.14  2  Distal Radius  Tt.Ar (mm )  0.49 **  0.75 **  0.74 **  0.68 **  0.67 **  0.55 **  0.40 **  3  0.40 **  0.55 **  0.53 **  0.52 **  0.51 **  0.75 **  0.01  3  Ct.BMD (mg HA/cm )  0.53 **  0.70 **  0.69 **  0.67 **  0.63 **  0.93 **  0.06  Ct.Po (%)  -0.46 **  -0.56 **  -0.56 **  -0.55 **  -0.49 **  -0.79 ** -0.09  Ct.Th (mm)  0.51 **  0.66 **  0.64 **  0.63 **  0.61 **  0.81 **  -0.02  BV/TV (%)  -0.10  -0.02  -0.06  -0.06  -0.03  -0.05  0.08  Tb.N (1/mm)  -0.20 *  -0.27 **  -0.32 **  -0.25 **  -0.27 **  -0.39 ** 0.16  Tb.Th (mm)  0.07  0.19 *  0.19 *  0.15  0.19 *  0.26 **  -0.05  Tb.Sp (mm)  0.19 *  0.23 **  0.29 **  0.23 **  0.24 **  0.35 **  -0.16  0.64 **  0.86 **  0.76 **  0.81 **  0.60 **  0.75 **  0.16 *  0.45 **  0.69 **  0.63 **  0.62 **  0.44 **  0.74 **  0.05  0.47 **  0.53 **  0.45 **  0.53 **  0.46 **  0.20 *  0.26 **  Tt.BMD (mg HA/cm )  0.47 **  0.69 **  0.62 **  0.63 **  0.41 **  0.75 **  0.09  Ct.BMD (mg HA/cm3)  0.55 **  0.76 **  0.77 **  0.71 **  0.56 **  0.94 **  0.06  Ct.Po (%)  -0.35 **  -0.41 **  -0.47 **  -0.41 **  -0.32 **  -0.67 ** 0.01  Ct.Th (mm)  0.55 **  0.77 **  0.71 **  0.72 **  0.54 **  0.80 **  0.06  BV/TV (%)  0.24 **  0.38 **  0.26 **  0.34 **  0.11  0.29 **  0.15  Tb.N (1/mm)  0.08  -0.03  -0.12  0.02  -0.05  -0.23 ** 0.30 **  Tb.Th (mm)  0.17 *  0.39 **  0.33 **  0.31 **  0.14  0.45 **  -0.11  Tb.Sp (mm)  -0.10  -0.03  0.07  -0.07  0.03  0.18 *  -0.30 **  Tt.BMD (mg HA/cm )  Distal Tibia FLoad (N) UStress (MPa) 2  Tt.Ar (mm ) 3  * p < 0.05; ** p < 0.01 Tt.Ar = total area; Tt.BMD = total bone mineral density; Ct.BMD = cortical bone mineral density; Ct.Po = cortical porosity; Ct.Th = cortical thickness; BV/TV = trabecular bone volume fraction; Tb.N = trabecular number; Tb.Th = trabecular thickness; Tb.Sp = trabecular separation.  138  Boys Radius  Table 12. Unstandardized multivariable regression model coefficients (95% confidence interval) and R2 for each model at the distal radius for BOYS. Constant (B0)  Age (B1)  Ulna Length (B2)  Ethnicity (B3)  Lean Mass (B4)  Fat Mass (B5)  R2  FLoad  1036 (-104, 2175) p = 0.07  70 (29, 111) p = 0.001  -9 (-15, -3) p = 0.005  -35 (-184, 113) p = 0.64  60 (46, 74) p < 0.001  -23 (-34, -11) p < 0.001  0.77  UStress  30.7 (5.1, 56.3) p = 0.02  1.7 (0.8, 2.6) p < 0.001  -0.2 (-0.3, -0.03) p = 0.02  -3.6 (-6.9, -0.2) p = 0.04  0.6 (0.3, 0.9) p < 0.001  -0.01 (-0.3, 0.3) p = 0.95  0.50  Load: Str  3.35 (2.46, 