Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Effects of a school-based physical activity intervention on adolescent bone strength, structure and density… Tan, Vina Phei Sean 2015

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

Item Metadata

Download

Media
24-ubc_2015_september_tan_vina.pdf [ 6.87MB ]
Metadata
JSON: 24-1.0167282.json
JSON-LD: 24-1.0167282-ld.json
RDF/XML (Pretty): 24-1.0167282-rdf.xml
RDF/JSON: 24-1.0167282-rdf.json
Turtle: 24-1.0167282-turtle.txt
N-Triples: 24-1.0167282-rdf-ntriples.txt
Original Record: 24-1.0167282-source.json
Full Text
24-1.0167282-fulltext.txt
Citation
24-1.0167282.ris

Full Text

EFFECTS OF A SCHOOL-BASED PHYSICAL ACTIVITY INTERVENTION ON ADOLESCENT BONE STRENGTH, STRUCTURE AND DENSITY: THE HEALTH PROMOTING SECONDARY SCHOOLS (HPSS) BONE HEALTH STUDY  by  VINA PHEI SEAN TAN MSc., Universiti Sains Malaysia, 2005   A Thesis Submitted in Partial Fulfillment of the Requirements for the Degree of  DOCTOR OF PHILOSOPHY  in  The Faculty of Graduate and Postdoctoral Studies (Experimental Medicine)   THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  June 2015 © Vina Phei Sean Tan, 2015ii  Abstract  Physical activity (PA) benefits bone strength in children but little is known of the effects of PA on bone strength in adolescents. In this thesis, my primary aim was to determine the effect of a secondary school based PA intervention on bone strength, structure and density in adolescents.  This 8-month cluster, randomized-controlled, whole school-based intervention study had four intervention and five control schools. Participants were 210 Grade 10 students who were 15.3 years old, on average, at baseline. The Health Promoting Secondary Schools (HPSS) intervention was a choice-based model based on self-determination theory that aimed to increase PA, promote healthy eating and reduce screen time in adolescents. I used peripheral quantitative computed tomography (pQCT) to assess bone strength, structure and density at the distal and shaft sites of the tibia and radius. I assessed PA using a validated PA self-report questionnaire and I measured a sub-set of participants’ PA objectively using accelerometry.  Part I is a systematic review and narrative synthesis of PA and pediatric bone literature. High-quality randomized-controlled trials (RCTs) with weight-bearing PA increased bone strength in children. Bone structure adaptations in response to PA were more common than adaptations in bone density (RCTs and observational studies). Only one RCT involved adolescents (average age 13.8 years) and studies often overlooked the influence of muscle on bone responses to PA. In Part II, moderate-to-vigorous PA (MVPA), vigorous PA (VPA) and grip strength positively influenced bone strength in boys and girls after controlling for ethnicity, maturity, limb length and muscle mass. Sedentary time (SED) negated the positive influence of MVPA, but not VPA, on bone strength in girls.  iii  In Part III, the HPSS intervention did not lead to significant gains in bone strength, structure or density in adolescents. The external factor of a province-wide teacher job action possibly hindered the execution of the HPSS intervention.  In summary, MVPA and VPA benefit bone strength in adolescents but further investigations are warranted to determine the effects of SED on bone strength. It remains to be determined the effects of a choice-based intervention on bone strength adaptations in adolescent boys and girls.   iv  Preface The Health Promoting Secondary Schools (HPSS) study was conceived and designed by Professor Joan Wharf Higgins and Associate Professor Patti-Jean Naylor (University of Victoria; UVic) in collaboration with Professor Heather McKay (University of British Columbia; UBC). The HPSS Bone Health Study (BHS) was primarily designed by Professor Heather McKay and Assistant Professor Heather Macdonald (UBC) while I provided input on aspects of study design, methods and targeted outcomes. Both studies received ethical approval from the UBC Behavioural Research Ethics Board (H10-01917) and UVic Ethics Board (10-168).  My role was to coordinate the BHS from planning to execution. Thus, I organized and led data collection across three regions of British Columbia – Greater Vancouver, Interior and Vancouver Island, during both 6-week data collection periods. Prior to data collection, I trained and supervised research staff on all research protocols with the exception of bone imaging training. I acquired and analysed all of the dual-energy X-ray absorptiometry (DXA) scans, analysed all of the peripheral quantitative computed tomography (pQCT) scans and completed all anthropometry measurements. I entered and cleaned the data (except accelerometry data) and conducted all of the statistical data analysis for this thesis. A version of Chapter 3 is published in the Journal of Bone and Mineral Research (Tan VPS, Macdonald HM, Kim SJ, Nettlefold L, Gabel L, Ashe M and McKay HA. The influence of physical activity on bone strength in children and adolescents: A systematic review and narrative synthesis. J Bone Miner Res, 2014, 29(10):2161-2181), the top peer-reviewed journal in our field. I was lead author on the review and collaborated with Professor McKay, Associate Professor Maureen Ashe and Dr. SoJung Kim (post-doctoral fellow) to design, conduct and write-up the systematic review. Professor McKay conceived the idea. I conducted the literature searches with guidance from information specialist Madeline Doyle-Waters. I extracted and compiled data from studies and conducted assessments of study quality with Dr. v  Kim. I synthesized the data and wrote the initial draft of the systematic review. Assist. Prof Macdonald and Professor McKay provided detailed edits and feedback. Assoc. Professor Ashe, Dr. Lindsay Nettlefold and Leigh Gabel (doctoral candidate) also provided feedback on near final drafts of the systematic review. I am currently preparing a version of Chapter 4 for submission, title: Sedentary time negates the positive influence of moderate-to-vigorous physical activity but not vigorous physical activity on bone strength in adolescent girls. I defined the research question, entered and cleaned the data and conducted the statistical analyses. Douglas Race (HPSS accelerometer coordinator) cleaned, uploaded and processed the accelerometer data. Professor McKay and Dr. Macdonald provided guidance on the focus of the paper and feedback on all versions of the manuscript. For Chapter 5, I defined the research question, entered, cleaned and compiled all the data, with exception for accelerometry data where Douglas Race uploaded and processed the accelerometry data. I and conducted the statistical analyses. I wrote the draft for the chapter. Dr. Macdonald and Professor McKay provided feedback and guidance for every version of the chapter.   vi  Table of Contents Abstract ................................................................................................................................................ ii Preface..................................................................................................................................................iv Table of Contents ..................................................................................................................................vi List of Tables ..................................................................................................................................... xvii List of Figures .................................................................................................................................... xxi List of Abbreviations ....................................................................................................................... xxvii Acknowledgements ............................................................................................................................ xxx Dedication ........................................................................................................................................ xxxi Chapter 1: Introduction, Literature Review and Research Hypotheses .................................................... 1 1.1 Introduction ..................................................................................................................................... 1 1.2 Literature Review ........................................................................................................................ 5 1.2.1 Bone anatomy and physiology .............................................................................................. 5 1.2.1.1 Whole bone composition and structure .................................................................... 6 1.2.1.2 Physiology of bone growth, development and maintenance ................................... 10 1.2.2 Bone biomechanics ............................................................................................................. 14 1.2.2.1 Biomechanical properties of bone .......................................................................... 14 1.2.2.2 Mechanostat theory and mechanotransduction ....................................................... 17 1.2.2.3 Principles governing bone adaptation to mechanical stimuli................................... 19 vii  1.2.2.4 Differences in bone adaptation .............................................................................. 22 1.2.3 Bone imaging ..................................................................................................................... 26 1.2.3.1 Dual-energy X-ray absorptiometry (DXA) ............................................................ 26 1.2.3.2 Peripheral quantitative computed tomography (pQCT) .......................................... 29 1.2.3.3 Strengths and limitations of pQCT ........................................................................ 35 1.2.3.4 Other bone imaging technologies .......................................................................... 39 1.2.4 Growth and maturation ....................................................................................................... 41 1.2.4.1 Definition of adolescence ...................................................................................... 41 1.2.4.2 Assessment of puberty........................................................................................... 41 1.2.5 Maturity- and sex-specific differences in bone .................................................................... 48 1.2.5.1 Sex differences in bone mass ................................................................................. 49 1.2.5.2 Sex differences in bone size and strength ............................................................... 49 1.2.5.3 Sex differences in bone at weight-bearing bone sites ............................................. 50 1.2.5.4 Sex differences in bone at non-weight bearing bone sites ....................................... 52 1.2.6 Factors that influence bone growth and development .......................................................... 54 1.2.6.1 Genetics ................................................................................................................ 54 1.2.6.2 Ethnicity and race.................................................................................................. 56 1.2.6.3 Hormones ............................................................................................................. 58 1.2.6.4 Dietary calcium ..................................................................................................... 62 viii  1.2.6.5 Muscle mass and force .......................................................................................... 65 1.2.7 Physical activity and sedentary behaviour ........................................................................... 67 1.2.7.1 Physical activity - definition, concept and components .......................................... 68 1.2.7.2 Measurement of PA............................................................................................... 68 1.2.7.3 Sedentary behaviour – definition and concept ........................................................ 72 1.2.8 Studies of PA and bone strength in children and adolescents ............................................... 73 1.2.8.1 The influence of PA on bone health ....................................................................... 74 1.2.8.2 Association between objectively measured PA and bone outcomes ........................ 77 1.2.8.3 School-based intervention studies on PA and bone health ...................................... 78 1.2.9 Summary ............................................................................................................................ 81 1.3 Rationale, Objectives and Hypotheses ........................................................................................ 82 1.3.1 Part I: A systematic review of the effects of physical activity on bone strength and structure in children and adolescents ................................................................................................. 82 1.3.2 Part II: Determinants of bone strength and structure in adolescent boys and girls ................ 83 1.3.3 Part III: Effect of a secondary school physical activity intervention on bone strength and structure in adolescents. ...................................................................................................... 84 Chapter 2: Methods ............................................................................................................................. 86 2.1 Part I - Systematic Review Protocol ........................................................................................... 86 2.1.1 Search strategy ................................................................................................................... 87 ix  2.1.2 Quality assessment ............................................................................................................. 89 2.1.3 Narrative synthesis ............................................................................................................. 91 2.2 Part II – Health Promoting Secondary Schools ........................................................................... 92 2.2.1 Development of the HPSS model ........................................................................................ 93 2.2.2 Implementation of the HPSS model .................................................................................... 94 2.2.3 HPSS organizational structure ............................................................................................ 96 2.3 HPSS BHS ................................................................................................................................ 98 2.3.1 Study design ....................................................................................................................... 98 2.3.2 Sample size calculation....................................................................................................... 98 2.3.3 Recruitment ........................................................................................................................ 99 2.3.3.1 School and teacher recruitment .............................................................................. 99 2.3.3.2 Student recruitment ............................................................................................. 101 2.3.4 Data collection ................................................................................................................. 102 2.3.4.1 Measurement training .......................................................................................... 102 2.3.4.2 School-site measurement ..................................................................................... 105 2.3.4.3 Anthropometry and body composition ................................................................. 107 2.3.4.4 Bone strength, structure and density .................................................................... 109 2.3.4.5 Physical activity .................................................................................................. 114 2.3.4.6 Muscle strength ................................................................................................... 116 x  2.3.4.7 Maturity .............................................................................................................. 117 2.3.4.8 Health history...................................................................................................... 118 2.3.4.9 Dietary calcium intake......................................................................................... 119 2.3.5 Statistical analysis ............................................................................................................ 119 Chapter 3: The Influence of Physical Activity on Bone Strength, Structure and Density in Children and Adolescents: A Systematic Review and Narrative Synthesis .............................................................. 120 3.1 Introduction ............................................................................................................................. 120 3.2 Methods ................................................................................................................................... 123 3.2.1 Search strategy and resources ........................................................................................... 123 3.2.1.1 Working definitions ............................................................................................ 124 3.2.2 Study selection and inclusion criteria ................................................................................ 126 3.2.3 Assessment and data extraction ......................................................................................... 126 3.2.4 Data synthesis and analysis ............................................................................................... 127 3.3 Results…………………………………………………………………………………...………129 3.3.1 Search results ................................................................................................................... 129 3.3.2 Assessing studies based on ‘quality’ ................................................................................. 129 3.3.2.1 Characteristics of intervention and observational studies ..................................... 132 3.3.3 Differences in research methods across studies ................................................................. 135 3.3.4 Influence of PA on bone strength, mass and structure........................................................ 155 xi  3.3.4.1 Effects of PA on bone strength – Intervention studies .......................................... 155 3.3.4.2 Effects of PA on bone mass and structure – Intervention studies .......................... 156 3.3.4.3 Associations between recreational PA and bone health ........................................ 157 3.3.4.4 Associations between recreational PA and bone mass and structure ..................... 158 3.3.4.5 Association between organized sports participation and bone strength ................. 159 3.3.4.6 Association between organized sports participation and bone mass and structure . 160 3.4 Discussion ............................................................................................................................... 161 3.4.1 Weight-bearing PA enhanced bone strength in children and adolescents ........................... 162 3.4.2 Changes in bone structure rather than bone mass most often accompanied gains in bone strength ............................................................................................................................ 164 3.4.3 PA-related adaptations in bone structure and strength were related to maturity .................. 166 3.4.4 PA-related adaptations in bone structure and strength were sex related -- but generally PA was associated with improved bone strength in both boys and girls ................................... 168 3.4.5 The contribution of muscle to bone strength ...................................................................... 168 3.4.6 PA prescription for bone strength...................................................................................... 169 3.4.7 Limitations to conducting our systematic review ............................................................... 170 3.5 Conclusions ............................................................................................................................. 171 Chapter 4: The Influence of Physical Activity, Sedentary Time and Muscle Strength on Bone Strength, Structure and Density in Older Adolescents....................................................................................... 173 4.1 Introduction ............................................................................................................................. 173 xii  4.2 Methods ................................................................................................................................... 176 4.2.1 Participants ....................................................................................................................... 176 4.2.2 Measurements .................................................................................................................. 178 4.2.2.1 Anthropometry .................................................................................................... 178 4.2.2.2 Bone image acquisition ....................................................................................... 178 4.2.2.3 Bone image analysis ............................................................................................ 178 4.2.2.4 Physical activity .................................................................................................. 179 4.2.2.5 Body composition ............................................................................................... 179 4.2.2.6 Muscle outcomes ................................................................................................ 180 4.2.2.7 Health history, dietary calcium intake and maturity ............................................. 181 4.2.3 Statistical analysis ............................................................................................................ 181 4.3 Results…………………………………………………………………………………………...183 4.3.1 Descriptive characteristics of participants ......................................................................... 183 4.3.2 Sex differences in bone outcomes ..................................................................................... 184 4.3.3 Determinants of tibial bone strength, structure and density in boys and girls ..................... 188 4.3.4 Determinants of radial bone strength, structure and density in boys and girls ..................... 194 4.4 Discussion ............................................................................................................................... 200 4.4.1 Sex differences in bone strength, structure and density ...................................................... 200 4.4.2 Influence of MVPA, VPA and SED on tibial bone strength, structure and density ............. 204 xiii  4.4.3 Independent contribution of muscle strength to radial bone strength and structure ............. 209 4.4.4 Strengths and limitations .................................................................................................. 210 4.5 Conclusion............................................................................................................................... 211 Chapter 5: Effect of the Health Promoting Secondary School (HPSS) Intervention on Bone Strength, Structure and Density in Adolescent Boys and Girls. ......................................................................... 213 5.1 Introduction ............................................................................................................................. 213 5.2 Methods ................................................................................................................................... 217 5.2.1 HPSS Intervention ............................................................................................................ 218 5.2.2 Descriptive variables ........................................................................................................ 220 5.2.3 Bone outcomes ................................................................................................................. 221 5.2.4 Statistical analysis ............................................................................................................ 222 5.3 Results…………………………………………………………………………………………..224 5.3.1 Participants ....................................................................................................................... 224 5.3.2 Descriptive variables ........................................................................................................ 227 5.3.3 Primary outcomes ............................................................................................................. 232 5.3.3.1 Boys ................................................................................................................... 232 5.3.3.2 Girls .................................................................................................................... 232 5.3.4 Secondary outcomes ......................................................................................................... 237 5.3.4.1 Boys ................................................................................................................... 237 xiv  5.3.4.2 Girls .................................................................................................................... 238 5.4 Discussion ............................................................................................................................... 239 5.4.1 HPSS: a choice-based, theory-driven intervention ............................................................. 239 5.4.2 Changes in bone strength, structure and density ................................................................ 242 5.4.2.1 Boys ................................................................................................................... 242 5.4.2.2 Girls .................................................................................................................... 245 5.4.3 Changes in PA levels ........................................................................................................ 247 5.4.4 Strengths and Limitations ................................................................................................. 250 5.5 Conclusion............................................................................................................................... 251 Chapter 6: Integrated Discussion ....................................................................................................... 252 6.1 Overview of Findings .............................................................................................................. 252 6.1.1 Systematic review of studies on PA and bone strength in children and adolescents ............ 252 6.1.2 Determinants of bone strength in adolescents .................................................................... 253 6.1.3 Effects of the HPSS intervention on bone strength, structure and density in adolescents .... 255 6.2 Challenges Associated with the Study of the PA-Bone Strength Relationship in Adolescents ... 257 6.2.1 pQCT imaging .................................................................................................................. 258 6.2.2 Assessment of maturity..................................................................................................... 260 6.2.3 PA assessment .................................................................................................................. 261 6.2.4 Muscle-bone relationship .................................................................................................. 263 xv  6.3 Challenges Associated with School-Based Interventions (Community Trials) ........................... 264 6.4 Public Health Implications of Adolescence PA and Sedentary Time on Bone Health ................ 266 6.5 Summary and Conclusions ....................................................................................................... 266 6.5.1 Part I: Influence of PA on bone strength in children and adolescents: A systematic review and narrative synthesis ...................................................................................................... 266 6.5.2 Part II: Determinants of bone strength, structure and density in adolescent boys and girls . 268 6.5.3 Part III: Effect of a school-based, community-trial intervention on bone strength in adolescents ....................................................................................................................... 270 Bibliography ...................................................................................................................................... 273 Appendices ........................................................................................................................................ 312 Appendix A: Effective Public Health Practice Project Tool & Dictionary ........................................... 313 Appendix B: HPSS Schools with Geographical and Socioeconomic Factors ....................................... 322 Appendix C: HPSS BHS Participant Information and Consent Package.............................................. 324 Appendix D: Dual Energy X-ray Absorptiometry Screening Form ..................................................... 331 Appendix E: pQCT Screening Form ................................................................................................... 333 Appendix F: Physical Activity Questionnaire for Adolescents (PAQ-A) ............................................. 335 Appendix G: Girl’s Maturity Questionnaire........................................................................................ 340 Appendix H: Boy’s Maturity Questionnaire ....................................................................................... 342 Appendix I: Health History Questionnaire (HHQ) .............................................................................. 344 xvi  Appendix J: Food Frequency Questionnaire (FFQ) ............................................................................. 350 Appendix K: Systematic Review Search Terms .................................................................................. 354 Appendix L: Additional Data for Chapter 4 ........................................................................................ 358 Appendix M: Additional Data for Chapter 5 ....................................................................................... 364   xvii  List of Tables Table 1.1. General overview of bone modeling and remodeling. Reproduced from Allen and Burr (62), with permission from Academic Press. .......................................................................................... 11 Table 1.2. Peripheral quantitative computed tomography outcomes, analysis sites and modes and description of bone outcomes. Adapted from Macdonald (161), with permission. .......................... 37 Table 2.1. Inclusion and exclusion criteria for systematic review studies. ............................................... 89 Table 2.2. Participant inclusion and exclusion criteria for Health Promoting Secondary Schools (HPSS) Bone Health Study (BHS). .......................................................................................................... 101 Table 2.3 Analysis modes, thresholds and outcome variables for pQCT assessments at the radius (7% and 30% sites) and tibia (8% and 50% sites). ..................................................................................... 114 Table 3.1. Quality assessment of intervention studies and observational studies of recreational physical activity (PA) and organized sport (n=37). .................................................................................... 131 Table 3.2. Randomized controlled trials and controlled trials (n = 14) of physical activity effects on bone strength, structure and mass in children and adolescents. Ratings are indicated as: *WEAK, **MODERATE, ***STRONG. .................................................................................................. 136 Table 3.3. Observational studies (n=10) of the relationship between recreational physical activity and bone strength, structure and mass in children and adolescents. Ratings are indicated as: *WEAK, **MODERATE and ***STRONG .............................................................................................. 141 Table 3.4. Observational studies (n=13) of the relationship between participation in organized sports and bone strength, structure and mass in children and adolescents. Ratings are indicated as: *WEAK, **MODERATE and ***STRONG. ............................................................................................. 146 Table 3.5. Bone mass and structure outcomes from studies with positive bone strength (n=26) grouped into weight-bearing physical activity (PA) and mixed type of PA (Mixed PA) studies. ................ 165 Table 4.1. Descriptive characteristics, physical performance measures and physical activity levels for boys and girls (mean differences; 95% CI)........................................................................................... 185 Table 4.2. Summary statistics for tibia and radius bone strength, structure and density in boys and girls and mean difference (boys – girls) unadjusted and adjusted for limb length and total body lean mass. ................................................................................................................................................... 186 xviii  Table 4.3. Hierarchical multivariable regression models to demonstrate the independent contribution of moderate-to-vigorous physical activity (MVPA) to bone strength, structure and density at the distal tibia (8% site) and tibial shaft (50% site) in boys and girls (controlled for ethnicity, tibial length, age (boys), maturity (girls) and lean mass). ........................................................................................ 190 Table 4.4. Hierarchical multivariable regression models that show the contribution of moderate physical activity (MPA), vigorous PA (VPA) and sedentary (SED) time to bone strength and density at the distal tibia (8% site) in girls. ........................................................................................................ 194 Table 4.5. Hierarchical multivariable regression models to demonstrate the independent contribution of grip strength to bone strength, structure and density at the distal radius (7% site) and radial shaft (30% site) in boys and girls ......................................................................................................... 197 Table 5.1 Boys’ baseline and follow-up mean (SD) values for age, maturity, anthropometry, body composition, muscle cross-sectional area (MCSA) with within-group absolute change (follow-up - baseline) with 95% confidence intervals (CI) and averaged (baseline and follow-up) dietary calcium and Physical Activity Questionnaire for Adolescents (PAQ-A) outcomes in Intervention and Control groups. ........................................................................................................................................ 229 Table 5.2. Girls’ baseline and follow-up mean (SD) values for age, age at menarche, anthropometry, body composition, muscle cross-sectional area (MCSA) with within-group absolute change (follow-up - baseline) with 95% confidence intervals (CI) and averaged (baseline and follow-up) dietary calcium and Physical Activity Questionnaire for Adolescents (PAQ-A) outcomes in Intervention and Control groups. ........................................................................................................................................ 230 Table 5.3. Differences in unadjusted 30-week change between groups (Intervention - Control) with 95% confidence intervals (CI) and p-value for age, anthropometry, grip strength, body composition, and physical activity by accelerometer outcomes in boys and girls. .................................................... 231 Table 5.4. Boys’ baseline and follow-up mean (SD) values for tibial and radius bone strength, structure and density by Intervention and Control groups and within-group change (follow-up - baseline) with 95% confidence intervals (CI). .................................................................................................... 233 Table 5.5. Differences in change of tibial and radius bone outcomes between groups (Intervention - Control) in boys with adjusted variance estimates for 95% confidence intervals (CI) and p-value based on the interclass correlation (ICC) values. .......................................................................... 234 Table 5.6. Girls’ baseline and follow-up mean (SD) values for tibial and radius bone strength, structure and density by Intervention and Control groups and within-group change (follow-up - baseline) with 95% confidence intervals (CI). .................................................................................................... 235 xix  Table 5.7. Differences in change of tibia and radius bone outcomes between groups (Intervention - Control) in girls with adjusted variance estimates for 95% confidence intervals (CI) and p-value based on the interclass correlation (ICC) values. .......................................................................... 236 Table B.1. Schools matched based on socioeconomic and geographical setting and randomized into intervention (INT) or control (CON) schools ............................................................................... 320 Table L.1. Baseline descriptive variables of participants with and without accelerometry data in boys and girls, reported as mean (SD). ....................................................................................................... 357 Table L.2. Baseline correlations of bone and dependent variables in boys (n=89) and girls (n=109) at the 8% tibia. ..................................................................................................................................... 358 Table L.3. Baseline correlations of bone and dependent variables in boys (n=89) and girls (n=109) at the 50% tibia .................................................................................................................................... 359 Table L.4. Baseline correlations of bone and dependent variables in boys (n=88) and girls (n=110) at the 7% radius. ................................................................................................................................... 360 Table L.5. Baseline correlations of bone and dependent variables in boys (n=89) and girls (n=111) at the 30% radius .................................................................................................................................. 361 Table M.1. Boys’ (n=76) Pearson correlations of descriptive variables (baseline and 30-week changes) – age, maturity-offset, height, weight, tibial length, lean and fat mass, muscle cross-sectional area (MCSA) at 50% tibial site, 30-week average physical activity questionnaire for adolescents (PAQ-A) score, PAQ-A moderate-to-vigorous physical activity (MVPA) and dietary calcium with change in bone strength index (BSI), total bone area (Tt.Ar) and total bone density (Tt.Dn) at the distal tibia (8% site, n=75) and polar stress-strain index (SSIp), total bone area (Tt.Ar), cortical bone area (Ct.Ar), medullary area (Me.Ar) and cortical bone density (Ct.Dn) at the tibial shaft (50%, n=75) and bivariate correlations (using Spearman’s rank correlation) with ethnic grouping and axillary hair stage ........................................................................................................................................... 363 Table M.2. Boys’ (n=76) Pearson correlations of descriptive variables (baseline and 30-week changes) – age, maturity-offset, height, weight, tibial length, lean and fat mass, muscle cross-sectional area (MCSA) at 30% radius site, 30-week average physical activity questionnaire for adolescents (PAQ-A) score, PAQ-A moderate-to-vigorous physical activity (MVPA) and dietary calcium with change in bone strength index (BSI), total bone area (Tt.Ar) and total bone density (Tt.Dn) at the distal radius (7% site, n=74) and polar stress-strain index (SSIp), total bone area (Tt.Ar), cortical bone area (Ct.Ar), medullary area (Me.Ar) and cortical bone density (Ct.Dn) at the radius shaft (50%, n=76) and bivariate correlations (using Spearman’s rank correlation) with ethnic grouping and axillary hair stage. Pearson correlations of anthropometric, body composition, grip strength with baseline values and 30-week change, average of physical activity questionnaire for adolescents (PAQ-A) scores and xx  minutes of moderate-to-vigorous physical activity (MVPA), average dietary calcium and accelerometer changes in girls (n=107). ....................................................................................... 364 Table M.3. Pearson correlations of anthropometric, body composition, grip strength with baseline values and 30-week change, average of physical activity questionnaire for adolescents (PAQ-A) scores and minutes of moderate-to-vigorous physical activity (MVPA), average dietary calcium and accelerometer changes in boys (n=76) ......................................................................................... 365 Table M.4. Girls’ (n=107) Pearson correlations of descriptive variables (baseline and 30-week changes) – age, age at menarche, maturity-offset, height, weight, tibial length, lean and fat mass, muscle cross-sectional area (MCSA) at 50% tibial site, 30-week average physical activity questionnaire for adolescents (PAQ-A) score, PAQ-A moderate-to-vigorous physical activity (MVPA) and dietary calcium with change in bone strength index (BSI), total bone area (Tt.Ar) and total bone density (Tt.Dn) at the distal tibia (8% site, n=104) and polar stress-strain index (SSIp), total bone area (Tt.Ar), cortical bone area (Ct.Ar), medullary area (Me.Ar) and cortical bone density (Ct.Dn) at the tibial shaft (50%, n=104) and bivariate correlations (using Spearman’s rank correlation) with ethnic grouping. .................................................................................................................................... 366 Table M.5. Girls’ (n=107) Pearson correlations of descriptive variables (baseline and 30-week changes) – age, age at menarche, maturity-offset, height, weight, tibial length, lean and fat mass, muscle cross-sectional area (MCSA) at 30% radius site, 30-week average physical activity questionnaire for adolescents (PAQ-A) score, PAQ-A moderate-to-vigorous physical activity (MVPA) and dietary calcium with change in bone strength index (BSI), total bone area (Tt.Ar) and total bone density (Tt.Dn) at the distal radius (7% site, n=101) and polar stress-strain index (SSIp), total bone area (Tt.Ar), cortical bone area (Ct.Ar), medullary area (Me.Ar) and cortical bone density (Ct.Dn) at the radius shaft (30%, n=103) and bivariate correlations (using Spearman’s rank correlation) with ethnic grouping ..................................................................................................................................... 367 Table M.6. Pearson correlations of anthropometric, body composition, grip strength with baseline values and 30-week change, average of physical activity questionnaire for adolescents (PAQ-A) scores and minutes of moderate-to-vigorous physical activity (MVPA), average dietary calcium and accelerometer changes in girls (n=107). ....................................................................................... 368xxi  List of Figures Figure 1.1. Long bone structure. Adapted with permission from the National Institute of Health, through personal communication. ................................................................................................................. 7 Figure 1.2. Structural elements in cortical (compact) and trabecular (spongy) bone. Osteons shaped from lamellar sheets surround blood vessel canals creating the Haversian system. Adapted with permission from the U.S. National Institute of Health. ..................................................................... 8 Figure 1.3. The importance of bone architecture, independent of the amount of material (i.e. mass). Trabecular bone in which the cross struts are disconnected (C) cannot support the same load as trabecular bone with connected cross struts (A), despite similar bone mass. A smaller bone mass that is buttressed properly (B) may be able to support more load than a greater mass in which the connectivity is compromised (C). Figures indicate maximum load. Reproduced from Burr and Akkus (51) with permission from Elsevier. ...................................................................................... 9 Figure 1.4. During dynamic growth, the growth plate contributes to linear growth (I) while resorption around the metaphyseal periosteum occurs to keep the bone shape (II). Resorption of trabeculae shifts the metaphysis-diaphysis border (III). Bone is added periosteally (IV) and resorbed endosteally (V) to increase bone width. Images reproduced from Rauch et al. (64), open access. ... 12 Figure 1.5. From the resting bone stage, bone remodeling is triggered by signals from osteocytes, which recruit preosteoclasts to start osteoclastogenesis where mature osteoclasts are formed to resorb bone. Next, preosteoblasts are recruited in the reversal stage to form mature osteoblasts. Osteoblasts then deposit bone matrix from the cement line (red line) to fill the lacuna. Finally, bone is mineralized by osteoblasts and subsequently osteoblasts undergo apoptosis, are entombed into osteocytes or transforms into lining cells. Reproduced from Kapinas and Delany 2011 (65), open access............ 13 Figure 1.6. Load-deformation curve for whole bone properties. The slope (orange line) of the curve represents the extrinsic stiffness (S); the height of the curve denotes ultimate force (Fu); the yellow triangle is the yield point; the red diamond is the failure point; the area under the curve represents A, elastic region and B, plastic region with the combination of A+B equal to the work to failure; and the total deformation to fracture is ultimate deformation (du). Adapted from Turner (70) with permission from New York Academy of Sciences. ........................................................................ 15 Figure 1.7. Stress-strain curve for bone tissue. The linear part of the curve represents the intrinsic stiffness (Young’s modulus (E)). The height of the curve denotes ultimate strength. The yellow triangle is the yield point; the red diamond is the failure point. Strains above the yield point cause permanent damage to bone structure. Post-yield strain before fracture is equivalent to ductility (converse to brittleness). Adapted from Turner (70) with permission from New York Academy of Sciences...... 16 xxii  Figure 1.8. Illustration of the mechanostat theory in bone modeling and remodeling. Bone remodeling occurs in the upper limit of disuse window (DW); MESm – minimal effective strain for bone modeling occurs between the adaptive window (AW) and the mild overload window (MOW); MESp – minimal effective strain for microdamage repair is at the lower limit of the pathologic overload window (POW); Fx: bone fracture strain and με, microstrain. Adapted from Frost (77) with permission from A.R. Liss (etc). .................................................................................................... 18 Figure 1.9. Bone adaptations to loading in a cross section sample of the ulnar midshaft of a rodent. The schematic shows the strain distributions (upper panel) and bone formed (lower panel) after ulnar loading. Mechanical loading of the right ulnar, 3 days/week for 16 weeks, resulted in apposition of new bone predominantly on the periosteal surface. New bone formed as a result of the experiment can be visualized (orange area) at the outer medio-lateral bone surface. Adapted from Robling et al. (79) with permission from Academic Press. ................................................................................... 25 Figure 1.10. Schematic of the tibial midshaft that shows radial plots of periosteal and endosteal surface changes (left) and cortical thickness changes (right), measured by peripheral quantitative computed tomography. The space between the dotted and solid lines is the difference between pre-pubertal boys randomly assigned to intervention (20 months of 15 min/week extra classroom physical activity and bone loading program) or control groups. Adapted from Macdonald et al. (113), with permission from Springer-Verlag London...................................................................................... 25 Figure 1.11. Effects of size on bone parameters assessed by dual-energy X-ray absorptiometry (DXA). Both samples have identical volumetric density but the areal density of the 3x3x3 specimen is greater because of the projected area captured by DXA. Adapted from Carter et al. (126), with permission from Mary Ann Liebert, Inc. ........................................................................................ 28 Figure 1.12. Image of peripheral quantitative computed tomography (pQCT), model XCT 3000 (Stratec Medizintechnik GmbH) and illustration of leg positioning in the pQCT. Illustration by Vicky Earle, Medical Illustrator, UBC, The Media Group. ................................................................................. 30 Figure 1.13. Illustration of the partial volume effect (PVE) whereby the voxels at the bone edges (blue voxels) contain both bone and soft tissue densities, resulting in a lowered density for the blue voxels. Partially filled voxels will not be accounted for during the analysis. Adapted from Zemel et al. (148), with permission from Elsevier. ...................................................................................................... 31 Figure 1.14. Scout scans for reference line placements in pQCT scans. Manufacturer recommended reference line (A) placements based on growth plate if (i) open, at the outer most medial border, and (ii) closed, at the medial border of the distal endplate (shown as scout scans at the radius). An alternative placement of reference line (A) at (iii) the most distal edge of bone end/surface (in this example, at the tibial plafond). Adapted from Ashby et al. (for images i and ii) (155) with permission from Springer-Verlag London and Burrows et al. (image iii) (156) with permission from Elsevier BV. ............................................................................................................................................... 33 xxiii  Figure 1.15. Illustration of polar stress-strain index (SSIp) calculated from peripheral quantitative computed tomography (pQCT) that estimates torsional strength along the z-axis of the bone shaft. CSMI, cross-sectional moment of inertia (mm4); d, distance of voxel area to the central axis; a, voxel area (mm2); Z, section modulus (mm3); dmax, maximum distance of voxel area to the central axis; voxel.Dn, voxel density; P.Dn, physiological “maximal” cortical bone density at 1,200 mg/cm3. ... 34 Figure 1.16. Total body bone mineral content (BMC) gain in boys and girls from longitudinal DXA data (11). Age of peak lean mass (LM) (181) and BMC gains in relation to age at peak height velocity (APHV) in boys and girls. The lag period between PHV and peak BMC (shaded area) is approximately 8 months and represents a period of low bone mass for age. Adapted from Bailey et. al, 1999 (11), with permission from the American Society for Bone and Mineral Research. ........... 43 Figure 1.17. This schematic illustrates the wide variation in maturation phases as assessed by the method of Tanner across pubertal development (i.e. adrenarche and gonadarche) in girls relative to chronological age. Reproduced from Dorn et al. (175), with permission from Taylor & Francis Informa UK Ltd. ........................................................................................................................... 46 Figure 1.18. A functional model of bone development based on the mechanostat theory where bone continuously adapts to external challenges with the known modulators.  Reproduced from Rauch and Schoenau (223), with permission from Lippincott, Williams & Wilkins, Inc. ................................. 54 Figure 1.19. This schematic depicts trends in hormonal levels, longitudinal growth (height velocity) and bone mineral accrual (bone mineral content (BMC) velocity) in girls in relation to maturation stages and chronological age. Boys display similar patterns of hormonal and bone growth but with a delay of about a year later than girls to attain peak height and BMC levels. Reproduced from MacKelvie et al. (250), with permission from BMJ Publishing. ........................................................................... 59 Figure 1.20 A schematic that describes the effects of sex hormones on longitudinal bone growth and bone surfaces. ♀, girls; ♂, boys; E, estrogen; T, testosterone; GH, growth hormone; IGF-1, insulin-like growth factor-1; ERα, estrogen receptor-α, ERβ, estrogen receptor-β, AR, androgen receptor. Reproduced from Bellido and Hill Gallant (251), with permission from Academic Press. .............. 62 Figure 1.21. Schematic that illustrates the effect of exercise on the skeleton. During exercise, load is transmitted to the skeleton through direct stimulation on bone mechanosensors and by indirect stimulation through dynamic muscle activity. Hormones from fat and the liver modulate loading by affecting bone and muscle growth as well as muscle performance, and act indirectly through potential changes in the mineral reservoir. Reproduced from Bonnet and Ferrari (336), with permission from International Bone and Mineral Society. .............................................................. 