4.23) p < 0.001  -0.07 (-0.10, -0.04) p < 0.001  0.001 (-0.003, 0.01) p = 0.62  0.03 (-0.09, 0.15) p = 0.62  -0.03 (-0.04, -0.02) p < 0.001  0.01 (0.004, 0.02) p = 0.005  0.75  Tt.Ar  70.6 (-37.6, 178.7) p = 0.20  -1.3 (-5.2, 2.6) p = 0.51  0.2 (-0.3, 0.8) p = 0.40  18.3 (4.2, 32.4) p = 0.01  3.1 (1.8, 4.5) p < 0.001  -2.4 (-3.5, -1.3) p < 0.001  0.56  Tt.BMD  384.3 (236.6, 532.1) p < 0.001  14.2 (8.8, 19.5) p < 0.001  -1.6 (-2.4, -0.8) p < 0.001  -22.7 (-42.0, -3.4) p = 0.02  3.7 (1.8, 5.5) p < 0.001  0.9 (-0.6, 2.4) p = 0.25  0.54  Ct.BMD  673.6 (526.9, 820.4) p < 0.001  25.7 (20.4, 31.0) p < 0.001  -1.5 (-2.3, -0.7) p < 0.001  -33.9 (-53.0, -14.7) p = 0.001  2.2 (0.4, 4.1) p = 0.02  1.4 (-0.04, 2.9) p = 0.06  0.70  Ct.Po  5.2 (0.2, 10.2) p = 0.04  -0.5 (-0.7, -0.3) p < 0.001  0.02 (-0.003, 0.05) p = 0.09  0.8 (0.1, 1.4) p = 0.02  0.003 (-0.06, 0.07) p = 0.92  -0.05 (-0.1, 0.004) p = 0.07  0.36  Ct.Th  0.99 (0.48, 1.50) p < 0.001  0.05 (0.03, 0.07) p < 0.001  -0.005 (-0.01, -0.001) p = 0.001  -0.13 (-0.20, -0.06) p < 0.001  0.01 (0.01, 0.02) p < 0.001  0.002 (-0.003, 0.007) p = 0.46  0.64  BV/TV  0.250 (0.174, 0.326) p < 0.001  -0.0004 (-0.003, 0.002) p = 0.80  -0.001 (-0.001, -0.0003) p = 0.001  0.006 (-0.004, 0.016) p = 0.22  0.002 (0.001, 0.003) p < 0.001  -0.001 (-0.001, 0.0002) p = 0.12  0.22  Tb.N  3.17 (2.55, 3.78) p < 0.001  -0.04 (-0.06, 0.02) p < 0.001  -0.004 (-0.01, -0.001) p = 0.01  0.09 (0.01, 0.17) p = 0.03  0.01 (0.004, 0.02) p = 0.003  0.002 (-0.005, 0.01) p = 0.63  0.28  Tb.Th  0.082 (-0.001, 0.001) p < 0.001  0.002 (0.001, 0.003) p = 0.005  -0.0002 (-0.0004, -0.00002) p = 0.03  0.0002 (-0.004, 0.004) p = 0.91  0.001 (0.0002, 0.001) p = 0.005  -0.0003 (-0.001, 0.001) p = 0.07  0.33  Tb.Sp  0.137 (-0.031, 0.305) p = 0.11  0.011 (0.005, 0.017) p = 0.001  0.001 (0.0004, 0.002) p = 0.005  -0.024 (-0.046, -0.002) p = 0.03  -0.004 (-0.006, -0.002) p < 0.001  0.0004 (-0.002, 0.002) p = 0.96  0.24  FLoad = failure load; UStress = ultimate stress; Load:Str = load-to-strength ratio; Tt.Ar = total area; Tt.BMD = total bone mineral density; Ct.BMD = cortical bone mineral density; Ct.Po = cortical porosity; Ct.Th = cortical thickness; BV/TV = trabecular bone volume fraction; Tb.N = trabecular number; Tb.Th = trabecular thickness; Tb.Sp = trabecular separation; Ethnicity = Caucasian compared with Asian.  