75 Figure 1.22. A schematic that illustrates the effect of physical activity on bone mass at different periods across the life span. The red curve represents a continuous exercise effect. Without a bone loading xxiv  stimulus through exercise, bone mass accrual could be attenuated over time. The yellow curve represents a person engaging in normal activities (e.g., walking). Reproduced from Bonnet and Ferrari (336), with permission from International Bone and Mineral Society. ................................ 75 Figure 2.1. Key words and subject headings specific to the search concepts applied in electronic databases searches. ....................................................................................................................................... 88 Figure 2.2. Search concepts, the use of Boolean operators and search filters to identify relevant articles for the systematic review. RCTs, randomized controlled trials; CTs, controlled trials ......................... 88 Figure 2.3. The HPSS model with four Action Zones - school environment/culture, community partnerships, student support and teaching & learning to enable a flexible and sustainable program. ..................................................................................................................................................... 94 Figure 2.4. Timeline of Health Promoting Secondary Schools (HPSS) study and Bone Health Study (BHS) measurements at baseline (T1) and at follow-up (T2)................................................................... 100 Figure 2.5. The Centre for Hip Health and Mobility Mobile Research Laboratory; uncoupled truck and trailer (top); entrance to the Mobile Lab (bottom left) and deployed rooms for DXA and pQCT measurements (bottom right). ...................................................................................................... 105 Figure 2.6. A schematic diagram that illustrates the horizontal Frankfort plane and the position of the head for measurement of stretch stature. .............................................................................................. 108 Figure 2.7. The figure illustrates a participant’s bone health being assessed by pQCT (left) and DXA (right) in the Mobile Lab. ............................................................................................................ 109 Figure 2.8. Position of the reference line and scan sites (7% and 30%) assessed at the radius using peripheral quantitative computed tomography (pQCT). Region of interest, scan direction and start point at the 7% site are also identified. ........................................................................................ 110 Figure 2.9. Positioning of the reference line and scan sites (8% and 50%) sites at the tibia using peripheral quantitative computed tomography (pQCT). Region of interest, scan direction and start point at the 8% site are also identified. ........................................................................................................... 111 Figure 2.10. Images of the radius acquired using peripheral quantitative computed (pQCT); scout view that shows the reference line position (left). Images (middle) acquired at the 7% site with trabecular (orange) and cortical (white) bone compartments and at the 30% site (right) that demonstrates the high proportion of cortical bone. Images also illustrate the separation using bubble wrap between the measured limb and the support platform. This assists with accuracy when drawing the region of interest for analyses. .................................................................................................................... 113 xxv  Figure 2.11. Tibial images from peripheral quantitative computed (pQCT); scout view of reference line position (left), tomographic image at the 8% site (middle) with trabeculae (orange) and cortical (white) sections and 50% site (right) with mainly cortical bone. .................................................. 113 Figure 3.1. Search concepts and the use of Boolean operators within databases. RCTs, randomized-controlled trials; CTs, Controlled trials. ....................................................................................... 124 Figure 3.2. Schematic of the effects of bone mass (areal bone mineral density, aBMD) and structure (cross-sectional) on bone strength. The influences of bone mass distribution from the neutral axis are not captured by DXA-based bone density measures that greatly affect bone strength. With the increase in d, distance of the outer bone mass to the neutral axis or plane of bending, approximate bone strength increases are compared with the values in the left column (Adapted from Bouxsein and Krasik, 2006, with permission from Springer). ...................................................................... 125 Figure 3.3. Flow diagram of study selection. ........................................................................................ 130 Figure 3.4. Overall view of bone strength, mass and structure outcomes based on weight-bearing physical activity (PA) or mixed type of PA (Mixed PA). RCTs, randomized-controlled trials. ................... 163 Figure 4.1.Flow chart of participants recruited, measured, excluded and the number of pQCT images analysed. ..................................................................................................................................... 177 Figure 4.2. Significant relationships (p<0.05) of moderate-to-vigorous physical activity (MVPA) regression residuals and bone strength (SSIp, BSI), structure (Ct.Ar) and density (Tt.Dn) at the 8% and 50% tibia sites in boys and girls. SSIp, polar strength strain index; BSI, bone strength index; Ct.Ar, cortical area and Tt.Dn, total density. ................................................................................ 193 Figure 4.3 Significant relationships (p<0.05) of vigorous physical activity (VPA) regression residuals and bone strength index (BSI) at the 8% tibia site in girls (p<0.05). ................................................... 193 Figure 4.4. Significant relationship (p<0.05) of grip strength regression residuals and bone strength (BSI) and structure (Tt.Ar) at the 7% radius site (p<0.05) in boys and girls. BSI, bone strength index and Tt.Ar, total bone area. ................................................................................................................. 195 Figure 4.5. Scatterplots of grip strength regression residuals and bone strength (SSIp) and structure (Tt.Ar , Ct.Ar ) at the 30% radius site (p<0.05) in boys and girls. SSIp, polar strength strain index; Tt.Ar, total bone area and Ct.Ar, cortical area. ....................................................................................... 196 Figure 5.1. Flow diagram of school and participant recruitment based on CONSORT guidelines for clustered randomized trials. ......................................................................................................... 226 xxvi  Figure 5.2. Flowchart of accelerometer wear and valid accelerometry based on a minimum three 10-hour days criterion. ............................................................................................................................. 227 xxvii  List of Abbreviations ABBREVIATION TERMS 2-D/3-D Two-/Three-dimensional aBMD Areal bone mineral density by dual energy X-ray absorptiometry APHV Age at peak height velocity BC British Columbia BHS Bone Health Study BMAD Bone mineral apparent density BMC Bone mineral content BMD Bone mineral density BSI Bone strength index CCS Canadian Cancer Society CHHM Centre for Hip Health and Mobility CI Confidence interval CIHR Canadian Institutes of Health Research CON Control cpm Counts per minute CRD Centre for Reviews and Dissemination CSA Cross-sectional area Ct.Ar Cortical bone area Ct.Dn Cortical bone mineral density Ct.Th Cortical thickness DXA Dual energy X-ray absorptiometry FEA Finite element analysis FFQ Food frequency questionnaire G Gravitational force or ground reaction force xxviii  ABBREVIATION TERMS GH Growth hormone HBS Healthy Bones Study HHQ Health history questionnaire HPSS Health Promoting Secondary Schools  HR-pQCT High resolution peripheral quantitative computed tomography IGF-1 Insulin-like growth factor - 1 Imax Maximum second moment of inertia Imin Minimum second moment of inertia INT Intervention Ip Polar second moments of area IU International units  LM Lean body mass without bone LPA Light physical activity µSv microSieverts MCSA Muscle cross-sectional area Me.Ar Medullary area MES Minimal effective strain MESm Minimal effective strain for modeling MESr Minimal effective strain for remodeling MESp Minimal effective strain for repair MPA Moderate physical activity MRI Magnetic resonance imaging MVPA Moderate-to-vigorous physical activity NHANES National Health and Nutritional Examination Survey OI Osteogenic Index xxix  ABBREVIATION TERMS PA Physical activity PAQ-A Physical Activity Questionnaire for Adolescents PAQ-C Physical Activity Questionnaire for Children PBMAS Pediatric Bone Mineral Accrual Study PBMCV Peak bone mineral content velocity PE Physical education PHV Peak height velocity pQCT Peripheral quantitative computed tomography PYPAQ Past Year Physical Activity Questionnaire PVE Partial volume effect RCT(s) Randomized controlled trial(s) ReaCT Real Community Trial SED Sedentary time SDT Self-determination theory SSIp Polar strength-strain index Tb.Ar Trabecular area Tb.BV Trabecular bone volume Tb.Dn Trabecular density Tt.BMC Total bone mineral content Tt.Ar Total bone area Tt.Dn Total bone mineral density UBC University of British Columbia vBMD Volumetric bone mineral density VPA Vigorous physical activity Z Section modulus xxx  Acknowledgements It all started with a long distance call from Canada to Malaysia that brought me to beautiful Vancouver for my doctoral studies. My thanks to Heather McKay for her leap of faith in taking me on as her doctoral student. Heather, you have inspired me in so many ways with your leadership and vision. Your guidance in research and writing are invaluable and your support in my endeavours are priceless. The other awesome Heather – Heather Macdonald, I am indebted to your many questions on ‘whys’, ‘hows’ and ‘whats’ that stimulated my thought process and challenged me to look at things differently. Heather and Heather, your contributions made my PhD journey richer. Thank you! My thanks also to Joan Wharf Higgins, for the HPSS study. It was a privilege to be part of such a unique study that gave respect and ownership to secondary school students to be decision-makers in their actions for health. To Lindsay Nettlefold, your friendship and enthusiasm in providing endless statistic discussions and for being my sifu in STATA, gamsa hamida! There are numerous others that have contributed to my learning in one way or another at CHHM and special thanks goes to Maureen Ashe for her mentorship during my systematic review paper and Danmei Liu for sharing her bone imaging expertise and her ‘expert eye’.  To my HPSS Bone Health Study crew and posse, Sophie Kim So Jung and Paul Drexler, forgive me for being such a slave driver during our grueling data collection period. Your dedication, commitment and humour made it possible to obtain the quality data that we currently have. Christine Voss and Leigh Gabel, your timely appearance during our data collection period made us sane and provided the extra hands and smiles. Douglas Race, our research maestro, thank you for the many calls and your patience in handling the accelerometer data.   My doctoral experience in Vancouver was made extra special by the ‘birds of a feather’ friends that I will forever hold dear to my heart. Sarah Moore, you picked me up, literally, from the airport, and your encouragement and support at times when it seems to be the darkest, will always be cherished. The Salmonians -- Campbells, Gonzalezs, Mackenzies; Jungs and Ratus, your kindness, compassion and acceptance combined with water-gun fights, pumpkin smashin’ and shared feasts taught me (and my family) that there is much to enjoy amidst our PhD ‘sufferings’.   To my Universiti Sains Malaysia sponsors, your support and faith in me made this happen. My heartfelt thanks to the School of Health Sciences Dean, Prof. Ahmad Zakaria, and Exercise and Sports Science program chairman, Assoc. Prof. Mohamed Saat Ismail for giving me the chance to further my education. My dearest colleagues and cheerleaders Chen SC, Lim BH, Chen CK and Ooi FK, Ang BS that gave me sound advice and encouragement, ribuan terima kasih.  To my family both in Penang and Kota Bharu, I am sorry I missed out on so much and I am amazed at how you let me be worry-free during the years away. Your constant understanding and encouragement gave me peace to focus on what needs to be done. Kam sia! And last but not least, my deepest appreciation to my darling boys – Jayden, Lewis and SuperDad-Husband, Khuan. There were times when I wished I was alone so that I would be able to complete my studies without distractions and obligations. But without those smile-inducing, comforting ‘distractions’ and ‘obligations’, I would have been a very lonely soul. Thank you for sticking by me when I was a grumpy mommy or a distant, distracted wife. This thesis is possible because of you. I love you for being there with me. Thank you!   xxxi  Dedication  Khuan, this is for you. Thanks for ‘hanging out’ with me…    1   Chapter 1: Introduction, Literature Review and Research Hypotheses 1.1 Introduction  Osteoporosis is a chronic condition that manifests in later adulthood increasing fracture risk and creating immense financial and societal burden. One in two women and one in three men over the age of 50 years will have an osteoporosis-related fracture in their lifetime (1). This leads to an up to three-fold increased risk of mortality, compared with healthy men and women without fractures (2).  In 2001, the annual direct health care cost for osteoporotic hip fractures in Canada was an estimated $650 million and is projected to reach $2.4 billion by the year 2041 if trends are not reversed (3). More recently, the cost of osteoporosis for 2007/08 in Canada was $2.3 billion or 1.3% of total Canadian health care costs (4). Thus, the estimated cost to treat osteoporosis-related hip fractures alone by 2041 is tantamount to what it costs to treat osteoporosis and all related fractures in 2007/08. Also, by 2022, an estimated 18% of Canadian women and 15% of Canadian men over the age of 65 with a prior fragility fracture will have a major hip fracture (5). Approximately 40% of individuals (aged 50 years and above) with all types of fragility fractures required rehabilitative and home care services after discharge and up to 47% reported receiving informal care from friends and relatives (6). Overall, osteoporosis and related fractures also affected quality of life and those afflicted were less able to provide self-care, be mobile or to live without pain (7).  Clearly, there is a need to prevent osteoporosis and related fractures. Prevention can start early during the childhood and adolescent years (1,8). From longitudinal pediatric bone studies, we learned that the amount of bone mass accrued in the two years around puberty (26% of adult status) (9) is equivalent to the average amount of bone mass lost in men and women from age 20 to 90 years old (10). This provides a great window of opportunity for bone health optimization during adolescence.    Chapter 1 – Introduction, Lit Review & Research Hypotheses 2  Physical activity (PA) positively influences bone accrual in adolescents (11–13) as mechanical forces acting upon bone elicit differentiation of progenitor cells and/or proliferation of osteoblasts to form more bone (14). Sadly, the positive effects of weight-bearing PA on bone mass acquisition (11,12,15) and maintenance (16) are diminished by the reality that 93% of adolescents in Canada are not currently meeting the recommended guidelines for acquiring 60 minutes of moderate-to-vigorous PA (MVPA) daily (17). Worldwide, 80% of 13-15 year olds across 105 countries do not meet these guidelines (18). Taken together, this suggests that today’s youth may be at increased risk of osteoporosis in later life.  Assessing PA, especially in children and youth has always posed a challenge. Studies that focus on PA and bone health rely mainly upon PA questionnaires as a data collection tool. Although many questionnaires have demonstrated validity (19,20), measures of type and duration of PA children and adolescents partake of, remains subjective.  Direct, objective measures of PA are now commonly used within the realm of child and youth PA research (21). Accelerometry-based activity monitors assess actual movements (most often based on vertical acceleration) (22). Validated thresholds are then used to classify raw accelerometer counts into PA intensity categories which can then be quantified with regards to duration and frequency (23). More recently, accelerometers have also been used to assess time spent being sedentary (24,25). However, very few child or adolescent bone health research studies used accelerometers to assess PA and even fewer objectively measured sedentary behaviour and reported its potential influence on bone health. Standardization and reproducibility of accelerometer-based PA outcomes was a concern in children and adolescents due to differences in energy expenditure if using adult-based thresholds to classify PA intensity (26). Also many studies did not report fully accelerometry protocols used and this hindered comparing results across studies (27). In light of this, more youth-based recent calibration studies (28), recommendations of standardized reporting in accelerometer studies (27), and newer models of   Chapter 1 – Introduction, Lit Review & Research Hypotheses 3  accelerometers (29) advanced the use of accelerometers to objectively measure PA in children and adolescents. Thus, the uses of accelerometry in pediatric bone studies are helpful to understand bone responses to PA in this population.  While a number of intervention studies targeted primarily children and bone mass (by dual energy x-ray absorptiometry (DXA)) as a primary outcome (30–34), we know relatively little about the effects of PA and sedentary time (SED) on adolescent bone structure and strength. For the purposes of my thesis I define bone structure as bone geometry and macroarchitecture such as cortical and trabecular area, cortical thickness, periosteal and endosteal circumference (35) and bone strength as the ability to resist fractures driven by bone’s extrinsic (mass, structure) and intrinsic properties (degree of mineralization) (36). A detailed description of bone strength, structure and mass are discussed in section 1.2.3, where I discuss imaging tools used in assessing pediatric bone. A recent systematic review investigated the role of PA on bone strength and found only four publications from two randomized controlled trials (RCTs) in children and one RCT in adolescents (37). Four of the five studies were DXA-based and only one study (from our research group) used peripheral quantitative computed tomography (pQCT) to assess 10-year old children. All five studies implemented school-based interventions and all used subjective measures (questionnaires) to assess PA levels.  Schools may be the most effective setting to roll out PA interventions targeting children and adolescents (38). Health interventions conducted in schools can change adolescent health behaviour regardless of family environment (39,40). Those PA and bone health intervention studies conducted in the school setting that yielded significant results, reported high compliance (≥70%) (41,32,33). However, most PA intervention studies focused on elementary-age children; there are very few studies of adolescents (42). Skeletal changes observed in younger children in response to effective interventions cannot be applied to adolescents for a variety of reasons; 1) the influence of stage of maturity on bone   Chapter 1 – Introduction, Lit Review & Research Hypotheses 4  development (43), 2) the relevance of the PA intervention to adolescents and 3) differences in compliance to the intervention (44). Adolescents are not larger children—growth and development is occurring at an accelerated tempo across a wide range of time frames and youth are more autonomous and selective in the activities they choose to adopt (or not), compared with younger children.  The process of being physically active is closely coupled with the basic action of muscles – to generate force. Based on mechanostat theory, skeletal muscle generates forces that influence bone modeling and bone adapts to mechanical loading to be functionally competent (45). This is supported by longitudinal studies where peak lean mass accrual preceded peak bone mass (46–48) and bone strength accrual (49). In addition to biomechanical influences of muscle on bone development, muscle-bone interactions are mediated through various biochemical pathways (50). Thus, to examine how PA influences bone strength, structure and density, it is essential to consider the specific role that muscle plays in this adaptive process.   The primary aim of my thesis is to extend previous studies of elementary school, and the few secondary school, based PA interventions that focused primarily on bone mass (by DXA) as an outcome. To do so I will evaluate the effects of a unique choice-based PA intervention program on adolescent bone strength, structure and density. My secondary aim is to identify modifiable factors related to bone strength, structure and density in adolescents. Two other novel aspects of my thesis are that I used pQCT to assess bone variables and accelerometry to measure PA objectively, in combination with a validated PA questionnaire. My thesis is divided into three parts; Part I is a comprehensive systematic review of peer-reviewed, published RCTs and observational trials of PA and bone strength in children and adolescents. Part II examines the determinants of bone strength, structure and density in adolescents (including PA, muscle and sedentary time). Part III investigates the effect of a choice-based PA intervention (Health Promoting Secondary School (HPSS) study) on bone strength, structure and density in adolescents.    Chapter 1 – Introduction, Lit Review & Research Hypotheses 5  In Chapter 1, I review relevant literature on bone biology and biomechanics, imaging modalities used to assess bone strength and structure, maturity- and sex-related differences in bone, determinants of bone strength, structure and density in children and adolescents with a focus on the effects of PA and SED, measurement of PA and PA-related constructs (specifically muscle strength) on the growing skeleton. I then provide the rationale, specific objectives and hypotheses for the three main parts of this thesis. In Chapter 2, I describe the methods and protocols I used to assess my outcomes and steps taken to conduct the systematic review; the HPSS and Bone Health Study (BHS) design, protocols and primary outcomes, and the HPSS intervention program. In Chapters 3 to 5, I provide the results, discussion and conclusions for three different parts of my study. In the final integrative chapter (Chapter 6) I discuss the global outcomes of the thesis, propose recommendations for future research in pediatric bone, summarize the results and provide a conclusion related to the thesis as a whole.    1.2 Literature Review In this section, I discuss the relevant literature that forms the basis for this dissertation in six parts: bone anatomy and physiology, bone biomechanics, bone imaging modalities, bone growth and development during adolescence, modulators of bone health and lastly, PA and bone health studies in adolescents.  1.2.1 Bone anatomy and physiology In this section, I briefly describe basic bone anatomy related to long bones in the human body, as they are the focus of my study. I follow this with a brief overview of bone physiology that contributes to bone growth, development, repair and maintenance.   Chapter 1 – Introduction, Lit Review & Research Hypotheses 6  1.2.1.1 Whole bone composition and structure Bone is composed of 65% mineral in the form of hydroxyapetite crystals, 25% organic material and 10% water. Ninety percent of the organic material consists of type-1collagen (cartilage) and the remaining 10% consists of non-collagenous proteins such as extracellular and cellular protein (51). The hard crystals and ductility of the flexible collagen matrix confer the stiffness of bone.  Bone features are distinct to their functions: birds have light and hollow bones for flight purposes while human ear bones are highly mineralized to transmit sounds for their acoustic role. There are numerous types of bones in the human body and the adage ‘form follows function’ adequately describes bone anatomy and its purpose. For example, the cranial bones are flat bones, comprised of a bony sponge-like network that absorbs and dissipates forces to protect the brain (52). Long bones, on the other hand, function to support our body mass and act as levers to enable movement (53). Long bone anatomy is uniquely shaped as being longer rather than wider, with articulate ends called epiphyses that are connected by a long shaft called the diaphysis (Figure 1.1) (54). The epiphyses are shaped to efficiently disperse compressive loads and contain growth plates that form trabeculae1, also known as cancellous or spongy bone. As bone grows in length, trabecular bone is replaced by cortical or compact bone that forms the diaphysis. When linear growth ceases, the growth plate fuses and forms the epiphyseal line. A closer look at the formation of trabecular and cortical bone reveals highly organized structures that I discuss in more detail below.                                                      1 In this thesis, I use the term ‘trabecular’ and ‘cortical’ to refer to spongy or cancellous and compact bone tissue, respectively.   Chapter 1 – Introduction, Lit Review & Research Hypotheses 7   Figure 1.1. Long bone structure2. Adapted with permission from the National Institute of Health, through personal communication.   Formation of trabecular and cortical bone in long bones occurs via endochondral ossification where cartilage forms (from chondrocytes) prior to bone tissue deposition. The same chondrocytes form the growth plates in immature bones (55). After birth, primary ossification occurs within three weeks while secondary ossification can take up to a year to create complex Haversian systems (Figure 1.2), complete with osteons and vascular canals (51). The adult skeleton comprises 80% cortical and 20% trabecular bone (56).                                                      2 The term ‘structure’ is used interchangeably in this thesis with ‘geometry’ and ‘architecture’ to describe dimensions of bone (e.g., shape, size, area, thickness, spatial distribution) Cortical bone   Chapter 1 – Introduction, Lit Review & Research Hypotheses 8   Figure 1.2. Structural elements in cortical (compact) and trabecular (spongy) bone. Osteons shaped from lamellar sheets surround blood vessel canals creating the Haversian system. Adapted with permission from the U.S. National Institute of Health.   Woven bone is the first bone matrix to be laid down before any trabecular or cortical bone is formed due to the rapid formation and quick mineralization of randomly placed type-1 collagen fibres. Therefore, woven bone is also created when injury occurs to bone tissue (55). As bone tissue repairs, a callus forms at the injury site, which provides stability during the healing process. Woven bone is later replaced by structured osteons. Half-osteons or hemi-osteons (with no vascular channels) form trabecular plates and rods with an average thickness of 200 μm (51). Although trabecular bone appears randomly constructed, it actually undergoes targeted modeling where the formation is highly organized and structured along the stress lines (51). The architectural construction of trabeculae support weight bearing and dissipate forces to the tougher cortical bone through the connected struts (Figure 1.3). The lattice-like network of trabecular bone provides a light yet strong and supportive structure.    Chapter 1 – Introduction, Lit Review & Research Hypotheses 9   Cortical bone functions like a brick wall to defend the structural integrity of bone. It is made up of osteons (about 100-250 μm in diameter). The osteon is composed of a central canal (Haversian) channeling a blood vessel, nerves and lymphatics surrounded by concentric layers of lamellae. Each lamella is about 3-7 μm thick and is separated by an interlamellar layer (about 1μm thick) (51). Throughout cortical bone, Haversian and Volkmann (transverse links to Haversian) canals (Figure 1.2) provide a vast network of capillaries running through an otherwise dense and stiff tissue, contributing to 3-5% of the porosity in cortical bone. Interstitial bone is the remains of primary or secondary bone that fills the space between cylindrical osteons that are no longer remodelled (51). A cement line separates osteons from interstitial bone and serves as an important mechanical structure. The cement line controls fatigue and fracture processes by absorbing energy to stop fracture propagation and provides viscous dampening in cortical bone (51). Thus, cortical bone structure is tough not only due to the compact arrangement of osteons but also from built-in mechanisms that deter further fracture proliferation (57).  Figure 1.3. The importance of bone architecture, independent of the amount of material (i.e. mass). Trabecular bone in which the cross struts are disconnected (C) cannot support the same load as trabecular bone with connected cross struts (A), despite similar bone mass. A smaller bone mass that is buttressed properly (B) may be able to support more load than a greater mass in which the connectivity is compromised (C). Figures indicate maximum load. Reproduced from Burr and Akkus (51) with permission from Elsevier.   Chapter 1 – Introduction, Lit Review & Research Hypotheses 10  1.2.1.2 Physiology of bone growth, development and maintenance  Three basic bone cells – osteoclasts, osteoblasts and osteocytes – are involved in bone growth, development and maintenance. These cells act on four bone surfaces: periosteal (outer bone surface), endosteal (marrow bone surface), intracortical (inner canal of an osteon) and trabecular (51). Osteoclasts function to remove or resorb bone tissue (structural construction) and are derived from the hematopoietic monocyte macrophage lineage present in circulation and the bone marrow (58). Osteoclasts can resorb up to 200,000 μm3 of bone per day (53), paving the way for new bone matrix. Without osteoclasts’ resorption ability, bone formation would be disturbed. Dysregulation of osteoclasts can lead to excess or insufficient bone mass and compromised bone structure, two contradicting medical conditions known as osteopetrosis and osteoporosis, respectively (58). With osteopetrosis, bone is too stiff (brittle bones) due to high mineralization and is highly susceptible to fracture while osteoporosis, a condition of insufficient bone mass, leads to reduced bone strength, increasing fracture risks as well. The coordination of osteoclast activation depends on both mechanical forces and hormones. At the end of the resorption phase, osteoclasts undergo programmed cell death or apoptosis (58).  Osteoblasts are formed from mesenchymal progenitors, which also produce chondrocytes, myocytes and adipocytes. Osteoblast formation is regulated by several morphological transcription factors (58). Osteoblasts synthesize bone matrix for growth and repair by producing type-1 collagen and then deposit minerals to form new bone (58). Systemic, mechanical and local factors stimulate osteoblast activities. On average, osteoblasts lay down bone matrix or osteoid at a rate of 0.5-1.5 μm/day (53). During bone growth, osteoblasts can act independently on bone surfaces or be coupled with osteoclasts to form a bone multicellular unit to form and resorb bone, respectively. At the end of their lifecycle, osteoblasts are either embedded in bone matrix as osteocytes (5-20%) or transform into lining cells (15-30%) that cover the new bone surface. The remaining 50-80% of osteoblasts experience apoptosis (58).    Chapter 1 – Introduction, Lit Review & Research Hypotheses 11  Osteocytes formed from mature osteoblasts have a different morphology and role than their predecessors. They are the most abundant bone cells in skeleton, and are found in the bone matrix and on bone surfaces (58). Osteocytes form projections from their cell body, known as dendrites, which connect to other nearby osteocytes through the gap junction in canaliculi (microscopic canals in bone tissue) (58). These dendrites are thought to assist osteocytes in responding to mechanical and hormonal signals, and coordinate the function of osteoblasts and osteoclasts (59). When mechanical stimulus is low, osteocytes undergo apoptosis and this triggers osteoclast action to remove bone (58,60).  Bone microdamage also triggers osteocytes to direct osteoclasts and osteoblasts to remove damaged bone and replace it with new bone (61). It is postulated that osteocyte viability is maintained by adequate mechanical stimulus (60).   1.2.1.2.1 Bone modeling and remodeling  Modeling and remodeling are the two main physiological mechanisms of bone growth and maintenance. I provide a general description of bone modeling and remodeling in Table 1.1.   Table 1.1. General overview of bone modeling and remodeling. Reproduced from Allen and Burr (62), with permission from Academic Press.  Modeling Remodeling Goal  Shape bone, increase bone mass Renew bone Cells  Osteoblasts or osteoclasts and precursors Osteoblasts, osteoclasts and precursors Bone envelope Periosteal, endosteal and trabecular Periosteal, endosteal, trabecular and intracortical Mechanisms Activation-formation or activation-resorption Activation- resorption - formation  Timing Dominant in childhood but continues throughout life Throughout life Net effect on bone mass Increase Maintain or slight decrease  Bone modeling is prominent during growth and development in children and adolescents and functions to increase bone length and size and develop bone shape. Formation modeling and resorption   Chapter 1 – Introduction, Lit Review & Research Hypotheses 12  modeling signalled by local tissue strain consequently alter bone shape and structure on separate bone surfaces (62). During longitudinal bone growth, formation and resorption modeling occur on different bone surfaces. As illustrated in Figure 1.4, resorption modeling occurs on both periosteal and endosteal surfaces around the metaphysis (63). This causes ‘bone drift’ whereby bone mass is shifted away from the central bone axis to widen (and lengthen) bone (62). Modeling also occurs throughout our lifetime to reinforce cortical (through periosteal apposition) and trabecular (increased number and thickness) structure when load is applied on bone. I address bone loading more fully in section 1.2.2.   Figure 1.4. During dynamic growth, the growth plate contributes to linear growth (I) while resorption around the metaphyseal periosteum occurs to keep the bone shape (II). Resorption of trabeculae shifts the metaphysis-diaphysis border (III). Bone is added periosteally (IV) and resorbed endosteally (V) to increase bone width. Images reproduced from Rauch et al. (64), open access.  Bone remodeling is mainly carried out in ‘pockets’ by bone multicellular units in targeted areas that are activated by osteocyte apoptosis and microdamage (Figure 1.5) (62). Various stages of remodeling occur at any one time throughout the body but sequentially, the first stage involves recruitment of preosteoclasts to the damaged site through signals from osteocytes. Differentiation from preosteoclasts   Chapter 1 – Introduction, Lit Review & Research Hypotheses 13  into mature osteoclasts involves a multitude of signaling molecules, including macrophage colony stimulating factor (M-CSF), receptor activator for nuclear factor κB ligand (RANKL), tumor necrosis factor, interferon gamma, and inter-leukins (58). Next, mature osteoclasts create Howship lacunae, or resorption pits, to dissolve bone mineral and release collagen fragments. This process takes about four to six weeks. Then, osteoblasts are recruited to lay down new bone matrix at the resorption site during the reversal phase (62). Specific signaling pathways to recruit osteoblasts are still unconfirmed but we know several important signalling molecules regulate osteoblastic differentiation. These include bone morphogenetic proteins (BMPs), transforming growth factor (TGF)-β, Wnt glycolipoprotein, Indian Hedgehog protein, parathyroid hormone, insulin-like growth factor-1, fibroblast growth factors, and Notch (a transmembrane protein) (65). Bone matrix development and mineralization of bone by osteoblasts can take four to six months (66). After new bone is formed, some osteoblasts transform into lining cells or osteocytes while the rest undergo apoptosis (62).   Figure 1.5. From the resting bone stage, bone remodeling is triggered by signals from osteocytes, which recruit preosteoclasts to start osteoclastogenesis where mature osteoclasts are formed to resorb bone. Next, preosteoblasts are recruited in the reversal stage to form mature osteoblasts. Osteoblasts then deposit bone matrix from the cement line (red line) to fill the lacuna. Finally, bone is mineralized by osteoblasts and subsequently osteoblasts undergo apoptosis, are entombed into osteocytes or transforms into lining cells. Reproduced from Kapinas and Delany 2011 (65), open access.      Chapter 1 – Introduction, Lit Review & Research Hypotheses 14  1.2.2 Bone biomechanics In this section, I discuss the mechanical properties of bone, the theoretical basis of bone adaptation to mechanical stimuli and the animal and human studies that provided in-depth understanding of bone adaptation to mechanical loading.  1.2.2.1 Biomechanical properties of bone Overall, human bones need to be stiff and strong yet light enough to allow movement of the body (67). Adequate bone strength is required to conduct daily activities without injury and to withstand larger forces sustained from low impact falls (i.e. falls from standing height or less). The ability of bone to absorb energy and avoid fracture depends on bones’ material properties, mass and structure (67).  Mechanical testing of human cadaver bones defines several basic mechanical properties such as stiffness, ductility and toughness. Depending on the direction of applied loads, bones experience compressive, tensile, shear, bending and torsional forces (68,69). Results of mechanical tests on whole bone (including cortical and trabecular bone) are used to generate a load-deformation curve that defines the extrinsic properties of bone – that is the mechanical properties of whole bone structure (70). The two main regions below the curve are the elastic and plastic regions. Within the elastic region, an applied load to bone results in deformation; when the load is removed, the bone reverts back to its original form. The slope of the curve in the elastic region represents the extrinsic stiffness or rigidity of bone (68). When loads are applied past the yield point, within the plastic region, permanent deformation occurs. Consequently, bone will fracture under larger loads at the failure point. When controlled for bone size, the load-deformation curve converts to the stress-strain curve that describes bone’s intrinsic properties.   Chapter 1 – Introduction, Lit Review & Research Hypotheses 15                                  Intrinsic bone properties are represented by the stress-strain curve (Figure 1.7), which defines bone material properties at the tissue level. To understand these properties, I briefly present definitions for basic biomechanical terms. Stress at the tissue level, is defined as force per unit area (units: Pascal, (Pa) where 1 Pa = 1 N/m2) (68). Strain at the tissue level, is a percentage change in length from the original dimensions. As strain is a relative deformation, there are no units. If a bone is stretched to 101% of its original length, the strain would be 0.01 or 1% of 10,000 microstrain (με) (68). With a change in bone length, bone width also changes and this length-width strain ratio is known as Poisson’s ratio. Human cortical bone has a Poisson ratio of 0.28 to 0.45 which means a change of 0.28% to 0.45% in width is observed in the perpendicular (axial) direction of the 1% strain applied (71). Thus, bone tissue does not break instantaneously when loaded but deforms slightly due to bone’s intrinsic material properties. S   A Fu Deformation Load B du Figure 1.6. Load-deformation curve for whole bone properties. The slope (orange line) of the curve represents the extrinsic stiffness (S); the height of the curve denotes ultimate force (Fu); the yellow triangle is the yield point; the red diamond is the failure point; the area under the curve represents A, elastic region and B, plastic region with the combination of A+B equal to the work to failure; and the total deformation to fracture is ultimate deformation (du). Adapted from Turner (70) with permission from New York Academy of Sciences.   Chapter 1 – Introduction, Lit Review & Research Hypotheses 16                The stress-strain curve describes bone’s material properties with the linear slope representing bone’s intrinsic stiffness, or Young’s modulus (68). Young’s modulus is similar across individuals, as the biomaterial used to construct bone is mainly hydroxyapatite and collagen. However, Young’s modulus may differ depending on the direction of stress. Thus, bone tissue is known to be anisotropic (different properties in different directions) and anisotropy is attributed to the orientation of bone tissue. The direction osteons are laid down and formed directs the anisotropic properties and can be thought of as the ‘grain’ of a biomaterial such as the grain in wood. For example, the tensile strength of cortical femoral bone in the longitudinal direction is greater than in the transverse direction, 135 MPa and 53 MPa, respectively, while compressive strength (longitudinal) is even higher at 205 MPa (72). In elderly donors (cadaver femoral bones), cortical bone tensile strength was about 27% stronger compared with trabecular bone. Trabecular tensile and compressive strength was, on average, 85 MPa and 135 MPa, respectively (73). This aptly shows the impact that load direction and bone type (cortical or trabecular) have on the anisotropic characteristics of bone.  Figure 1.7. Stress-strain curve for bone tissue. The linear part of the curve represents the intrinsic stiffness (Young’s modulus (E)). The height of the curve denotes ultimate strength. The yellow triangle is the yield point; the red diamond is the failure point. Strains above the yield point cause permanent damage to bone structure. Post-yield strain before fracture is equivalent to ductility (converse to brittleness). Adapted from Turner (70) with permission from New York Academy of Sciences.  Ultimate strength E Fracture Post-yield strain Stress Strain   Chapter 1 – Introduction, Lit Review & Research Hypotheses 17  Bone strength is dependent on material (mineralization) and structural (size and shape) properties represented by the load-deformation and stress-strain curves. As strain increases past the yield point, permanent deformation occurs. The amount of energy needed to cause fracture is related to bone’s ductility (inverse to brittleness). During linear growth in childhood and adolescence, bone is less mineralized and is therefore considered more ductile. Bone does not increase in size and density arbitrarily, as more energy would be required to move a bigger and heavier skeleton. Instead, bone formation is regulated by mechanical and non-mechanical influences as proposed in the mechanostat theory, which I discuss in detail below. 1.2.2.2 Mechanostat theory and mechanotransduction More than a century ago, Julius Wolff proposed that mechanical stresses regulate bone architecture (74). More recently, Harold Frost (75) proposed the mechanostat theory which suggests that bone adaptation is regulated by several factors: mechanical (the dominant effector based on the Utah paradigm (45)) and non-mechanical, including a negative feedback mechanism that regulates bone physiology (76). According to the mechanostat theory, different strain thresholds, termed minimal effective strains (MES), dictate action of bone modeling (MESm), remodeling (MESr) or microdamage repair (MESp) depending on the quantity of mechanical force received (Figure 1.8) (77). Mechanical forces exerted on bone are mainly through muscle contraction from PA and from ground reaction forces (63). Mechanical forces are transformed to biological pathways through the mechanotransduction process to regulate bone adaptation.   Chapter 1 – Introduction, Lit Review & Research Hypotheses 18                  Figure 1.8. Illustration of the mechanostat theory in bone modeling and remodeling. Bone remodeling occurs in the upper limit of disuse window (DW); MESm – minimal effective strain for bone modeling occurs between the adaptive window (AW) and the mild overload window (MOW); MESp – minimal effective strain for microdamage repair is at the lower limit of the pathologic overload window (POW); Fx: bone fracture strain and με, microstrain. Adapted from Frost (77) with permission from A.R. Liss (etc).  There are four distinct phases in mechanotransduction, the process by which mechanical stimuli leads to bone formation or resorption. The first is mechanocoupling where the mechanical load perturbs the osteocytes and their dendrites (78). These perturbations result in fluid flow that produces a shear stress detected by mechanoreceptors located at cell membranes. Currently, there are three known groups of mechanoreceptors – mechanosensitive ion channels, cell adhesion/cytoskeletal molecules (e.g., connexin 43, β-integrin) and G-protein related molecules – that ‘open’ the channels to allow proteins, lipids and calcium to enter cells or be released into extracellular matrix (79). Movement of biochemical substances across membranes creates secondary biochemical signals that initiate the second phase known as biochemical coupling. Biochemical coupling occurs through various pathways, including the cyclooxygenase (COX)-prostaglandin E2 (PGE2), nitric oxide (NO) and Wnt-lipoprotein receptor-related ≈ 3000 με ≈ 25000 με ≈ 50-100 με ≈ 1000 με   Chapter 1 – Introduction, Lit Review & Research Hypotheses 19  protein (LRP 5/6) pathways (80). The third phase of mechanotransduction involves signal transmission to bone cells from previous biochemical pathways. For example, the Wnt-LRP5/6 pathway in mature osteocytes promotes survival of β-catenin, which activates osteogenesis (58), while down regulating sclerostin (a negative bone formation regulator, expressed from the SOST gene) (81,82), which suppresses bone resorption. During the final phase, effector cell response, bone cells (osteoblasts, osteoclasts) directed by the biochemical signals form or resorb bone (83).  Although mechanical stimuli are the main drivers of bone adaptation, non-mechanical factors related to growth including genetics, nutrition and hormones also influence bone adaptation. I discuss these non-mechanical factors and their role in bone adaptation in section 1.2.6.  1.2.2.3 Principles governing bone adaptation to mechanical stimuli Three basic principles govern bone adaptation to mechanical stimuli (84) illustrated by classic animal studies and supported by studies in humans. I review the key animal and human studies, below. The first principle: bone adapts to dynamic loads and not static loads. In their classic experiment, Lanyon and Rubin experimented with functional loads and controlled external loading on turkey ulnas. They reported that static loads decreased bone formation and increased bone porosity while dynamic loads increased bone formation (85). Dynamic loads also effectively increased bone mineral content (BMC) by 134% through periosteal bone apposition; 36 cycles/day were as effective as 1800 cycles/day (86). When loaded, bone cells deform and exude fluids into interstitial space (87), causing fluid shear stress to initiate mechanocoupling (first step in mechanotransduction) (78). Intuitively, it makes sense that dynamic strain applied and removed in an on-off pattern, would elicit more fluid movement across cell membranes than a one-time static strain on bone cells. Hsieh and Turner showed that a strain frequency of 1 Hz, 5 Hz and 10 Hz on adult female rats increased ulnar bone formation rate in a dose-dependant manner (88).   Chapter 1 – Introduction, Lit Review & Research Hypotheses 20  Furthermore, bone is stimulated by strain magnitudes above strain thresholds. Rubin and colleagues used turkey ulnas to examine the dose-response of loading on bone cross-sectional area (CSA). They found a positive linear correlation between load magnitude and bone area whereby loads above 1000 με stimulated bone formation compared with the contralateral non-loaded ulnas (89). Turner and colleagues aimed to locate a threshold for bone formation in rats and found that a minimal load of 40N or a strain of 1050 με was required to induce bone formation and mineral apposition (90). The minimum strain threshold of 1000 με appears to be constant across several species but bone formation thresholds in humans are unknown. The second principle: short periods of loading are more osteogenic compared with long periods of loading. Lengthy loading bouts are counterproductive for bone formation as bone cells lose their sensitivity to the mechanical stimulus (dependent on strain rate and magnitude) (91). Robling and colleagues showed this with their rat tibia loading model whereby 4 bouts/day of 90 bending cycles/bout resulted in greater gains in bone strength, structure and density compared with a single loading bout of 360 cycles (92–94). These findings support the theory that rest periods, anywhere from 14 seconds to 8 hours (92), are required to restore bone’s sensitivity to mechanical loads. This is related to the fluid shifts in bone cells whereby continuous dynamic loading does not provide a refractory period sufficient enough to restore mechanical sensitivity (95).  The third principle: unique distribution of strains on bone elicits bone formation. Bone cells respond to unusual mechanical loading environments and are less responsive to routine or customary loading signals. Thus, in order to stimulate bone formation, mechanical loading needs to be ‘abnormal’ and ‘error-driven’. Rubin and Lanyon demonstrated how rooster ulnas gain bone mass due to atypical direction of loading (loaded at 90 degrees – transverse direction from the natural wing-flapping strain distribution) (86). Since the peak strain magnitude (10,000-12,000 με), customary strain rate (30,000/sec   Chapter 1 – Introduction, Lit Review & Research Hypotheses 21  and 36,000/sec) and number of loading cycles (4-1800 cycles/day) were below hyperphysiological levels, the investigators attributed the bone changes observed through post-mortem histology and microradiography to the altered strain distribution (86). In a recent micro-CT study, Wallace and colleagues reported that young female mice exposed to diverse directional loading demonstrated greater trabecular volume fraction (1/Tb.BV) (12%), cortical area (Ct.Ar) (11%), cortical area fraction (1/Ct.Ar) (15%) and cortical thickness (Ct.Th) (12%) at the proximal humeri (100 μm distally) compared with mice exposed only to linear directional loads (96).  Based on these principles of bone adaptation, Turner and Robling proposed the osteogenic index (OI), which is a formula to quantify the osteogenic properties of an exercise protocol (97). The equation is derived from well-designed animal studies and takes into account intensity (ground reaction force) (98), number of loads/jumps and duration between exercise sessions (recovery sessions) (92) to generate a score for the particular exercise program (97). The OI equation is presented below,                                                          where GRF is the peak ground reaction force or intensity, N is the number of loads per session, t is the time (hours) between sessions and τ a time constant of 6 hours. Although the OI provides a useful tool to quantify the osteogenic potential of an exercise regimen, it has, to my knowledge, not been validated in studies of children or adolescents. Studies using the OI in adults (99–102) are inconsistent. This is due to differences between studies in how the formula was used to calculate OI values (e.g., OI for a week vs. lifetime score) and the variability in bone outcomes (e.g., bone resorption vs. formation biomarkers).  A study by Lester and colleagues illustrates one application of the OI. They tested three different exercise programs in adult women (20.3±1.8 years) over an 8-week period: 1) resistance exercise (OI = OI (n-session/day)   Chapter 1 – Introduction, Lit Review & Research Hypotheses 22  16.0), 2) running (OI = 20.6) and 3) combined running and resistance training (OI = 36.9). Trabecular volumetric bone mineral density (vBMD; by pQCT) increased 1.3% in the running and combined group (101). Erikson and colleagues conducted a proof-of-concept study (n=7 per group) and found that an 8-week jumping program in men (single or double jump sessions/day, equal number of jumps in both groups, resistance set at 80% of body mass) increased bone formation marker levels over time with significant changes in the twice/day group but bone marker levels were no different compared with controls (102). The investigators highlighted that although the OI might be useful, studies on the optimal OI score for bone modeling have not been conducted (102).  1.2.2.4 Differences in bone adaptation  Dynamic, high-magnitude loads that are of short duration and associated with abnormal strains are known to induce bone formation in animal models. However, the specific bone response varies with age and skeletal site. Further, bone adaptation to mechanical loads differs between cortical and trabecular bone. These factors come into play when results from studies are evaluated. I discuss several animal and human studies that contributed to our understanding of bone adaptation.  