139  Boys Tibia  Table 13. Unstandardized multivariable regression model coefficients (95% confidence interval) and R2 for each model at the distal tibia for BOYS. Constant (B0)  Age (B1)  Tibia Length (B2)  Ethnicity (B3)  Lean Mass (B4)  Fat Mass (B5)  R2  FLoad  2158 (86, 4230) p = 0.04  49 (-33, 130) p = 0.24  -5 (-11, 2) p = 0.20  -208 (-516, 99) p = 0.18  125 (99, 151) p < 0.001  -16 (-39, 7) p = 0.17  0.79  UStress  19.1 (2.0, 36.1) p = 0.03  1.0 (0.3, 1.7) p = 0.003  -0.03 (-0.1, 0.02) p = 0.28  -2.3 (-4.9, 0.2) p = 0.07  0.4 (0.2, 0.6) p < 0.001  -0.2 (-0.4, 0.01) p = 0.07  0.52  Tt.Ar  328.9 (95.8, 562.0) p = 0.006  -9.4 (-18.5, -0.2) p = 0.045  0.5 (-0.3, 1.3) p = 0.20  14.2 (-20.3, 48.8) p = 0.42  7.0 (4.0, 10.0) p < 0.001  1.3 (-1.3, 4.0) p = 0.30  0.48  Tt.BMD  281.9 (179.2, 384.7) p < 0.001  5.9 (1.8, 10.0) p = 0.004  -0.5 (-0.8, -0.1) p = 0.009  -10.5 (-25.7, 4.7) p = 0.18  3.0 (1.7, 1.3) p < 0.001  -1.0 (-2.1, 0.1) p = 0.09  0.52  Ct.BMD  580.9 (479.2, 682.6) p < 0.001  19.1 (15.1, 23.1) p < 0.001  -0.3 (-0.7, -0.002) p = 0.048  -20.8 (-35.9, -5.8) p = 0.007  1.6 (0.3, 2.9) p = 0.02  -0.3 (-1.4, 0.8) p = 0.61  0.76  Ct.Po  6.4 (1.6, 11.3) p = 0.01  -0.4 (-0.6, -0.2) p < 0.001  0.01 (-0.01, 0.03) p = 0.22  0.4 (-0.3, 1.2) p = 0.22  0.03 (-0.04, 0.09) p = 0.40  -0.01 (-0.1, 0.04) p = 0.76  0.20  Ct.Th  0.66 (0.11, 1.22) p = 0.02  0.02 (0.002, 0.05) p = 0.03  -0.001 (-0.003, 0.001) p = 0.22  -0.09 (-0.17, -0.01) p = 0.04  0.02 (0.01, 0.03) p < 0.001  -0.01 (-0.01, -0.001) p = 0.03  0.60  BV/TV  0.211 (0.156, 0.266) p < 0.001  -0.0003 (-0.002, 0.002) p = 0.80  -0.0002 (-0.0005, -0.0001) p = 0.003  -0.0004 (-0.009, 0.008) p = 0.93  0.002 (0.001, 0.002) p < 0.001  -0.0004 (-0.001, 0.0002) p = 0.121  0.27  Tb.N  2.85 (2.24, 3.47) p < 0.001  -0.05 (-0.07, -0.03) p < 0.001  -0.003 (-0.005, -0.001) p = 0.006  0.08 (-0.01, 0.17) p = 0.084  0.02 (0.01, 0.03) p < 0.001  0.003 (-0.004, 0.01) p = 0.47  0.22  Tb.Th  0.066 (0.038, 0.094) p < 0.001  0.002 (0.001, 0.003) p < 0.001  -6.84∙10-6 (-0.0001, -0.0001) p = 0.89  -0.004 (-0.008, 0.0001) p = 0.06  -0.0001 (-0.0004, 0.0003) p = 0.75  -0.0002 (-0.001, 0.0001) p = 0.07  0.30  Tb.Sp  0.148 (-0.012, 0.309) p = 0.07  0.014 (0.008, 0.021) p < 0.001  0.001 (0.0004, 0.001) p = 0.001  -0.021 (-0.044, 0.003) p = 0.09  -0.006 (-0.008, -0.004) p < 0.001  -0.0002 (-0.002, 0.002) p = 0.80  0.26  FLoad = failure load; UStress = ultimate stress; Load:Str = load-to-strength ratio; Tt.Ar = total area; Tt.BMD = total bone mineral density; Ct.BMD = cortical bone mineral density; Ct.Po = cortical porosity; Ct.Th = cortical thickness; BV/TV = trabecular bone volume fraction; Tb.N = trabecular number; Tb.Th = trabecular thickness; Tb.Sp = trabecular separation; Ethnicity = Caucasian compared with Asian.  