1.2.2.4.1 Age-specific bone adaptations Results of several animal studies indicate that the growing skeleton has a greater capacity to adapt to mechanical loads compared with the mature skeleton. I highlight two classic animal and two recent human studies below. First, Turner and colleagues tested strain loads that ranged from 30 N to 64 N in young and old rats (103). Both young and old rats had periosteal bone apposition above 40 N but only increased 59% in the old rats compared with 100% gains in young rats. The investigators also reported that the minimal mechanical threshold for endosteal bone formation was 1050 µε for young rats and 1700 µε for old rats to achieve the same bone adaptation (103). Second, Jarvinen and colleagues exposed young   Chapter 1 – Introduction, Lit Review & Research Hypotheses 23  and adult rats to a 14-week running program. Femoral neck bone strength increased similarly in both groups of animals (104). In young rats, this was a result of increased CSA (25%) compared to vBMD (11%). In adult rats the reverse occurred; a 10% increase in CSA (not significantly different from controls) and a 23% increase in vBMD (104). This suggests that with aging, bone responds differently to mechanical loads (sensitivity, structural vs. density).  In human studies, increased bone strength in response to mechanical loading appears to be a result of changes in bone geometry rather than bone. A series of landmark contralateral comparison studies in racquet sport players (aged between early 20s to mid 50s) reported that bone size (by pQCT) was larger (9% to 32% more) on the playing arm compared with the non-playing arm due to sport-induced adaptations while side-to-side comparisons of vBMD were almost identical (-1% to 2%) (105,106). Similarly, young tennis players (girls, age 10 to 15 years) had stronger bones (by magnetic resonance imaging (MRI), polar second moment of area +11-23%) in their playing arm (humerus) versus their contralateral arm due to a greater Ct.Ar (also periosteal and endosteal circumferences). In contrast, there was no side-to-side difference in total humerus BMC (by DXA) (107). In older master athletes (athletes that train and compete above the age of 35 years; age 54.9±12.4 years in this study), there were no side-to-side differences in bone outcomes (BMC, vBMD, total bone area (Tt.Ar) by pQCT) between the dominant and non-dominant leg (108). However, there were sport-specific differences in Tt.Ar and BMC (but not vBMD) at all three bone sites, 4%, 14% and 66% of the tibia, between sprinters, hurdlers and triple jumpers. This indicates that bone adapts mainly by geometrical changes even in middle-aged athletes that are exposed to stringent exercise training and high impact forces.  To my knowledge, RCTs that looked at bone adaptations to PA across ages (youth to older adulthood) are missing in the literature. There are trials that either focused on children and adolescents (30,32,33) or adults (109–111) and each intervention program differed in frequency, intensity, type and   Chapter 1 – Introduction, Lit Review & Research Hypotheses 24  duration of PA. Overall, most PA intervention trials resulted in positive bone adaptations regardless of age. From a large cohort study, Tobias and colleagues (112) indicated that the level of impact forces required to positively affect bone adaptation may also be different between youth and adults. Thus, it appears that PA can benefit bone health across all ages but further investigation is required to establish the PA characteristics that optimize bone adaptations across different age groups.   1.2.2.4.2 Site and compartment-specific bone adaptations Bone reacts to strains experienced and adapts where it is most needed. For example, rodent bone tends to bend in the medio-lateral direction when loaded mechanically. Thus, resultant bone formation was observed at medial and lateral sections more so than at anterior-posterior sections relative to the neutral axis (Figure 1.9) (79). Similar site-specific adaptations were observed in an RCT that investigated the adaptation of cortical bone surfaces in boys and girls after a 20-month intervention (113). Prepubertal boys who participated in jumping exercises (3 min/day of countermovement jumps) and classroom PA (additional 15 min 5 days/week on top of regular PE), displayed bone adaptations at the anterior, posterior and medial sections, but not the lateral side, of the tibial midshaft (Figure 1.10) (113).  In addition to local adaptations (anterior-posterior, medial-lateral), peripheral and axial sites also respond differently depending on the type of loading, timing of assessment and the bone compartment (cortical, trabecular) assessed. As it is, timing of bone mineral accrual was different depending on the bone site assessed. From a longitudinal study following 8 year olds until age 30 years (114), the femoral neck was the first to achieve peak bone mass accrual, 2 years after age of peak height velocity (APHV). This was followed by peak bone mass accrual at the lumbar spine (4 years after APHV) and total body peak BMC accrual at six years post APHV (114). An understanding of site-specific characteristics that change across time also informs on the PA-induced bone adaptations. In adults, aged 20 to 99 years, Macdonald and colleagues used high resolution pQCT (HR-pQCT) to assess differences in trabecular and   Chapter 1 – Introduction, Lit Review & Research Hypotheses 25                  Figure 1.10. Schematic of the tibial midshaft that shows radial plots of periosteal and endosteal surface changes (left) and cortical thickness changes (right), measured by peripheral quantitative computed tomography. The space between the dotted and solid lines is the difference between pre-pubertal boys randomly assigned to intervention (20 months of 15 min/week extra classroom physical activity and bone loading program) or control groups. Adapted from Macdonald et al. (113), with permission from Springer-Verlag London.  cortical bone at the distal radius and tibia as age increases (115). They found that trabecular bone reduced in volume with advancing age depending on skeletal site. Distal tibia reductions in trabecular bone volume Figure 1.9. Bone adaptations to loading in a cross section sample of the ulnar midshaft of a rodent. The schematic shows the strain distributions (upper panel) and bone formed (lower panel) after ulnar loading. Mechanical loading of the right ulnar, 3 days/week for 16 weeks, resulted in apposition of new bone predominantly on the periosteal surface. New bone formed as a result of the experiment can be visualized (orange area) at the outer medio-lateral bone surface. Adapted from Robling et al. (79) with permission from Academic Press.   Chapter 1 – Introduction, Lit Review & Research Hypotheses 26  were observed starting at young adulthood while loss of bone volume at the distal radius began in middle adulthood (115). Looking at bone compartments, the distal tibia experiences almost equal reduction in trabecular number and thickness in women and men (115). But for the distal radius, women have reduced trabecular number and men have thinner trabeculae as they age (115). This trend was also reported by Khosla and colleagues who also used HR-pQCT to assess similar bone sites (116). For the cortical bone compartment, again, the distal tibia seemed to deteriorate earlier (pre and peri-menopausal) compared with the distal radius (after menopause) in women (115,117). The changes that occur at different sites (local, peripheral, axial) and within different bone compartments (trabecular and cortical) need to be considered when trying to assess whether bone adaptation are truly from PA or natural progression due to age and sex differences.  1.2.3 Bone imaging  As bone strength is the ‘bottom line’ of fracture prevention (118), there is a need to understand how bone strength is estimated using different imaging modalities in order to better interpret bone outcomes. In this section, I present a general overview of DXA and pQCT, along with the strengths and limitations of each. I also briefly describe HR-pQCT, MRI and quantitative ultrasound (QUS) as the other imaging tools used in pediatric bone research.  1.2.3.1 Dual-energy X-ray absorptiometry (DXA) With the development of DXA in 1987 (119), safety and image resolution of bone imaging improved. This provided a catalyst for pediatric bone health research. Briefly, DXA emits two high- and low-energy photons through the body and a detector picks up photons not attenuated by bone and soft tissue (119). DXA scan data is processed pixel-by-pixel depending upon the degree of beam attenuation,   Chapter 1 – Introduction, Lit Review & Research Hypotheses 27  to map out soft tissue regions adjacent to bone (119). The total projected bone area (cm2) is a sum of the pixels within the bone edges. The reported value of areal bone mineral density (aBMD, g/cm2) is defined as the integral mass of bone mineral per unit of this projected area (119). Bone mineral content (BMC, g) is obtained by multiplying aBMD by the area scanned (119).  For DXA, the effective dose – defined as the uniform whole body dose of radiation that would be equivalent to risk of carcinogenic and genetic effects - depends on scan time and size of the individual being scanned (119). Longer scan time means a longer exposure to the radiation energies. In a 15 year-old, the effective dose equivalent for a whole body scan using DXA (Discovery-A, Hologic Inc; scan time of 60 seconds) is 4.2 μSv (120). This dose is less than 0.2% of annual natural background radiation (121). Thus, DXA is a safe as well as a valid, reliable and precise tool to assess clinically relevant sites such as the hip, spine and forearm (122). It is also easy to use and low cost, making it highly desirable for both clinical and research settings (123). Finally, DXA is the current clinical gold standard to diagnose osteoporosis and monitor treatment in adults.  Despite the clinical advantages of DXA, there are limitations when assessing the growing skeleton and the first of which is DXA’s planar technology. In general, this remains the primary weakness, as DXA is unable to assess volume. For example, Figure 1.11 depicts the overestimation of aBMD from DXA projections of large versus small specimens (124). There have been efforts made to develop algorithms (122,125–127) and to correct aBMD for size (126,128) in order to estimate bone strength from DXA images of the hip and spine. One approach was size-adjustments of vertebral segments measured by DXA to assess bone mineral apparent density (BMAD, g/cm3) (126,128). Carter and colleagues used a cuboidal formula (126) while Kroger and colleagues used an elliptical or cylindrical formula (128) to obtain vertebral BMAD. However, there was no consensus as to the most acceptable way to account/correct for size. Other than accounting for size, the International Society for Clinical Densitometry (ISCD) cautioned   Chapter 1 – Introduction, Lit Review & Research Hypotheses 28  that aBMD results need be considered in conjunction with factors such as height, sex, age, pubertal stage, ethnicity and fracture history to avoid misinterpretation in children and adolescents (129).     Figure 1.11. Effects of size on bone parameters assessed by dual-energy X-ray absorptiometry (DXA). Both samples have identical volumetric density but the areal density of the 3x3x3 specimen is greater because of the projected area captured by DXA. Adapted from Carter et al. (126), with permission from Mary Ann Liebert, Inc.   Hip structural analysis (HSA) (125) was developed to assess bone structure and estimate bone strength at the proximal femur. However, this approach is not without its own limitations. The ISCD 2014 position statement highlighted that it may not be possible to accurately identify bone markings required by hip assessment protocols (e.g., HSA) in the growing femur (129). Other than the planar nature of DXA, several other DXA limitations needs to be noted. First, a limitation of DXA is its spatial resolution (about 0.5 mm) (130). Thus, DXA cannot discriminate between trabecular and cortical bone compartments (131). Second, DXA outcomes can also be affected by scanning position, movement and body composition (132). Inaccuracies occur as soft tissue absorbs and attenuates low energy beams from DXA. Thus, soft tissue thickness will affect assessment of bone mineral   Chapter 1 – Introduction, Lit Review & Research Hypotheses 29  (133). Also, the relative distance of bone from the detector due to increased soft tissue thickness leads to systematically higher BMC, aBMD and apparent bone area values (133). In spite of this, analysis and monitoring of body fat and bone mineral free lean mass by DXA are considered valid and reliable in children and adolescents (134,135). Lastly, outcomes across different DXA manufacturers (Lunar, Hologic and Norland) are not comparable due to the different systems used to generate (K-absorptiometry vs. high-low voltage switch), calibrate (different bone and fat mass sources) and detect (remove vs. correct beam hardening) X-rays in the imaging systems (133,136). 1.2.3.2 Peripheral quantitative computed tomography (pQCT) In light of DXA’s limitations, the use of 3-D imaging tools is becoming more widespread in pediatric bone research. To date, pQCT has been used most frequently in research, as it provides a means to obtain valid and reliable measures of bone structure and estimates of bone strength in the growing skeleton with low radiation exposure. In addition, unlike DXA, pQCT can separate trabecular and cortical bone compartments at peripheral sites (distal and shaft sites of the radius and tibia) (137). In this section, I provide an overview of pQCT technology, image acquisition and analysis and common estimates of bone strength obtained using pQCT.  Briefly, pQCT consists of an X-ray tube that produces a narrow beam with a focal point of 250 x 250 μm. pQCT operates at 60 kV (0.3A) and uses an aluminium and copper filter which minimizes radiation exposure and assists to obtain a good contrast image (138). The diameter of the central gantry opening is 300 mm (XCT 3000 model) (Figure 1.12). The gantry rotates in 12o steps for a total of 15 translations and obtains a single image with a slice thickness of 2.5 mm (139). Attenuated images from pQCT are calibrated based on the density of water (60 mg of hydroxyapatite density) to obtain vBMD values (140). The effective radiation dose for a scout scan (to place the reference line) and a pQCT scan is   Chapter 1 – Introduction, Lit Review & Research Hypotheses 30  less than 0.1 μSv (139). As a system, pQCT is compact (1280 x 740 x 910 mm) and is easily portable (weight=90kg) (141).                   Figure 1.12. Image of peripheral quantitative computed tomography (pQCT), model XCT 3000 (Stratec Medizintechnik GmbH) and illustration of leg positioning in the pQCT. Illustration by Vicky Earle, Medical Illustrator, UBC, The Media Group.  1.2.3.2.1 Image acquisition  In contrast to DXA, pQCT image acquisition is not standardized and thus, there exists considerable variability in the current literature regarding acquisition protocols used in pediatric studies. The user can choose from various image acquisition parameters related to resolution, scan speed, reference line and scan sites. As per ISCD guidelines (137), it is vital that pQCT protocols in children and adolescents focus on the appropriate methods and calibrations to optimize results based on the research question being asked or the medical condition being evaluated so as to minimize radiation exposure.   Radiation exposure is related to the voxel (volume pixel) size used to capture the pQCT image and the scan speed. In general, a pQCT scan (per slice) takes about 3 minutes and has an effective dose of 0.43 µSv (142). Voxel sizes range from high (0.2 mm) to low (0.6 mm) resolution. High-resolution scans incur more time, thus increasing radiation exposure. Although a shorter scan time minimizes movement artifacts and effective radiation dose, scan quality is compromised using a lower resolution and a faster scan speed   Chapter 1 – Introduction, Lit Review & Research Hypotheses 31  (143). Several pediatric pQCT studies employed a minimum 0.4 mm voxel size and a scan speed at 30 mm/s to acquire quality pQCT images (30,32,144–146).   During scan acquisition, the choice of voxel size also influences cortical bone parameters (density, thickness) due to the partial-volume effect (PVE). The PVE refers to the presence of different densities of tissues (soft tissue and bone) in a single voxel (Figure 1.13) that results in an underestimation of true BMD. Rittweger and colleagues proposed correction equations (different detection thresholds) for the PVE based on cylindrical specimens of aluminum phantoms that corrected up to 80% of cortical density measured (147). Selecting the suitable threshold and analysis parameters, PVE can be minimized. I discuss this further in section 0.  Figure 1.13. Illustration of the partial volume effect (PVE) whereby the voxels at the bone edges (blue voxels) contain both bone and soft tissue densities, resulting in a lowered density for the blue voxels. Partially filled voxels will not be accounted for during the analysis. Adapted from Zemel et al. (148), with permission from Elsevier.  Locating scan sites, especially for longitudinal studies, must be consistent in order to attribute any changes accurately to medical conditions, interventions or growth. There is currently no consensus on the   Chapter 1 – Introduction, Lit Review & Research Hypotheses 32  most suitable site to assess cortical and trabecular bone in youth that are still growing (137). Manufacturer’s guidelines (149) suggest the reference line bisect (midline) of the most distal end of the growth plate (Figure 1.14, i) and trabecular bone be evaluated at the 4% site (of measured limb length) and cortical bone be evaluated at the 38% or 66% sites (149). However, studies have evaluated 3%, 4%, 7% and 8% for distal sites and 14%, 20%, 38%, 50% and 66% for shaft sites, in some cases using different reference lines (137). If the growth plate is fused, the manufacturer recommends the reference line bisects the mid-medial edge of the articular surface of the bone (Figure 1.14, ii) (149).  To address challenges associated with placing a reference line on a ‘moving’ growth plate and with accurately reproducing estimates of the mid-epiphyseal line; our research group adapted the protocol for our pediatric bone studies. That is, we position the reference line on the surface of a bony landmark that is reproducible and identifiable as children transition through maturity and into adulthood (i.e. the tibial plafond, Figure 1.14, iii) (150). To assess trabecular bone, we scan at a site 7% and 8% of the distance from this landmark, proximally, along the radius and tibia, respectively. Our measurement sites were also selected to ensure that the region of interest does not include the growth plate in most children (151). To assess cortical bone at the radius, we scan the 30% site from the medial border of the lunate fossa, also known as the medial edge of the distal radius. Ashe and colleagues demonstrated that in adults, cortical thickness at this site explained 80% of the variance in bone strength obtained from compressive failure load testing (152). At the tibial shaft, we assess cortical bone at the 50% site. Kontulainen and colleagues showed that in human cadaver bone, estimated bone strength (stress-strain index; SSIy) at the 50% site explained 80% of the variance in failure load (four-point bending model) (153). In comparison, the manufacturer recommended site (66%) explained 76% of the variance in failure load (four-point bending model) (153). It is recommended that muscle parameters be assessed at the 66% site as it is has largest MCSA, on average (154). However, we showed that muscle cross-sectional area (MCSA) at the   Chapter 1 – Introduction, Lit Review & Research Hypotheses 33  50% site is highly and systematically correlated with MCSA at the 66% site (r=0.95, n=20 boys and girls, age 9-11 years; unpublished data from the Healthy Bones Study (HBS)). By also acquiring muscle parameters at the 50% site, children were not subjected to an additional pQCT scan at the 66% site.  Bone outcomes from pQCT (BMC, bone strength, structure and density) at the radius and tibia shaft sites are reproducible (coefficient of variance (CV) 0.02 to 2.2%) (157) and precise (CV 0.4% to 3.5%) from repeated scans of human cadavers (158). However, at the distal radius and tibia sites, reproducibility was less precise. For example, precision for Tt.Ar at the radius was 0.9 to 2.3 (CV%) (157) and at the tibia precision was 1.8 to 7.9 (CV%). This is due to the presence of smaller trabeculae of bone that are susceptible to the partial volume effect (158).            Figure 1.14. Scout scans for reference line placements in pQCT scans. Manufacturer recommended reference line (A) placements based on growth plate if (i) open, at the outer most medial border, and (ii) closed, at the medial border of the distal endplate (shown as scout scans at the radius). An alternative placement of reference line (A) at (iii) the most distal edge of bone end/surface (in this example, at the tibial plafond). Adapted from Ashby et al. (for images i and ii) (155) with permission from Springer-Verlag London and Burrows et al. (image iii) (156) with permission from Elsevier BV.  iii A   Chapter 1 – Introduction, Lit Review & Research Hypotheses 34   1.2.3.2.2 Image analysis Similar to image acquisition, pQCT image analysis protocols are also not standardized. Users have various options they can choose from to analyse pQCT images, using either default settings or user-defined thresholds and modes. Briefly, the image captured requires a defined region of interest for analysis, which the user can determine automatically or manually. To separate soft tissue and bone, pQCT software uses two basic techniques: 1) voxels below a researcher pre-determined threshold (outer area) are removed, then an outer bone edge is formed from remaining voxels (Contour mode 1) and 2) the software identifies the first voxel containing the specified threshold, then finds the next voxel closest in a clockwise direction until a bone edge contour is defined (Contour modes 2 and 3). The Peel mode option then separates trabecular and cortical bone at distal sites. CALCBD analysis modes are used for distal sites that require trabecular and cortical bone separation while CORTBD analysis modes are used to analyze shaft sites that are predominantly cortical bone.  Figure 1.15. Illustration of polar stress-strain index (SSIp) calculated from peripheral quantitative computed tomography (pQCT) that estimates torsional strength along the z-axis of the bone shaft. CSMI, cross-sectional moment of inertia (mm4); d, distance of voxel area to the central axis; a, voxel area (mm2); Z, section modulus (mm3); dmax, maximum distance of voxel area to the central axis; voxel.Dn, voxel density; P.Dn, physiological “maximal” cortical bone density at 1,200 mg/cm3. CSMI = Σ(d2 x a) Z = Σ(d2 x a)/dmax SSIp = Σ(d2 x a x voxel.Dn/ P.Dn )                             dmax Torsion forces d dmax a z-axis   Chapter 1 – Introduction, Lit Review & Research Hypotheses 35  Using CALCBD and CORTBD analysis modes, we are able to derive bone parameters as listed in Table 1.2 and are able to estimate bone strength. At distal bone sites, compressive bone strength is estimated based on Hooke’s Law where stress is directly proportional to strain (69), and reorganized into the equation of Stress=F/A (153). As ultimate compressive stress is proportional to the square of total bone density (Tt.Dn2) (126), force (F) is directly proportional to Tt.Dn2 multiplied by Tt.Ar (perpendicular to the force direction). Thus, compressive bone strength can be estimated using the bone strength index (BSI (mg2/mm4) = Tt.Ar (mm2) * Tt.Dn2 (mg/cm3) /10,000) (159). Stratec software incorporates an analysis of bone strength for shaft sites called the strength-strain index that addresses torsion (Figure 1.15) and bending forces. The software uses measurements of cortical bone (area and density) to produce SSIp (strength in torsion) and SSIx/y (strength in bending) (153). 1.2.3.3 Strengths and limitations of pQCT Using pQCT to assess bone strength, structure and density in children and adolescents is safe, precise and reliable. However, pQCT image acquisition and analysis would benefit from standardized methodologies related to similar populations, interventions or medical conditions. It is otherwise not possible to compare results across studies due to differences in image acquisition parameters (i.e. scan sites, reference lines, image resolution) and image analyses (i.e. modes and thresholds) (137). As pQCT measurements are limited to the peripheral skeleton it is also not possible to assess some clinically relevant sites (proximal femur, spine) and bone parameters from one site cannot be ascribed to other sites. The exception is pQCT’s ability to assess the distal radius, a common fracture site in children at puberty. Further, it is not possible to accurately assess cortical density and thickness due to a PVE resulting from pQCT voxel sizes (>0.5 mm) (131,160). Despite these limitations, pQCT is able to provide structural assessment and volumetric density of cortical bone at the shaft and trabecular bone at distal sites with low   Chapter 1 – Introduction, Lit Review & Research Hypotheses 36  radiation exposure. Peripheral QCT also provides accurate and reliable estimations of bone strength in children and adolescents.    37  Table 1.2. Peripheral quantitative computed tomography outcomes, analysis sites and modes and description of bone outcomes. Adapted from Macdonald (161), with permission.  Outcomes (unit) Analysis site and mode Description Bone mass  Bone mineral content (BMC, mg/mm) Distal & shaft: Contour mode The amount of bone mineral within a cross-sectional slice of 1-mm thickness. Can be multiplied by the slice thickness to obtain total BMC in a 2.5 mm slice. Bone geometry  Total bone cross-sectional area (Tt.Ar, mm2) Distal & shaft: Contour mode The surface area of the entire bone cross-section including the cortex and marrow cavity. Tt.Ar directly reflects changes in bone size resulting from periosteal apposition.  Trabecular bone cross-sectional area (Tb.Ar, mm2) Distal: Contour mode, Peel mode The surface area of the trabecular bone cross-section. This area is influenced by endocortical apposition or resorption.  Cortical bone cross-sectional area (Ct.Ar, mm2) Shaft: Separation mode The surface area of cortical bone within the cross-section. Ct.Ar is influenced by periosteal apposition and endocortical apposition and resorption.   Cortical thickness (Ct.Th, mm) Shaft: Separation mode The distance between the outer and inner border of the cortical shell. Ct.Th can be determined with the circular ring model that assumes a circular cross-section or through an auto-detection of the outer border. Bone density Total density (Tt.Dn, mg/cm3) Distal: Contour mode Tt.Dn is the volumetric density averaged over the entire cross-section. It is influenced by the relative contributions and densities within both the cortical and trabecular bone compartments.   Trabecular density (Tb.Dn, mg/cm3) Distal: Peel mode Tb.Dn is the volumetric density averaged over the trabecular area of the cross-section. Tb.Dn is influenced by trabecular number and thickness and the degree of mineralization at the material level.    38  Table 1.2. continued  Outcomes (unit) Analysis site and mode Description Bone density (cont.) Cortical density (Ct.Dn, mg/cm3) Shaft: Separation mode Ct.Dn is the volumetric density averaged over the cortical area of the cross-section. Ct.Dn is influenced by cortical porosity and the degree of mineralization at the material level.  Bone strength  Cross-sectional moment of inertia (CSMI, mm4) Shaft: Separation mode CSMI is proportional to the distribution of bone mass about the neutral axis and is an indicator of bone strength in bending or torsion (depending on which axis is used as a reference). It is calculated as the integral sum of the products of area (A) of each voxel and the squared distance (d2) of the corresponding voxel to the bending (x, y) or torsion (z) axes.   Section modulus (Z, mm3) Shaft: Separation mode Z is CSMI divided by dmax, the maximum distance from the bending axis to the outer surface, in the plane of bending. Thus, Z approximates a cross section’s resistance to bending in a given plane.   Strength-strain index (SSI, mm3)  Shaft: Separation mode SSI is calculated as the integrated product of Z and Ct.Dn. The ration of Ct.Dn and normal physiological density (ND = 1200 mg/cm3) provides an estimate of the modulus of elasticity. Similar to CSMI, SSI can be determined with respect to the polar (z) axis or the bending (x, y) axes.  Bone strength index (BSI, mg2/mm4) Distal: Contour mode At distal sites, compressive strength is estimated as the square of the total density and the total cross-sectional area (the load-bearing area).  Chapter 1 – Introduction, Lit Rewiew &Research Hypotheses 39  1.2.3.4 Other bone imaging technologies Other bone imaging technologies used less frequently in pediatric bone research include HR-pQCT, MRI and QUS. I provide a brief overview of these three technologies below.  Similar to pQCT, HR-pQCT provides a 3-D analysis of bone microstructure and density. It is an accurate (151,162), reproducible (117) and safe (effective dose of 3 μSv for a single scan (163)) approach to assess these parameters in children. HR-pQCT scans encompass a bone cross-section with a thickness of 9.02 mm region of interest. Depending on the resolution desired (41, 82 or 123 μm), the region of interest is comprised of a specific number of slices – 220, 110 or 73 slices, respectively (164). From my review of literature, the most common HR-pQCT protocol uses a resolution of 82 μm and scans a total of 110 slices. In addition to the bone outcomes generated from pQCT, HR-pQCT also assesses cortical bone porosity and trabecular microarchitecture (162) (trabecular thickness (mm), number (1/mm) and separation (mm)) (164). High-resolution images enable post-acquisition finite element analyses (FEA), that simulate bone loading to provide an accurate estimate of bone strength (162). However, due to the high resolution and longer scanning time, HR-pQCT scans are more sensitive to movement artefact compared with pQCT (137). A short gantry also limits the measurement of more proximal peripheral sites (165). Lala and colleagues assessed cadavers (age unspecified) and compared tibiae outcomes obtained from HR-pQCT and pQCT at the same bone site and relative slice thickness (the first 27 slices acquired using HR-pQCT) (165). Cortical area, thickness and density were notably different (but highly correlated r = 0.96 – 0.99) at the 20% site of the tibia. These differences increased with greater cortical thickness, regardless of voxel size used in pQCT (0.2 or 0.5 mm) (165). Thus, it is important to consider the pQCT image acquisition protocols and the type of analysis conducted if bone outcomes are to be compared with results from HR-pQCT. Chapter 1 – Introduction, Lit Rewiew &Research Hypotheses 40  Two non-ionizing bone imaging technologies used to assess bone in children and adolescents are MRI and QUS. MRI bone images are created when fat and water (found in muscle mass) protons are excited after subjected to magnetic forces (bone does not have free protons) and a ‘picture’ (image) forms that depicts various tissues (166). In general, MR image acquisition varies based on image weightage (T1 to T2), pulse sequence (e.g., gradient or spin echo) and magnet strength (1.5 – 11.7 Tesla (T)). A 1.5 T magnet generates a resolution of about 300 μm (166). With stronger magnets, higher resolutions ranging from 150-200 μm are possible. MRI is used to assess bone structure and strength (but not density) and to assess other components such as water, fat and cartilage (e.g., erosions, effusion) (167). Imaging of bone using MRI is limited to peripheral sites such as the distal radius, humerus, distal tibia and calcaneus (168) and proximal femur (169). MRI is unable to evaluate the axial skeleton as there is insufficient contrast due to the higher volume of hematopoietic marrow (166). Cross-sectional geometry outcomes obtained using MRI to assess the 50% tibia site of 68-80 year old cadavers, was comparable to values obtained using pQCT (r = 0.55 to 0.85) (170).   QUS is the least expensive bone imaging tool among approaches I present. QUS provides quick scan times, is highly portable and does not require technical expertise (166). QUS measures the distance and speed that ultrasound waves travel from a source to a detector. As ultrasound waves are completely attenuated by air (166), QUS cannot examine axial sites as results are compromised by air in the lungs and bowels. Bone sites most commonly examined are calcaneus, tibia, radius and phalanx (123). Primary outcomes from QUS are speed of sound (SOS) and broadband attenuation (BUA). More dense bone results in greater ultrasound wave attenuation and loss of energy (slower velocity). Therefore, QUS outcomes are mainly linked to BMD and provide little information about bone structural changes (123). However, the central limitation of QUS is a fundamental lack of understanding of the relation of QUS outcomes to bone material and structural parameters. Thus, new paradigms are proposed for further research using QUS (171). Chapter 1 – Introduction, Lit Rewiew &Research Hypotheses 41  1.2.4 Growth and maturation  In this section, I define the terms childhood and adolescence that I have adopted in this thesis. As there are differences in chronological and biological age within an individual, I discuss various methods to assess biological maturation. Biological maturation is closely related to bone growth and development in children and adolescents.  1.2.4.1 Definition of adolescence In order to define adolescence, I first provide a definition of childhood. Childhood is defined as the developmental period after the end of infancy and extends to the beginning of adolescence, whereby there is no development of secondary sexual characteristics. Chronologically, childhood extends until age 11 years in girls and 13 years in boys (172). Girls and boys are considered adolescents from age 12 to 18 years and 14 to 18 years, respectively (172). However, I adopt the definition from the Canadian Pediatric Society, which states that adolescence begins with the onset of physiologically normal puberty and ends when adult status is achieved, typically the life period between childhood and adulthood (173). Therefore, adolescence is more challenging to identify due to different maturational timing among individuals. 1.2.4.2 Assessment of puberty Maturation is a developmental process to reach adulthood and can be divided into several domains such as physical, sexual, intellectual, emotional and even social maturation (174). Puberty consists of physical and sexual maturation that are observed through skeletal and somatic (body size) growth and development of secondary sexual characteristics (175). As bone development is more closely related to pubertal changes than chronological age, I discuss common methods to assess pubertal development in the following sections.   Chapter 1 – Introduction, Lit Rewiew &Research Hypotheses 42  1.2.4.2.1 Skeletal maturation Skeletal age, or bone age, is traditionally assessed by clinicians by comparing hand radiographs (length, width, ratio) and fusion of the epiphyseal plate to a set of sex-specific atlases with ages ranging from birth to adulthood (176). The three most popular methods are Greulich-Pyle, Tanner-Whitehouse (three versions) and Fels, which differ based on the set (and number) of hand-wrist bones assessed, scoring procedure and reference sample (177). Overall, skeletal age assessments are precise and reliable (176) but the outcomes are not comparable across methods (178). Skeletal age measurements are usually used in clinical setting, as the process itself requires highly trained personnel to obtain and interpret the radiographs. Furthermore, it involves exposure to ionizing radiation. The reference radiographs were mainly derived from a Caucasian/white cohort, which may not be suited to assess skeletal age in other ethnicities due to known ethnic differences in maturation (179).  1.2.4.2.2 Somatic maturation In relation to skeletal growth, maturation can also be assessed based on somatic changes. For example, individual growth trajectories can be mapped with data from longitudinal studies where serial measurements of height are available. Age at PHV (APHV), an indicator of somatic maturation, can be calculated from such trajectories (Figure 1.16) (11,180). Recent Canadian data suggests that APHV occurs around 11.8±1.0 years in girls and 13.5±1.0 years in boys (114). Approximately 90-92% of adult height is achieved by APHV (9) with average gains of 8.5±1.1 cm/year and 10.4±1.3 cm/year in girls and boys, respectively at APHV (181). In relation to sexual maturation, girls reach PHV one year prior to menarche (first menstrual period), on average (182). Importantly, data from the University of Saskatchewan 7-year longitudinal Pediatric Bone Mineral Accrual Study (PBMAS) reported that peak BMC accrual followed PHV by about 7-8 months, on average (11). It has been proposed that the lag between PHV and peak BMC accrual creates a temporary period of low bone mass and transient bone fragility (183). Chapter 1 – Introduction, Lit Rewiew &Research Hypotheses 43  Epidemiological studies indicate that this corresponds to a period of peak fracture incidence in adolescent girls and boys (184–188).   Figure 1.16. Total body bone mineral content (BMC) gain in boys and girls from longitudinal DXA data (11). Age of peak lean mass (LM) (181) and BMC gains in relation to age at peak height velocity (APHV) in boys and girls. The lag period between PHV and peak BMC (shaded area) is approximately 8 months and represents a period of low bone mass for age. Adapted from Bailey et. al, 1999 (11), with permission from the American Society for Bone and Mineral Research.  Assessment of somatic maturation is non-invasive and easy to conduct (i.e. no costly equipment or highly skilled assessors are required), but it relies upon longitudinal data and is therefore time consuming and costly to follow a cohort over a number of years. To obtain PHV, a child’s height is assessed from prepuberty, through puberty, to maturity. From these data height velocities can be calculated and the growth trajectory can be mapped. Accurate measurement of height across time requires Girls’ PHV = 11.8 yrs Boys’ PHV = 13.5 yrs Boys (solid line):  Age of peak LM gain: 13.8 yrs Peak LM: 1128 g/yr  Age of peak BMC gain: 14.1 yrs Peak BMC: 409 g Size adjusted peak BMC: 394 g/cm2  Girls (dash line):  Age of peak LM gain: 12.2 yrs Peak LM: 507 g/yr  Age of peak BMC gain: 12.5 yrs Peak BMC: 325 g Size adjusted peak BMC: 342 g/cm2  Chapter 1 – Introduction, Lit Rewiew &Research Hypotheses 44  standardized protocols and strict quality control. Longitudinal measurement also requires prolonged commitment from study participants, and thus, is often subject to high attrition rates. If it is not feasible to collect serial measurements across growth, a model by Mirwald and colleagues that predicts sex-specific APHV has been proposed using single measures of height, sitting height, leg length, body mass and chronological age (189). The authors cross-validated the equation with two different sets of longitudinal data and set a limit to predict the mean up to a year surrounding APHV (mean at zero, SD of 0.5 yrs). The equation predicted 89% and 88% of variance in APHV from the other datasets for boys and girls, respectively (189). However, independent assessment of the equation compared with actual APHV with skeletal age using the Fels method showed poor associations with maturity status in boys aged 13-14 years (Cohen kappa = 0.13, 57% agreement, Spearman’s r = 0.29) (190). Differences between predicted versus actual APHV increased with chronological age (and age from actual APHV) and tended to differ between early and late maturity groups (191,192). When tested, the predicted APHV tends to overestimate actual APHV as chronological age increases; the largest difference was observed in 11 to 13 year old girls (+0.4 to +0.6 years) (192) and 13 to 15 year old boys (+0.2 to +0.3 years) (191). As the equation was derived from a white Canadian cohort, they may have different body proportions (leg length, sitting height, etc.) compared with adolescents of other ethnicities. For example, Asians tend to have shorter limbs than their white counterparts (193,194). 1.2.4.2.3 Sexual maturation Sexual maturation includes adrenal and gonadal maturation. Adrenarche is responsible for maturation of the gonads and pubic and axillary hair growth and contributes to behavioural and psychological changes (195). Subsequently, gonadal maturation leads to maturation of primary sex organs (ovaries and testes) and secondary sexual characteristics (breasts and genital development) through Chapter 1 – Introduction, Lit Rewiew &Research Hypotheses 45  increased levels of estrogen or testosterone (195). Thus, assessment of secondary sexual characteristics or measures of hormone levels can be used to estimate a child or adolescent’s stage of maturity.  A popular method for assessing sexual maturation in children and adolescents is by the method of Tanner (195) that classifies secondary sexual characteristics into five distinct stages. When assessed by a qualified physician, the Tanner method is considered a gold standard (175). Children and adolescents are classified as pre-pubertal (Tanner stage 1), early pubertal (Tanner stage 2 and 3), peri-pubertal (Tanner stage 4) or post-pubertal (Tanner stage 5) based on standardized photographs or line drawings depicting the stages of secondary sexual maturation (breast and genital development, pubic hair growth). This is based on the work of Marshall and Tanner who studied a cohort of British children (192 girls, 228 boys) in the Harpenden Growth Study (196,197). Marshall and Tanner reported differences in both the timing of pubertal onset (anywhere between 8 to 15 years old) and the tempo of maturation. On average, a boy’s pubertal development takes about 1.8 to 4.7 years to complete (197) while girls take anywhere from 1.5 to 6 years (196). Figure 1.17 illustrates the wide variation in maturation for girls at a certain chronological age, where one stage of maturation (e.g., onset of puberty) may overlap with other maturational stages (Tanner 2 to 5) of different individuals of the same age. To illustrate, a 10-year old girl might be Tanner stage 1 or Tanner stage 3 for breast development depending on the tempo and timing of her maturation. Boys’ timing of maturation follows a similar trend as to girls, albeit with a lag time of one to two years compared with girls (175). Boys’ tempo of growth is of greater magnitude especially at puberty, on average (198)   Chapter 1 – Introduction, Lit Rewiew &Research Hypotheses 46    Figure 1.17. This schematic illustrates the wide variation in maturation phases as assessed by the method of Tanner across pubertal development (i.e. adrenarche and gonadarche) in girls relative to chronological age. Reproduced from Dorn et al. (175), with permission from Taylor & Francis Informa UK Ltd.  When physician-determined Tanner staging is not feasible (e.g., due to the intrusive nature of assessment, field-based research setting, resource limitations), self-assessment methods are often used. During self-assessment, participants select the closest description and picture (or line drawing) that represents their current stage of secondary sexual maturation (175). There are moderate to high correlations (0.48-0.91) between self-assessed and physician-assessed maturation with stronger correlations in normal body mass girls compared with boys (174,199). However, self-assessment may be less accurate in overweight and obese individuals, particularly in girls, as excess adipose tissue could be mistaken for breast development. Obese girls’ self-assessment of breast development was only moderately correlated (r=0.37) with physician assessment (200). Due to the sensitive nature of the Tanner method (i.e., physical examinations or graphic depictions of genitals and breasts), this method may not be appropriate in all research settings (e.g., field studies in Chapter 1 – Introduction, Lit Rewiew &Research Hypotheses 47  schools) (175). Thus, self-assessments of secondary sexual characteristics such as axillary and facial hair growth are an alternative especially for boys; however, results are less precise to the exact stage of maturation. Self-assessments of axillary hair in boys were significantly correlated with physician assessments for the appearance of axillary hair (r=0.68) (201). In a 6-year longitudinal study, 77% of boys were in Tanner stage 4 of pubic hair growth by the first appearance of axillary hair (202). In addition, age of first appearance of axillary hair in boys was moderately correlated with Tanner stage 3-5 of pubic hair growth (r=0.55-0.66) and to APHV (r=0.51) (202). For girls, although self-assessments of axillary and leg hair growth were significantly correlated with physician assessments (r=0.74 and r=0.44, respectively), menarcheal status may be a more reliable indicator of sexual maturity (195). Menarche is considered a late event in sexual maturation; at menarche, 60% of girls reported achieving Tanner stage 4 and 10% were at Tanner stage 5 for breast development (203). Mean age at menarche (12.7±1.0 years) also coincided with the mean age at peak BMC accrual velocity (12.6±0.86 years) (11). This close association between peak BMC accrual and age at menarche provides a suitable means to evaluate maturity-related changes in bone variables in girls. Menarcheal age is usually self-reported retrospectively and therefore, accuracy of reporting depends on time since menarche (204–206). For example, among post-menarcheal girls, 66% correctly reported the month and year of menarche after an average of 323 days (205). However, accuracy decreased with time; 59% reported correctly after an average of 430 days, while only 45% correctly reported after an average of 649 days (205). Recall can be refined if prompted by occasions or conditions related to the occurrence of menarche. For example, asking about school grade, season (Spring, Summer, Fall, Winter), and/or events (e.g., Christmas, birthday, school holidays) may assist in determining the exact timing of menarche. Hormonal assays to assess sexual maturity are challenging as they most often require a blood draw and therefore may be unappealing to healthy children. There are also only limited reference data for Chapter 1 – Introduction, Lit Rewiew &Research Hypotheses 48  children and adolescents. Although less invasive methods (e.g., saliva or blood spot) are available, the results from assays of different types (saliva, blood spot, venous blood) are not comparable and are not linked to Tanner stages (195). In addition, hormone levels follow a diurnal pattern and fluctuate by menstrual cycle phase, thus consistent timing is required when obtaining samples (195). Future advances in biochemical assays may enhance assessment of maturation using hormonal assays (195).  Overall, assessment of maturation can vary across research studies. The method selected depends largely on the aim of the study. The choice of assessment also has to be feasible based on study design (cross-sectional, longitudinal), setting (clinical, school, home) and available expertise (trained staff, health professional). A common limitation across methods is the dependence on mostly white cohorts for reference data. Yet, assessment of maturity in pediatric bone research is vital to isolate contributions of natural progression of growth from exercise and nutritional factors. In a recent study, Rantalainen and colleagues found marked site-specific sexual dimorphism of the tibia (by pQCT) and attributed these bone changes to the natural progression of growth (before and after APHV), and not from other factors such as exercise and nutrition (207). 1.2.5 Maturity- and sex-specific differences in bone  In this section, I provide an overview of the role of maturity in bone growth and development and the sex-specific differences in bone strength, structure and density. First, I discuss the findings from a seminal longitudinal study that followed children throughout growth into adulthood (11). This study provided great insight regarding differences between boys and girls in the tempo and timing of bone mineral accrual across maturity stages. Then, I focus on studies that utilized 3-D imaging tools studies to evaluate sex- and maturity-specific differences in bone strength and its contributors (structure and density) at weight-bearing and non-weight bearing sites.  Chapter 1 – Introduction, Lit Rewiew &Research Hypotheses 49  1.2.5.1 Sex differences in bone mass  Bone science has benefitted tremendously from longitudinal studies that observed the process of bone development from childhood into adulthood. The PBMAS followed children from age 8 to 14 years, acquired annual DXA scans for seven consecutive years and followed up participants five years later, in adulthood (11). Findings from the PBMAS demonstrated that not unlike height, the magnitude of BMC accrual for the total body and proximal femur was 15% greater in boys (394 g/year), on average, compared with girls (342 g/year) at peak BMC velocity (PBMCV) and during the two subsequent years (11). This was not the case at the lumbar spine where there were no between-sex differences in PBMCV or accrual during the 2 years before and after APHV (11). Evidently, there is site-specificity of bone mineral accrual between boys and girls. Overall, total body and femoral neck BMC is greater in boys compared with girls; these differences are related to stature (208). Also from the PBMAS study, girls who matured later (about a year later as assessed by APHV) had significantly less total body BMC than girls considered ‘average maturers’ when compared within chronological age categories. There were no differences for boys among early, average and late maturers for BMC accrual at the femoral neck or lumbar spine sites (209). This illustrates the influence of the timing and tempo of maturation on bone – and the need to account for maturity in studies of growing children.  1.2.5.2 Sex differences in bone size and strength Sex-differences in bone size and strength can be examined through changes on the surfaces of bone. Classic cross-sectional studies by Garn and colleagues examined radiographs of the second metacarpal in children from birth to adulthood (80 years old). They reported differences between boys and girls at the periosteal and endosteal surfaces by chronological age (210,211). They concluded that sex Chapter 1 – Introduction, Lit Rewiew &Research Hypotheses 50  differences in bone area stemmed from greater apposition at the endosteal surface in adolescent girls compared with boys while boys experienced greater periosteal apposition compared with girls (210,211). Our bone research group readdressed this finding using pQCT (Gabel et al., 2015; submitted for publication) to assess the tibia of boys and girls across age 8 to 23 years and accounted for maturity using APHV. The findings supported those of Garn and colleagues (210,211) whereby periosteal apposition of bone in boys was greater than for girls. However, at the endosteal surface, a decrease in endosteal resorption (rather than increased bone apposition) in girls accounted for sex differences in the medullary area. Early planar-based imaging technologies offered a means to evaluate maturity- and sex-related differences in bone. However, the evolution of 3-D imaging (pQCT, HR-pQCT, MRI) provides a better means to evaluate the structure, strength and surfaces of bone.  1.2.5.3 Sex differences in bone at weight-bearing bone sites  Maturity- and sex-specific differences in bone strength, structure and density were reported in studies of children and adolescents that used pQCT to assess these bone outcomes. In pre- and peri-pubertal boys, BSI at the distal tibia was greater (+29%) compared with girls at the same stage of maturity (150). This may be attributed to a greater Tt.Ar and higher Tt.Dn in pre-and peri-pubertal boys compared with girls (150,212). BSI (+15%) and Tt.Ar (+6%) remained greater for boys after adjusting for tibial length and MCSA (150). In post-pubertal adolescents, there were no differences between sexes for bone strength and density at the 4% distal tibia. However, Tt.Ar and trabecular BMC (Tb.BMC) were higher (unadjusted) in boys compared with girls (213). In a study of 8 to 15 year olds, boys had higher Tb.Dn at the 4% site compared with girls (adjusted for height, Tanner stage and age) (214). This suggests that the strength advantage conferred to boys compared with girls at the distal tibia is a result of increased bone area (size) and trabecular density.  