140  Table 14. Unstandardized multivariable regression model coefficients (95% confidence interval) and R2 for each model at the distal radius for GIRLS. Constant (B1) 811 (-113, 1734) p = 0.09  Centered Age (B2) 75 (56, 94) p < 0.01  Centered Age2 (B3) -10 (-15, -4) p < 0.01  Ulna Length (B4) 0 (-4, 5) p = 0.88  Ethnicity (B5) 16 (-87, 119) p = 0.76  Lean Mass (B6) 26 (10, 42) p = 0.001  Fat Mass (B7) -5 (-16, 5) p = 0.34  R2 0.76  UStress  39.3 (12.2, 66.5) p = 0.005  1.9 (1.4, 2.5) p < 0.001  -0.3 (-0.4, -0.1) p < 0.001  -0.01 (-0.1, 0.1) p = 0.90  -1.3 (-4.3, 1.7) p = 0.40  -0.04 (-0.5, 0.4) p = 0.87  0.01 (-0.3, 0.3) p = 0.94  0.51  Load: Strength  2.50 (1.09, 3.92) p = 0.001  -0.10 (-0.12, -0.07) p <0.001  0.02 (0.01, 0.02) p < 0.001  0.001 (-0.01, 0.01) p = 0.81  0.01 (-0.15, 0.17) p = 0.89  -0.04 (-0.06, -0.01) p = 0.004  0.01 (-0.01, 0.02) p = 0.29  0.69  Tt.Ar  31.9 (-46.1, 109.8) p = 0.42  -0.6 (-2.2, 1.0) p = 0.45  0.2 (-0.2, 0.6) p = 0.36  0.2 (-0.2, 0.6) p = 0.37  10.6 (1.9, 19.2) p = 0.017  3.7 (2.3, 5.0) p < 0.001  -0.9 (-1.8, -0.04) p = 0.04  0.61  Ct.Th  0.99 (0.51, 1.46) p < 0.001  0.05 (0.04, 0.06) p < 0.001  -0.005 (-0.01, -0.002) p = 0.001  0.0001 (-0.002, 0.003) p 0.91  -0.07 (-0.12, -0.02) p = 0.007  -0.001 (-0.01, 0.01) p = 0.90  0.004 (-0.001, 0.01) p = 0.11  0.72  BV/TV  0.142 (0.054, 0.230) p = 0.002  -0.001 (-0.002, 0.001) p = 0.55  -0.0001 (-0.001, 0.0004) p = 0.77  -0.0001 (-0.001, 0.0004) p = 0.74  0.004 (-0.005, 0.014) p = 0.37  0.001 (-0.001, 0.002) p = 0.42  -0.001 (-0.002, 0.0004) p = 0.17  0.03  Tb.N  1.85 (1.07, 2.62) p < 0.001  -0.03 (-0.04, -0.01) p = 0.001  0.003 (-0.001, 0.01) p = 0.15  -0.001 (-0.005, 0.003) p = 0.69  0.10 (0.01, 0.18) p = 0.03  0.004 (-0.01, 0.02) p = 0.59  -0.002 (-0.01, 0.01) p = 0.69  0.21  Tb.Th  0.079 (0.045, 0.112) p < 0.001  0.001 (-0.000004, 0.001) p = 0.051  -0.0002 (-0.0004, 0.00001) p = 0.07  -0.00001 (-0.0002, 0.0002) p = 0.87  -0.001 (-0.005, 0.002) p = 0.51  0.0001 (-0.0004, 0.001) p = 0.65  -0.0002 (-0.001, 0.0002) p = 0.29  0.11  Tb.Sp  0.462 (0.236, 0.689) p < 0.001  0.007 (0.003, 0.012) p = 0.002  -0.001 (-0.002, 0.001) p = 