Chapter 1 – Introduction, Lit Rewiew &Research Hypotheses 51  At the tibial shaft, boys had stronger bones as a function of their greater bone area, on average. Our research group reported that boys had greater bone strength (6%, SSIp (150) and 14%, Z (215)) at postpuberty (only), as compared with girls. Importantly, peri- to post-pubertal girls demonstrated greater cortical bone density (1% to 10%) at shaft sites compared with boys at the same maturation stage (150). These sex-specific differences in cortical density persisted after adjusting for tibial length and MCSA (150). For girls, greater bone density at the tibia shaft may be one means whereby they achieve bone strength when their bones have stopped growing in size by depositing bone endosteally. In a study of 13-year old black youths, boys had greater bone strength (12%, SSIp) at the 38% tibia compared with girls (212). This was a function of black boys greater cortical bone diameter while black girls had greater cortical density (212). Relatively, increases in bone size at shaft sites or more accurately, the displacement of bone away from the neutral axis provides exponential (third or fourth power) increases to bone strength (153) compared with bone density (second power) at distal sites (159). Two other studies reported greater bone area in boys compared with girls (213,216); however, no bone strength or density variables were assessed (216). A larger medullary area in boys compared with girls observed at early puberty was maintained into young adulthood (213,216). These sex-specific adaptations are related, in part, to the role of sex hormones. I discuss this role in greater detail in section 1.2.6.3. Adjusting for different covariates when evaluating sex-specific differences in weight-bearing bones could provide different results. Hölger and colleagues used MRI to examine the femoral shaft (66% site) in prepubertal boys and girls and young adult men and women. After adjusting for femur length and body mass, the bone outcomes (bone area, BMC and bone strength) were not significantly different between boys and girls. These results at the femur differ from the pQCT studies mentioned previously; this may be due to differences in variables controlled for. In young adults, men (21.4±2.4 years) had significantly greater total and cortical area (5-6%), greater BMC (3%) and greater bone strength (polar second moment of area (Ip), BSI, 6-7%), compared with women (20.4±3.2 years) (217). When femur Chapter 1 – Introduction, Lit Rewiew &Research Hypotheses 52  length and muscle area were controlled for in the analyses, bone differences diminished or favoured the young women. Young women demonstrated trends toward stronger bones, thicker cortices and larger medullary area, compared with young men relative to their body size and lean body mass.  1.2.5.4 Sex differences in bone at non-weight bearing bone sites Three pQCT studies (212,218,219) and one HR-pQCT study (220) examined sex differences in non-weight bearing bones in children and adolescents. At the distal radius, two studies examined bone strength and reported boys had greater bone strength (BSI and by FEA) compared with girls from pre- to post-puberty (212,220). Boys’ greater BSI in pre-and peri-puberty was attributed to their greater Tt.Ar, Tt.Dn and Tb.Dn, compared with girls’ (212). Neu and colleagues investigated maturity- and sex-related bone differences at the distal radius in a cohort aged 6 to 23 years using pQCT (218). Boys at Tanner stages 1, 2 and 5 had greater total bone area compared with girls (218). Boys at Tanner stage 1 (only), had greater Tt.Dn compared with girls (218). Kirmani and colleagues used HR pQCT to examine the distal radius (220). Boys had wider bones (larger periosteal and endosteal circumference) compared with girls during pre- (stage 1, assessed by Tanner-Whitehouse III) and post-puberty (stage 4 and 5), but not at stages 2 and 3 (220). Boys also had thicker cortices and greater Ct.Dn at stages 2 and 3 compared with girls. At stage 4, girls had thicker cortices and greater Ct.Dn (220). Thus, greater bone strength (by FEA) in boys is conferred by their larger bone size, achieved via increased periosteal apposition and higher trabecular volume. Sex-related differences in bone structure and density at non-weight bearing distal sites were highly associated with maturity. Variability across studies may be partially explained by differences in the site examined, different approaches to assess maturity and different measurement protocols.  The Dortmund Nutritional and Anthropometric Longitudinally Designed (DONALD) study assessed the radial shaft (65% site) using pQCT. Their cohort was comprised of 371 healthy 6 to 23 year Chapter 1 – Introduction, Lit Rewiew &Research Hypotheses 53  olds and their parents (n=107), aged 29-40 years (219,221,222). Boys had significantly greater bone strength (+16-25%, assessed as SSIp) compared with girls at all Tanner stages (except Tanner stage 4 where no differences were observed) (221). For Ct.Ar, boys had significantly larger bone area compared with girls at Tanner stages 1 (+11%) and 5 (+16%). Boys also had significantly greater marrow area (Me.Ar) compared with girls at Tanner stage 3 (+36%) and 5 (+26%) (219). In the same cohort, girls had greater Ct.Dn compared with boys at Tanner stages 4 (3.2%) and 5 (+3.5%) (222). Thus, boys’ greater bone strength (SSIp) compared with girls across all Tanner stages (except Tanner stage 4) may be a function of boys’ larger cortex (Tanner stage 1 and 5) or Me.Ar (Tanner stage 2 and 5). These bone parameters (Ct.Ar and Me.Ar) were significantly different at certain maturational stages. However, small size differences have an exponential influence on bone strength, which may have contributed to boys’ greater bone strength as compared with girls. Girls’ greater bone density as compared with boys at Tanner stage 4 did not confer girls greater bone strength. Importantly, DONALD study data were not adjusted for body size or any muscle parameters.  Generally, a few cross-sectional and prospective studies support maturity- and sex-specific bone strength differences in growing children. Differences appear to be a function of 1) assessment of weight-bearing versus non-weight bearing bones, 2) maturity stage and 3) site assessed (distal or shaft). Most sex differences in bone strength were explained by differences in bone structure rather than bone density. The magnitude of the difference between sexes increased as maturity progressed. Importantly, possible confounding variables (e.g., muscle, body size) should be controlled as they can otherwise drive the outcomes observed.    Chapter 1 – Introduction, Lit Rewiew &Research Hypotheses 54  1.2.6 Factors that influence bone growth and development In this section, I introduce factors that influence bone growth and development (Figure 1.18). Specifically, I aim to provide a brief overview of inherent and extrinsic factors that influence bone strength, structure and density in children and adolescents. Inherent factors known to influence bone growth and development include genetics, ethnicity, hormones and muscle mass. Extrinsic factors include PA and dietary intake of calcium. As the influence of PA on bone is a central tenet of my thesis, I discuss this association in greater detail in section 1.2.8.  Figure 1.18. A functional model of bone development based on the mechanostat theory where bone continuously adapts to external challenges with the known modulators.  Reproduced from Rauch and Schoenau (223), with permission from Lippincott, Williams & Wilkins, Inc.  1.2.6.1 Genetics  From family and twin studies, genetics accounted for 50-80% of the variability in aBMD (224–226). The heritability, i.e. inherent traits, of bone strength and structure are less well defined. Several retrospective studies used pQCT to assess adult bone strength, structure and density and demonstrated that Chapter 1 – Introduction, Lit Rewiew &Research Hypotheses 55  body length and body mass assessed in infants at 6 months to 1-year old, significantly predicted bone outcomes (r=0.12- 0.56) in adults (213,227–229). A prospective study that followed individuals across three generations (first generation: 72 to 96 years old, second generation: 40 to 80 years, third generation: 31 to 72 years) (230) and a study in twins (3,724 monozygotic and 12,050 dizygotic twin pairs, age 66 to 68 years) (231), reported that 33 to 53% of vertebral fractures were linked to genetics. While certain traits related to risk factors for fractures may be inherited, the traits may be different from genes that contribute to bone phenotype (structure, size). However, heritability of bone strength is practically unknown although several genes and loci have been associated with some bone structural components (232).      Genome-wide analysis studies (GWAS), i.e. genetic epidemiology, search widely across complete genomes to isolate single nucleopeptides (SNPs) that may be the cause of certain traits. From GWAS, researchers using pQCT and HR-pQCT, identified new loci associated with cortical and trabecular BMD, cortical thickness and bone microstructure (232–234). From large population cohort trials including the Gothenburg Osteoporosis and Obesity Determinants (GOOD) study, Avon Longitudinal Study of Parents and their Children (ALSPAC), Osteoporotic Fractures in Men (MrOS) and the Canadian MultiCentre Osteoporosis (CaMos) study, genotype samples were pooled to identify possible loci and single nucleotide proteins (SNPs) related to specific bone variables such as BMD and cortical thickness (233). Subsequently, the meta-analysis reported several SNPs associated with the Wnt16 gene (233). An example of the Wnt16 gene function is to produce receptor activator nuclear-factor kappa-β ligand (RANKL) protein that promotes differentiation and maturation of osteoclasts (233). More studies are on-going to identify the important role of genes in bone phenotype and even bone adaptation. On top of genetic influences on bone, bone is also moulded by the environment. In a Swedish cross-sectional study of twins, environmental factors explained one third of the variance in vertebral fractures that occurred between age 50 to 70 years and up to 85% of the variance in vertebral fractures for Chapter 1 – Introduction, Lit Rewiew &Research Hypotheses 56  those age 80 years and above (231). In another cross-sectional study (226), we can see that inherent properties of height, body mass and bone density are related between mothers and daughters. Daughters by age 12.8±0.8 years, had 96% and 76% of their mothers’ (age 42.4±4.2 years) height and weight, respectively (226). In addition, proximal femur aBMD of daughters was 56% and lumbar spine aBMD 70%, of their mothers’ values (226). Mother-grandmother (grandmothers, 67.6±8.8 years) spine aBMD showed 66% heritability (226). There was no relation between intake of dietary calcium (across generations) and aBMD; however, low PA levels in daughter-mother-grandmother triads were associated with low aBMD and high PA levels in daughter-mother-grandmother triads were associated with greater aBMD at the femoral neck (226). This suggests that environmental factors may interact with heredity to influence bone growth and development.  1.2.6.2 Ethnicity and race Ethnicity dictates through genetics the hair, eye and skin colour, and body build of an individual. Externally, it shapes their social-cultural diet, norms and behaviours (235). Thus, ethnicity can influence bone growth and development through various channels. To begin, I use the term ethnicity to mean hereditary and sociocultural influences while race is based on phenotype classified by a social construct prior to modern genetic studies (236). For example, the Caucasian race is identifiable with light skin and eye colour but may have stemmed from different ethnicities such as Irish, Dutch and Finnish. These terms are used interchangeably in the current bone literature. For my literature review I use terms used by authors of the studies I review; I include definitions if provided by authors. I discuss pediatric studies that are mainly cross-sectional association studies that examined the influence of ethnicity on bone health.  Studies of bone traits and fracture risk in children demonstrated distinct ethnic differences in bone structure and density, irrespective of geographical environment. Fracture risk of US white children was Chapter 1 – Introduction, Lit Rewiew &Research Hypotheses 57  twice (hazard ratio = 2.1) that of non-white children (aged 6-17 years) (237). Generally, non-white American children (i.e. blacks, Hispanics) had significantly stronger tibiae and radii (3-47%, BSI, Z, SSIp) and greater Tt.Dn and Ct.Dn (2-23%) at distal and shaft sites (assessed using pQCT), compared with white children (controlled for age, sex, PA, muscle mass and limb length) (238–240). Our research group used HR-pQCT to examine ethnic differences in a cohort aged 14 to 22 years. Asians (East, Southeast and South Asians) had significantly smaller bone area and size compared with white (North America or European ancestry); yet, Asians had greater cortical BMD (5-8%), thicker cortices (12-27%), less cortical porosity (25%, boys only), lower trabecular number (8%, girls only) and greater trabecular separation (10%, girls only) than did whites after accounting for age, lean mass, limb length, calcium intake, PA and age of menarche (girls only) (193). Bone strength did not differ between these groups (193).  Certain factors other than genetic programming may contribute indirectly to the observed ethnic differences in bone properties in children. One factor would be pubertal timing that is different between ethnicities. Non-white girls (i.e. black, Chinese) have an earlier onset of maturation compared with white girls, on average. The mean age at menarche in Asian, black, Hispanic and white girls was 12.0, 12.1, 12.2 and 12.7 years, respectively (179,193). This relates back to the influence of maturity on bone development as discussed in section 1.2.5.  Another ethnic difference related to bone development outside the control of genetics would be calcium intake, retention and resorption. For instance, Asian children (10 years old) and adolescents (16 years old) had a lower calcium intake than their white counterparts; however, calcium intake did not contribute to total body, total hip, femoral neck and lumbar spine and BMC in either ethnicity (241). Regarding calcium retention, black girls demonstrated higher calcium retention and absorption efficiency compared with white girls, on average, across a wide range of calcium intake levels (760-2195 mg/day) (242). I discuss the role of calcium on bone development in further detail in section 1.2.6.4.  Chapter 1 – Introduction, Lit Rewiew &Research Hypotheses 58  There may also be differences in muscle parameters between ethnicities – and this could influence ethnic differences in bone. For example, white boys and girls have significantly more lean mass than black, Asian and Hispanic children (243–245). Differences in muscle mass are commonly related to body size. Muscle CSA (by pQCT) at shaft sites of the radius and tibia was significantly larger in blacks compared with whites and Hispanics but these differences were no longer significant after adjusting for limb length (240). Yet, differences in bone outcomes between these ethnic groups remained after controlling for muscle (MCSA) in the analysis (193,240). I discuss the interaction of muscle and bone in detail in section 1.2.6.5. 1.2.6.3 Hormones  Endocrine and paracrine pathways that regulate bone growth and development are documented in the literature but how pathways are triggered at different phases of life is not well understood. In this section, I discuss the key role of growth hormone (GH), insulin-like growth factor-1 (IGF-1) and sex hormones in bone growth and development.  1.2.6.3.1 Growth hormone and IGF-1  GH and IGF-1 regulate metabolism and human growth across the lifespan. Levels of these key hormones remain relatively constant except during periods of accelerated growth (i.e. infancy and puberty). Secretion of GH from the anterior pituitary glands is triggered by growth-hormone releasing hormone (GHRH) from the hypothalamus. Higher levels of GH promote maturation of gonads (i.e., testes, ovaries) (246,247) and, in turn, estrogens and androgens secreted from the gonads further up regulate GH levels (248). During puberty, GH levels increase 1.5- to 3-fold from the otherwise constant levels observed during growth (249). Secretion of GH also stimulates release of IGF-1 from cartilage and hepatic Chapter 1 – Introduction, Lit Rewiew &Research Hypotheses 59  sources. This cascade of hormones orchestrates increases in height and bone mass as depicted in Figure 1.19.  Figure 1.19. This schematic depicts trends in hormonal levels, longitudinal growth (height velocity) and bone mineral accrual (bone mineral content (BMC) velocity) in girls in relation to maturation stages and chronological age. Boys display similar patterns of hormonal and bone growth but with a delay of about a year later than girls to attain peak height and BMC levels. Reproduced from MacKelvie et al. (250), with permission from BMJ Publishing.  GH acts directly on growing bone by interacting with prechondrocytes at the epiphyseal plate to encourage longitudinal growth. GH also directs mesenchymal stem cells located in bone marrow to adopt osteoblastic and chondrocytic lineages instead of adipocytic lineage (246,247,251). Previous DXA-based studies demonstrated that children with GH deficiency had low vertebral and total body bone mass along with short stature compared with healthy counterparts (252,253). However, once body size was accounted for, children with GH deficiency had normal bone mass (BMAD), i.e. volumetric BMD, and thinner cortices thought to be related to bone length, i.e. stature (254). In excess, GH causes a very rare condition Chapter 1 – Introduction, Lit Rewiew &Research Hypotheses 60  called gigantism in children, resulting in excess height and higher incidences of vertebral fractures when they are adults (255).  Indirectly, GH acts through IGF-1 to further promote bone formation by stimulating differentiation and activity of osteoblasts. IGF-1 is known to act upon the growth plate (256) and is a prime stimulator of somatic growth and bone elongation. Primary IGF-1 deficiency, i.e. below 2.5 standard deviations (SD), is identified when children with idiopathic short stature (ISS), i.e. height-for-age below 2.5 SD, have normal GH levels but are deficient in IGF-1 (257). Anwar and colleagues described children with ISS with primary IGF-1 deficiency as shorter, with smaller hip circumference, lower BMI and less adiposity and shorter arm span than the average child with ISS. Treatment of IGF-1 deficiency involved individualized treatment of GH that resulted in marked growth in height (257,258). Both GH and IGF-1 stimulate skeletal tissue growth independently, but have additive and synergistic effects when both are present in sufficient amounts (257,258).  1.2.6.3.2 Sex hormones The key sex hormones, testosterone and estrogen, have direct and indirect influences on skeletal development especially during puberty. I present results from in vitro studies that explain the basic cellular interactions of sex hormones and bone cells. Results from studies showed that androgens promote differentiation of osteoblast progenitor cells to osteoblasts at the epiphyseal resting zone and subsequently increase osteoblast numbers and size in the proliferation zone (259,260). Androgens also inhibit osteoblast apoptosis, further encouraging bone formation (259). Androgens increase linear bone growth by binding to androgen receptors (ARs) at the growth plate and influence transcriptional factors to form osteoblasts at the growth plate (259). Indirectly, androgens benefit bone growth through increases in lean mass (muscle), which increases mechanical loading on bone (261). Androgens also stimulate release of GH from the pituitary and IGF-1 from skeletal tissue to encourage bone growth periosteally in adolescents (259). In a Chapter 1 – Introduction, Lit Rewiew &Research Hypotheses 61  lesser, but no less vital role, aromatization of testosterones into estrogens in males is needed for healthy skeletal growth. This was illustrated in case studies of two men (aged 17 and 24 years) who had aromatase deficiency or inactive estrogen receptors (ERs), in the presence of normal levels of androgens. These men displayed thinner, smaller, and less dense bones compared with age- and sex-matched individuals as measured by DXA (lumbar spine, femoral neck and distal radius sites) (262,263). After three years of estrogen treatment, the man with aromatase-deficiency demonstrated significant gains in radius length (8.5%) and CSA (46%) and cortical thickness (12%) at the distal radius as measured by pQCT (263). This suggests that estrogen may be required for androgen action in healthy bone growth.  Estrogen has a bi-phasic function on human bone growth. Pre-puberty, low estrogen levels encourage bone growth at the epiphyseal plate by promoting osteoblast proliferation in both boys and girls (264,265). Low levels of estrogen are also thought to increase sensitivity to mechanical loads, augmenting further bone growth and periosteal apposition in boys and girls (261). Similar to testosterone, estrogen also indirectly enhances bone formation by increasing local IGF-1 secretion independent of GH action (266). As estrogen levels continue to increase during puberty, accelerated osteoblast formation exceeds the rate of chondrocyte formation and this eventually leads to fusion of the epiphyseal plate in late puberty, thus stopping longitudinal bone growth (264,265,267). Higher levels of estrogen, especially in females, appear to interact with different estrogen receptors and modulates bone apposition at the periosteal and endosteal surfaces (Figure 1.20) (261).  Chapter 1 – Introduction, Lit Rewiew &Research Hypotheses 62   Figure 1.20 A schematic that describes the effects of sex hormones on longitudinal bone growth and bone surfaces. ♀, girls; ♂, boys; E, estrogen; T, testosterone; GH, growth hormone; IGF-1, insulin-like growth factor-1; ERα, estrogen receptor-α, ERβ, estrogen receptor-β, AR, androgen receptor. Reproduced from Bellido and Hill Gallant (251), with permission from Academic Press.  1.2.6.4 Dietary calcium Calcium is the main building block of bone material, hydroxyapatite (Ca5(PO4)3OH) (268). Current recommended dietary allowance (RDA) guidelines suggest that children and adolescents 9 to 18 years of age consume 1300 mg/day of dietary calcium to ensure healthy bone development (269). This RDA was derived from results of DXA-based calcium retention and calcium balance studies conducted between 1999-2009 (269).   Dietary calcium is often recommended as a public health intervention for prevention of osteoporosis. However, results of systematic reviews and meta-analyses indicate that calcium supplementation in children only has a small (but positive) effect on aBMD (by DXA) (270–272). For Chapter 1 – Introduction, Lit Rewiew &Research Hypotheses 63  example, Winzenberg and colleagues conducted a meta-analysis of 19 RCTs that assessed the effects of calcium supplementation (range: 300-1200 mg/day) on aBMD in healthy children and adolescents, 3 to 18 years of age (n= 2859). They reported a small but significant positive effect of supplementation on whole body aBMD (standardized mean difference, SMD=0.14, 95% CI: 0.01-0.27) and upper limb aBMD (SMD=0.14, 95% CI: 0.04-0.24) (270), but not proximal femur or lumbar spine aBMD. After cessation of supplementation, a sustained effect was observed only for upper limb aBMD (SMD=0.14; 95% CI: 0.01-0.28) (270). The authors suggested the small benefit in upper arm aBMD translated into a 0.2% reduction in fracture risk, which they did not consider clinically significant (271). Notably, studies included by Winzenberg and colleagues involved children who were predominantly white, healthy, mostly pre- and peri-pubertal and used DXA or single photon absorptiometry (radius sites) to examine the effects of calcium supplementation on bone mass or density (271).  While calcium supplementation alone may have a negligible effect on bone accrual, there is some evidence to suggest that calcium may act synergistically with weight-bearing PA to promote bone accrual during childhood and adolescence. In children as young as 3 to 5 years old, over an average of 50 weeks of intervention, the calcium supplementation with weight-bearing PA (gross motor activities) resulted in larger cortical area and thickness compared with children who did not receive a calcium supplementation and who participated in non-weight bearing (light motor) activities (273). Girls aged 8 to 13 years with high levels of self-reported PA (average 7.2±4.0 h/week) and high calcium supplementation (800 mg/day) demonstrated significantly greater 12-month gains in BMC at the femoral neck, one-third distal radius and total body compared with girls with calcium only, PA only or placebo and sedentary group. Benefits of calcium combined with PA were not observed at the ultra- and mid-radius or the lumbar spine (274). This suggests that bone adaptation to calcium and exercise may be site-specific, or may depend on the dose of PA and/or calcium. To date, calcium and PA trials have mostly evaluated pre- and early pubertal children; Chapter 1 – Introduction, Lit Rewiew &Research Hypotheses 64  none utilized 3-D imaging tools. Thus, there is a need to confirm the possible synergistic effects of calcium and PA on bone strength and structure in the older adolescent skeleton.  One school of thought is that calcium supplementation studies did not show significant positive effects due to insufficient vitamin D intakes. Adequate amounts of vitamin D are needed to actively transport dietary calcium. In healthy vitamin D sufficient 9-13 year old children, higher doses of vitamin D supplementation did not increase calcium gut absorption (275). A double-blind, RCT used pQCT to assess effects of calcium-vitamin D supplementation on bone in peri-pubertal twin girls (age 9 to 13 years; n=20 pairs) (276). One twin received a calcium-vitamin D supplement (800 mg calcium and 400 IU vitamin D) for 6 months while the other received a placebo. The twin receiving the calcium-vitamin D supplement demonstrated greater increases in Tb.Ar (3-5%), Tb.Dn (5%) and bone strength (5-7%) at the distal radius and tibia compared with the non-supplemented twin (276). The calcium-vitamin D supplemented twin also had greater increases in Ct.Ar (6%) and reduced Me.Ar (6-8%) at the tibia shaft compared with their placebo-control twin sibling (276). One other calcium and vitamin D RCT showed different results. Moyer-Mileur and colleagues supplemented 12-year old girls (n=71, 35 supplementation, 36 controls) with 500 mg calcium and 400 IU vitamin D daily for a year in a double-blind, RCT (277). Bone outcomes (by pQCT) between supplementation and control groups were no different (277). Despite the multitude of nutrition and bone health studies in children and adolescents, results of supplementation trials are inconsistent. In my view, the evidence is not compelling to encourage calcium supplementation in an otherwise healthy population of children and youth. Furthermore, there is a glaring gap in bone and nutrition studies based on the absence of studies that used 3-D bone imaging modalities that are able to capture bone geometry and estimate of bone strength.   Chapter 1 – Introduction, Lit Rewiew &Research Hypotheses 65  1.2.6.5 Muscle mass and force We know that muscle development precedes peak bone mineral accrual during growth (181) as depicted in Figure 1.18. In adults, loss of muscle mass or sarcopenia is followed closely by bone loss (278). In his Utah paradigm, Harold Frost proposed that the aim of skeletal physiology is to ensure bones have sufficient mechanical competence to prevent voluntary physical loads from causing spontaneous fractures (45). Muscle forces are the biggest source of mechanical load applied to bone surfaces, as muscle contractions act upon lever arms to create movement (279). For example, muscle imparts up to 5.5 times body weight (BW) at the hip during recovery from a momentary loss of balance from a static single-leg balance, conveying a greater force than any external environmental force present (280).  Just as muscle contractions produce movement and PA, regular participation in PA increases muscle mass and function. While most bone health studies use muscle mass as a proxy for the amount of force that can be generated to influence bone adaptations, muscle function (i.e. strength/force and power) does not depend solely on muscle size or MCSA. Anliker and Toigo eloquently explained that muscle force generation is also influenced by neuromuscular factors, muscle fiber type distribution and muscle’s secondary role as a metabolic tissue in the body (281). Myostatin-deficient mice are prime examples of how muscle size does not translate to the equivalent muscle strength required for bone adaptations. Myostatin-deficient mice experience muscle hypertrophy and with body mass being equal, there were no differences in femoral bone strength, geometry and mass compared with control mice (282). However, when the myostatin-deficient mice were exercised, radii strength increased more than 30% and 25% compared with non-exercised counterparts and wild-type exercised mice, respectively (283).  