0.29  0.0003 (-0.001, 0.001) p = 0.57  -0.029 (-0.054, -0.004) p = 0.03  -0.002 (-0.005, 0.002) p = 0.44  0.001 (-0.002, 0.003) p = 0.51  0.18  Girls Radius  FLoad  Constant (B1)  Centered Age2 (B3) -0.4 (-1.5, 0.6) p = 0.43  Ulna Length (B4)  Ethnicity (B5)  Menarche (B6)  -0.2 (-1.0, 0.6) p = 0.65  -6.9 (-24.6, 10.7) p = 0.44  -0.9 (-1.8, -0.02) p = 0.045  -0.6 (-1.3, 0.1) p = 0.07 0.02 (-0.005, 0.04) p = 0.13  Tt.BMD  374.5 (207.7, 541.3) p < 0.001  Centered Age (B2) 11.2 (5.9, 16.5) p < 0.001  Ct.BMD  932.9 (798.3, 1067.4) p < 0.001  26.3 (21.8, 30.8) p < 0.001  Ct.Po  -2.4 (-6.7, 1.9) p = 0.27  -0.40 (-0.54, -0.26) 0.03 (0.001, 0.06) p < 0.001 p = 0.04  Fat Mass (B8)  R2  50.1 (5.4, 94.8) p = 0.03  Lean Mass (B7) -1.4 (-4.2, 1.4) p = 0.34  0.63 (-1.2, 2.5) p = 0.50  0.60  -10.7 (-25.1, 3.7) p = 0.15  51.1 (13.0, 89.2) p = 0.009  -1.3 (-3.5, 1.0) p = 0.27  1.6 (0.1, 3.0) p = 0.04  0.90  -0.3 (-0.8, 0.1) p = 0.16  -1.5 (-2.7, -0.3) p = 0.02  0.1 (0.02, 0.2) p = 0.02  -0.1 (-0.1, -0.02) p = 0.008  0.70  FLoad = failure load; UStress = ultimate stress; Load:Str = load-to-strength ratio; Tt.Ar = total area; Tt.BMD = total bone mineral density; Ct.BMD = cortical bone mineral density; Ct.Po = cortical porosity; Ct.Th = cortical thickness; BV/TV = trabecular bone volume fraction; Tb.N = trabecular number; Tb.Th = trabecular thickness; Tb.Sp = trabecular separation; Ethnicity = Caucasian compared with Asian.  141  Table 15. Unstandardized multivariable regression model coefficients (95% confidence interval) and R2 for each model at the distal tibia for GIRLS. Constant (B1) 3019 (1390, 4649) p < 0.001  Centered Age (B2) 63 (21, 106) p = 0.004  Centered Age2 (B3) -8 (-19, 4) p = 0.19  Tibia Length (B4) -7 (-13, -2) p = 0.012  Ethnicity (B5) -73 (-305, 158) p = 0.53  Lean Mass (B6) 148 (114, 183) p < 0.001  Fat Mass (B7) -16 (-40, 9) p = 0.20  R2 0.78  UStress  45.2 (30.6, 59.7) p < 0.001  1.2 (0.9, 1.6) p < 0.001  -0.2 (-0.4, -0.1) p < 0.001  -0.1 (-0.1, -0.02) p = 0.003  -0.05 (-2.1, 2.0) p = 0.97  0.6 (0.3, 0.9) p < 0.001  -0.2 (-0.4, 0.02) p = 0.07  0.67  Tt.Ar  151.1 (-6.1, 308.3) p = 0.06  -11.4 (-15.5, -7.3) p < 0.001  2.4 (1.3, 3.6) p < 0.001  0.3 (-0.2, 0.9) p = 0.22  -3.4 (-25.7, 18.9) p = 0.76  9.2 (5.9, 12.6) p < 0.001  0.4 (-1.9, 2.8) p = 0.71  0.47  Ct.Th  1.01 (0.62, 1.41) p < 0.001  0.04 (0.03, 0.05) p < 0.001  -0.005 (-0.01, -0.002) p = 0.001  -0.001 (-0.002, 0.0002) p = 0.10  -0.03 (-0.09, 0.02) p = 0.23  0.02 (0.01, 0.03) p < 0.001  -0.002 (-0.01, 0.004) p = 0.51  0.74  BV/TV  0.228 (0.174, 0.283) p < 0.001  -0.002 (-0.002, 0.001) p = 0.77  -0.0003 (-0.001, 0.0001) p = 0.12  -0.0004 (-0.001, -0.0003) p < 0.001  0.006 (-0.001, 0.014) p = 0.10  0.003 (0.002, 0.004) p < 0.001  -0.001 (-0.002, 0.0001) p = 0.07  0.28  Tb.N  1.56 (0.99, 2.11) p < 0.001  -0.03 (-0.04, -0.01) p < 0.001  0.01 (0.001, 0.01) p = 0.012  -0.001 (-0.003, 0.001) p = 0.38  0.12 (0.04, 0.20) p = 0.003  0.01 (-0.004, 0.02) p = 0.22  0.01 (-0.002, 0.01) p = 0.15  0.24  Tb.Th  0.141 (0.113, 0.169) p < 0.001  0.001 (0.0004, 0.002) p = 0.002  -0.0004 (-0.0006, -0.0002) p < 0.001  -0.0002 (-0.0003, -0.0001) p < 0.001  -0.002 (-0.006, 0.002) p = 0.32  0.001 (0.001, 0.002) p < 0.001  -0.001 (-0.001, -0.0002) p = 0.003  0.42  Tb.Sp  0.513 (0.350, 0.676) p < 0.001  0.008 (0.004, 0.013) p < 0.001  -0.001 (-0.002, -0.0001) p = 0.04  0.0005 (-0.0001, 0.001) p = 0.09  -0.033 (-0.057, -0.010) p = 0.005  -0.004 (-0.008, -0.001) p = 0.02  -0.001 (-0.003, 0.002) p = 0.50  0.22  Centered Age2 (B3) -0.8 (-1.6, -0.1) p = 0.0.03  Tibia Length (B4)  Ethnicity (B5)  Menarche (B6)  -0.6 (-0.9, -0.3) < 0.001  6.9 (-5.8, 19.5) p = 0.29  Girls Tibia  FLoad  Constant (B1)  Fat Mass (B8)  R2  40.3 (8.4, 72.3) p = 0.014  Lean Mass (B7) 3.3 (1.3, 5.2) p = 0.001  -0.8 (-2.1, 0.6) p = 0.25  0.68  Tt.BMD  373.8 (280.9, 466.6) p < 0.001  Centered Age (B2) 4.8 (1.1, 8.5) p = 0.012  Ct.BMD  785.3 (705.1, 865.5) p < 0.001  19.4 (16.2, 22.6) p < 0.001  -1.4 (-2.0, -0.7) p < 0.001  0.03 (-0.2, 0.3) p = 0.78  0.9 (-10.0, 11.8) p = 0.87  69.4 (41.8, 97.0) p < 0.001  -1.3 (-2.9, 0.4) p = 0.14  1.1 (-0.01, 2.3) p = 0.051  0.93  Ct.Po  1.7 (-2.4, 5.7) p = 0.42  -0.4 (-0.6, -0.3) p < 0.001  0.05 (0.02, 0.08) p = 0.003  -0.01 (-0.02, 0.01) p = 0.27  -0.3 (-0.8, 0.3) p = 0.30  -1.6 (-2.9, -0.2) p = 0.03  0.2 (0.1, 0.3) p < 0.001  -0.1 (-0.2, -0.04) p = 0.002  0.58  FLoad = failure load; UStress = ultimate stress; Tt.Ar = total area; Tt.BMD = total bone mineral density; Ct.BMD = cortical bone mineral density; Ct.Po = cortical porosity; Ct.Th = cortical thickness; BV/TV = trabecular bone volume fraction; Tb.N = trabecular number; Tb.Th = trabecular thickness; Tb.Sp = trabecular separation; Ethnicity = Caucasian compared with Asian.  142  

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.24.1-0073104/manifest

Comment

Related Items