In a racquet sports study, examination of bone differences between playing and non-playing arm of girls (pre- to post-pubertal stages), muscle area (by MRI) explained 12% to 16% of variance in bone Chapter 1 – Introduction, Lit Rewiew &Research Hypotheses 66  mass, size and strength (by MRI) in the playing arm (284). By default, the study design (contralateral limb comparison) controlled for genetics, body size, nutrition, hormones and other possible external confounders (284). In pre- and peri-pubertal groups, muscle strength (measured using force platform or isokinetic computerised dynamometer) in addition to muscle mass (by pQCT, DXA) predicted bone strength and structure, but not bone density, at the lower extremities (150,285,286). This suggests other possible mechanisms whereby when muscles are put to use (i.e. PA), enhanced muscle function will result in stronger bones. Assessment of muscle properties therefore becomes a focal point to better understand the muscle-bone unit. In children, three valid tests of muscle function are; 1) maximal isometric grip force (grip strength), 2) vertical jump test or Bosco test, and 3) 30-second Wingate test (287). They have been used in large European epidemiological youth studies (AVENA, EYHS, HELENA) (288). However, these protocols assess different muscle parameters. For example, the vertical jump test examines explosive strength while the Wingate examines lower limb peak power (289). Selection of the appropriate approach to evaluate muscle depends on the research question, which dictates the outcome of interest, feasibility of the measure for use in the field or laboratory as well as the validity and reproducibility of the approach. Moreover, the different ways to assess muscle function do relate to bone outcomes. Two decades ago, grip strength explained up to 87% of variance in BSI at the distal radius in children as young as 5 years and in adults as old as 57 years (290). Neu and colleagues examined the relation of grip strength and MCSA at the 65% site of the forearm in 6 to 23 year olds (n=366, 181 boys, 185 girls) (291). Grip strength per unit of MCSA, normalised for forearm length, increased by 44% from age 6 to 20 years and was similar between sexes (291). This meant that muscle function (strength) is not limited per se by the mass available during childhood growth and development. Furthermore, as muscle function is highly associated with body size, outcomes need to be adjusted for body size or be scaled allometrically (292). Other approaches to assess muscle function that were related to bone strength (and other bone outcomes) using isokinetic Chapter 1 – Introduction, Lit Rewiew &Research Hypotheses 67  dynamometers and force and power jump mechanography (force platforms) (285,286,293). Daly and colleagues reported that peak muscle force measured by an isokinetic dynamometer explained an additional 2-5% of variance in femoral neck bone area and section modulus (assessed using MRI) after accounting for limb length and total body lean mass (286). Despite the evidence of muscle (mass and function) association to bone strength, structure and density from cross-sectional and comparison studies in human, results from RCTs are contradictory. RCTs that intervened with resistance training in adolescents did not find significant differences in bone mass changes between intervention and control groups (294,295). While muscle strength improved in the study participants, there were no improvements in bone mass and density (by DXA) in 14 to 18 year old girls after a 26-week (294) or 15-month intervention (295). The lack of evidence may be due to the use of DXA that does not capture changes in bone geometry. Also, the limited literature on the minimal effective load of resistance exercises as opposed to known minimal impact forces required to obtain bone adaptations may be the cause of the non-significant findings in this instance. Overall, more investigations on the influence of muscle, direct PA loading and bone adaptations are required. With the current advances being made using new imaging technologies such as HR-pQCT combined with reliable assessments of muscle function, future investigations may better inform us as to the association between muscle parameters and bone strength, structure and density. 1.2.7 Physical activity and sedentary behaviour  In this section, I define PA, its components and generally, how it relates to bone health. I also briefly describe commonly used PA measurement tools in relation to issues of validity, reliability, strengths and weaknesses. I also define sedentary behaviour, measurement tools for SED and discuss the Chapter 1 – Introduction, Lit Rewiew &Research Hypotheses 68  limited findings related to the influence of sedentary behaviour on bone strength, structure and/or density in children and adolescents. 1.2.7.1 Physical activity - definition, concept and components PA and exercise are common terms that are often used interchangeably in the literature. By definition, PA refers to any bodily movement by skeletal muscles resulting in the use of energy, or incurring energy expenditure above basal metabolic rate (296). Exercise is an extension of PA whereby exercise is a structured, planned, repetitive activity with the aim of improving or maintaining physical fitness (296). I consider PA to encompass both structured and unstructured bodily movement and do not discriminate between sports, recreational or occupational activities in this thesis.  PA comprises four main components, 1) frequency, 2) intensity, 3) type (aerobic, weight-bearing, impact, flexibility), and 4) time or duration. By accounting for frequency, intensity and time/duration, we are able to quantify the volume of PA undertaken (297). This is an important concept as it enables PA to be assessed with good reproducibility and to determine the dose-response relationship between PA and health conditions by facilitating comparisons across studies (298). To better understand the influence of PA on bone strength and predict bone’s response to PA, it is also important to clearly note the type of PA (weight-bearing or non-weight bearing), number of sessions, duration and number/frequency of rest intervals conducted (299). 1.2.7.2 Measurement of PA Many methods are available to quantify children and adolescents’ participation in PA (300). However, self-report questionnaires are still common in field-based studies as they are cost-effective, easy to administer, and are accepted as valid measures of PA. This is considered to be particularly so if the PA Chapter 1 – Introduction, Lit Rewiew &Research Hypotheses 69  measures are associated with physiological-based measures such as indirect calorimetry (301). However, self-reported PA is subject to recall (302) and social desirability bias (303). These measures also do not accurately capture PA intensity levels (304).  Device-based methods offer an alternative means to measure PA, as they objectively quantify motion (e.g., accelerometers, pedometers) or physiological signals (e.g., heart rate monitors) without measurement bias (301). However, device-based methods are not without limitations. Compliance issues (wearing the device for the stipulated time), inability to capture certain types of PA (water-based activities, cycling, resistance exercises) and high costs, limit the use of these methods. Depending on study aims, the PA outcome of interest and the study feasibility and logistics, self-report questionnaires or device-based methods may be employed individually or in tandem to assess PA.  1.2.7.2.1 Subjective measures of PA: questionnaires There are many self-report questionnaires currently available for use with children and adolescents (305). I focus on the Physical Activity Questionnaire for Adolescents (PAQ-A), as it has been used extensively by our research group and allows for valid and reliable assessments of MVPA in school-based settings (20,306). The PAQ-A was designed to be self administered in large-scale studies. The questionnaire captures PA conducted in the past seven days and provides an overall mean PA score derived from eight items, each scored on a 5-point scale (307). Reliability and validity of the PAQ-A to assess general PA levels in secondary school students, are strong (307). Kowalski and colleagues examined the convergent validity of the PAQ-A with PA data from the Caltrac accelerometer, worn by adolescents between Grade 8-12 over a 7-day period (different from the 7-day period assessed by PAQ-A) (20). The study results indicated a low (r=0.33) but significant association between the PAQ-A score to objectively measured PA (20). Janz and colleagues compared the PAQ-A with Actigraph accelerometer data to assess concurrent validity. Participants completed the PAQ-A immediately after wearing the Chapter 1 – Introduction, Lit Rewiew &Research Hypotheses 70  accelerometer for five days (at least one weekend and at least an hour for each time segment present in PAQ-A – morning, lunch, afternoon, evening) (308). They reported a moderately high (r=0.63; p<0.05) significant association between PAQ-A score and accelerometer derived MVPA (308). Although PAQ-A is a valid and reliable tool to assess PA levels, it is unable to estimate intensity, time and frequency and its validity and reliability is specific to school day measurement. However, it is highly applicable in large research settings, easy to administer at low cost.      1.2.7.2.2 Objective measures of PA: accelerometers Accelerometers provide an objective measure of PA by detecting acceleration in body movements (reported as counts) using piezoelectric transducers and microprocessors. Counts are arbitrary units that are translated into levels of PA intensities from calibration studies. These counts are processed to generate output of PA intensity, duration and frequency (22). Specifically, they provide categories of PA intensity based on set ‘intensity’ cut-off points (28,309,310). Cut-points derived from calibration studies are variable depending on the calibration methods (direct calorimetry, doubly-labelled water, laboratory or field setting) and age of those studied (21). Thus, the application of different cut-points prevents direct comparison across studies that used accelerometry to assess PA. Accelerometers are calibrated to quantify energy expenditure. PA intensity levels are interpreted as metabolic equivalents (METs) – defined as the ratio of work metabolic rate to a standard resting metabolic rate of 1.0 (4.184 kJoule or 1 kcal per kg/hr) (311). In general, PA intensity is categorized as either light PA (LPA, <3 METs), moderate PA (MPA, 3-6 METs), vigorous PA (VPA, >6 METs) or moderate-to-vigorous PA (MVPA, ≥3 METs) (312). Sedentary behaviour typically does not exceed resting energy levels and is defined as activity levels between 1-1.5 METs (e.g., sitting, sleeping, lying down and watching television) (313).  Chapter 1 – Introduction, Lit Rewiew &Research Hypotheses 71  The parameters used to collect and assess PA from accelerometry – epoch length, wear time, non-wear time and cut points – may be different between adults and youth (21). I briefly discuss each of these parameters below.  First, prior to accelerometer use, an epoch length is selected. An epoch is the time interval over which counts are measured and integrated. Epoch lengths can affect estimates of PA intensities (314); a 15-second epoch length is recommended to accurately capture PA of adolescents (315) compared with a 1-min epoch length for adults. Second, the amount of wear time that constitutes a valid day (including criteria for establishing non-wear time), and the number of valid days required for inclusion are set to ensure that the measurement period provides an accurate representation of general PA. Generally, at least 10 hrs/day of wear time is used to identify a valid day (316). Seven days of wear represents ‘usual’ PA across all ages; however, majority of studies of adolescents adopted protocols that adopted 3-4 days of wear days in their analysis (317). Third, non-wear time cut-points are used to identify and excluded as non-wear periods. A summation of continuous zero counts per minute is the typical way to identify non-wear time (317). However, this may also include sedentary periods and not true non-wear periods. Furthermore, spurious counts may occur due to slight movements while sedentary or when moved (e.g., bumped or moved from a spot to another) but not worn during PA. In studies of adolescents, investigators have used a period of 20 min and 30 min of consecutive zeros of counts per minute as a criterion of non-wear time (317). Fourth, to translate accelerometer counts into accessible measures of PA, cut-points are used to define sedentary behaviour and PA intensity based on the population of interest. Trost and colleagues compared five different accelerometer cut-points for children and adolescents to measures of intensity assessed using indirect calorimetry. They reported that cut-points published by Evenson and colleagues (28) most accurately classified PA intensity (sedentary, LPA, MPA and VPA) among children and adolescents aged 5 to 15 years (310).  Chapter 1 – Introduction, Lit Rewiew &Research Hypotheses 72  Accelerometers can provide highly accurate assessments of PA level when appropriate methods are used. However, there are limitations to the use and interpretation of results from accelerometry. Accelerometers are unable to assess water based PA (as these devices are not waterproof) or PA that requires minimal trunk movement (e.g., cycling) (300). Accelerometers are also unable to account for increases in METs if a person carries an extra load (e.g., backpack) when conducting certain PA (e.g., hiking) (300). Without a consensus on the protocols to report, collect, process and analyze, results from accelerometry studies are often non-comparable (21). 1.2.7.3 Sedentary behaviour – definition and concept There are two commonly used definitions of sedentary behaviour. The first one is limited to the intensity of the activity, whereby sedentary behaviour is any waking behaviour of 1.5 METs. The second definition of sedentary behaviour takes into accounts both intensity and posture, whereby sedentary behaviour is any waking behaviour of 1.5 METs in a sitting or reclining position (318). The latter definition takes into account the positive benefits of standing from increased muscle activation (319) including lower mortality (320) and improved cardiometabolic markers (321). As there is no consensus currently on the precise definition, the definition adopted by researchers will depend on their research focus. As there are no studies of the association between health outcomes and standing or sitting in adolescents, I adopt the first definition of sedentary behaviour in my thesis as it is more general and has been used in epidemiological studies (322–324).  1.2.7.3.1 Measuring sedentary behaviour One novel use of accelerometers is to operationalize and quantify sedentary behaviour. Previously, sedentary behaviour was usually captured by questionnaires that depended on recall skills to estimate sedentary behaviour time, frequency and type (e.g., sitting down or lying down reading) Chapter 1 – Introduction, Lit Rewiew &Research Hypotheses 73  (325,326). Hardy and colleagues validated their sedentary behaviour questionnaire with accelerometer results in adolescents (325). They showed that the questionnaire and accelerometer results had criterion validity with less than 5% of data points outside the limits of agreement. This is despite the 3.2±11.2 hours/week discrepancy between accelerometer and questionnaire measures of sedentary behaviour. Nonetheless, as mentioned in section 1.2.7.3.1, SED can be accurately captured by accelerometry.  1.2.7.3.2 Association between measures of sedentary behaviour and bone outcomes  There are five studies, to my knowledge, that examined the influence of SED on bone properties. As ‘unloading’ has known detrimental effects on the skeleton (327) sedentary behaviours would theoretically be associated with low bone mass (by DXA). Results, however, are inconsistent. Some studies support this association (322,328–330) while one study from our group reported no association between SED and bone architecture and strength at the distal tibia (assessed by HR-pQCT) (331). From the National Health and Nutrition Examination Survey (NHANES), Chastin and colleagues showed that MVPA (by accelerometry) was no longer associated with BMC when sedentary activities (by self-report) were entered into regression models (329). One hour per day of screen time was associated with lower values for proximal femur BMC (0.77 g lower in girls and 0.45 g lower in boys) (329). These differences are likely associated with how sedentary behaviour was assessed, the bone site measured and the instrument used to assess bone. Thus, new evidence is emerging that highlights the need to examine the influence of PA and SED on bone outcomes during growth.  1.2.8 Studies of PA and bone strength in children and adolescents In this section, I briefly provide an overview of how PA can influence bone health, followed by a review of studies of PA and bone health conducted in children and adolescents. I focus on the highest Chapter 1 – Introduction, Lit Rewiew &Research Hypotheses 74  levels of evidence, systematic reviews, RCTs and longitudinal studies that assessed the influence of PA on bone strength, structure and density in children and adolescents.  1.2.8.1 The influence of PA on bone health Earlier, I discussed the mechanostat theory and how bone positively responds to loading and impact (section 1.2.2.3). I also provided an overview of the principles whereby weight-bearing exercise influences bone adaptation (section 1.2.2.3). Figure 1.21 illustrates how exercise affects bone through; 1) direct impact, (e.g., high ground reaction forces when the feet strike the ground, 2) muscle forces from contractions during PA, and 3) endocrine and paracrine related pathways (e.g., myokines that inhibit adipogenesis and positively stimulate osteogenesis and myogenesis (332). In previous child and youth studies that investigated the effect of mechanical loading on bone, it was not possible to identify pathways that played the most dominant role. However, a recent meta-analyses included RCTs conducted with children and reported a positive benefit for PA on aBMD and BMC accrual (assessed using DXA) (333,334). Importantly, PA during childhood and adolescence may impart lifelong benefits for bone health (335). The life stage when PA participation occurred influences bone mass outcomes, as illustrated in Figure 1.22. Chapter 1 – Introduction, Lit Rewiew &Research Hypotheses 75   Figure 1.21. Schematic that illustrates the effect of exercise on the skeleton. During exercise, load is transmitted to the skeleton through direct stimulation on bone mechanosensors and by indirect stimulation through dynamic muscle activity. Hormones from fat and the liver modulate loading by affecting bone and muscle growth as well as muscle performance, and act indirectly through potential changes in the mineral reservoir. Reproduced from Bonnet and Ferrari (336), with permission from International Bone and Mineral Society.  Figure 1.22. A schematic that illustrates the effect of physical activity on bone mass at different periods across the life span. The red curve represents a continuous exercise effect. Without a bone loading stimulus through exercise, bone mass accrual could be attenuated over time. The yellow curve represents a person engaging in normal activities (e.g., walking). Reproduced from Bonnet and Ferrari (336), with permission from International Bone and Mineral Society. Chapter 1 – Introduction, Lit Rewiew &Research Hypotheses 76   The University of Saskatchewan’s PBMAS longitudinal study followed the growth and development of girls and boys from childhood to adulthood (11,12,337). Physically active boys and girls followed for at least six years with entry age between 8 and 14 years, had greater bone mineral accrual (7% to 17 % more total body BMC assessed using DXA) than their less active counterparts (11,12). After adjusting BMC for body size (height and body mass), the investigators found that active boys (top quartile) had 9% more total body and 7% more femoral neck BMC compared with inactive (lowest quartile) boys one year after peak bone mineral accrual velocity (11). Active girls had 16% and 11% more total body and femoral neck BMC, respectively, a year after peak BMC accrual velocity (11). Fifteen years later, PBMAS participants returned for follow up measurement as young adults. Adults who were considered active adolescents had significantly greater bone CSA (4% - 7%) and Z (5% - 8%, by DXA-HSA) (controlled for sex, height, body mass, CSA and Z at APHV, sex and relative lean mass) compared with adults who were considered as inactive during adolescence (337). This long-term study, although difficult to conduct due to high costs and attrition, provided important insights regarding the positive role of PA in bone accrual and development of bone structure at the proximal femur. However, as they used self-report measures of PA (PAQ-C), they were unable to assess the amount (frequency and duration) and type (intensity, weight-bearing, endurance) of PA that conferred these benefits. RCTs designed to evaluate the effect of PA on bone strength would confirm and advance these early findings. A recent systematic review aimed to identify the influence of PA on bone strength across the lifespan (37). The authors focused on RCTs with a minimum 6-month intervention period and on bone strength as the primary outcome. They reported a small but significant effect of PA on bone strength in prepubertal boys (effect size = 0.17, 95% CI 0.02-0.32) from three different studies (5 publications). However, Nikander et al. grouped all bone strength parameters together regardless of skeletal site measured (radius or tibia) or instrument used to assess bone strength (DXA, pQCT). Thus, the findings may over-simplify complex and site-specific relationships. Although this first review makes an important Chapter 1 – Introduction, Lit Rewiew &Research Hypotheses 77  complication, results from the meta-analysis should be interpreted cautiously. I extended Nikander and colleagues’ review by broadening the scope of eligible studies to include RCTs and observational studies that assessed the influence of PA on bone strength and structure or density in children and adolescents. I include this systematic review as a part of my thesis, in section 3.1. 1.2.8.2 Association between objectively measured PA and bone outcomes  Some studies used accelerometry to investigate the relation between PA intensity and bone outcomes in children and adolescents. They reported that MVPA and VPA were related to bone strength, structure and density in children (338–340) and adolescents (13,341). In boys who were followed longitudinally over six years (mean age 5.2 years at baseline), 40 minutes of MVPA per day was associated with a 3-5% larger CSA and greater bone strength (Z, assessed by DXA-HSA) at the femoral neck compared with boys who performed 10 min/day of MVPA (adjusted for lean mass) (339). In adolescent boys (age 12.5-17.5 years, n=189), less than 45 min/day of MVPA was associated with reduced bone mass while more than 78 min/day of MVPA or more than 28 min/day of VPA were associated with higher values for aBMD at the femoral neck (by DXA) (13). In pre- and peri-pubertal boys and girls, VPA explained 2-11% of variance in bone strength at the proximal femur (assessed by DXA) after adjusting for lean mass (338,340). In another cross-sectional study, VPA was associated with bone area and BMC at the tibial shaft (assessed using pQCT) in boys and girls (mean age 15.5 years, n=1748) (341). There was a 7-mm2 greater bone area in those in the highest quartile of PA (VPA) compared with those in the lowest quartile of PA (341). Due to the different cut-points used to categorize MVPA and VPA, comparisons across studies should be interpreted with caution.  Accelerometers are unable to specifically record high impact PA levels that are most osteogenic (section 1.2.2.3, principles of bone adaptation). For example, Actigraph accelerometers are able to capture Chapter 1 – Introduction, Lit Rewiew &Research Hypotheses 78  accelerations as high as 2.5 G during weight-bearing PA (342). However, higher intensity activities (e.g., running at speeds > 10 km/h (equivalent to 10 METs) (343)) generate accelerations greater than 2 G, 30% to 45% of the time (344). Vertical accelerations greater than 4.2 G were not associated with higher values for femoral neck aBMD (assessed by DXA) in adolescents (344). Artistic gymnastics, a very high impact sport, generates accelerations of about 4 to 8 G (345) and has a MET value of 3.8 (343). This high-impact sport would be classified as MVPA. Thus, although accelerometers have advanced our field, an alternative means to assess the direct impact of PA on bone outcomes is still needed. 1.2.8.3 School-based intervention studies on PA and bone health Schools are targeted settings for PA intervention trials in children and adolescents for several reasons. First, school-based PA interventions reach a large number of children and adolescents from diverse backgrounds. Second, schools are considered ideal settings for health promotion research, as nearly 50% of waking hours for the first two decade of a child’s life are spent in schools (346). Third, schools provide a setting where successful interventions could be integrated into the curriculum. Targeted programs could potentially comprise a part of teacher development and training, supported by PA-promoting infrastructure.  Previous PA school-based intervention studies tended to focus on changes in physical education (PE) programs. However, multi-pronged, whole school-based interventions proved most effective (347). Importantly, these models have potential to translate research into practical application (348). A whole school-based intervention also eliminates stigmatization of specific individuals or groups (e.g., less physically active, the overweight or obese) while providing a platform to disseminate the good intentions of the intervention to all students (347). Based on the social ecological theory, changes in the social setting and environment encourage and support individuals to adopt positive health behaviours (349). Another Chapter 1 – Introduction, Lit Rewiew &Research Hypotheses 79  advantage of a whole school-based intervention is that it avoids contamination across individuals when the intervention is contained within an institution in contrast to a case where both intervention and control groups are within the same institution.  However, a meta-analysis that investigated the effect of school-based PA interventions on PA levels showed no significant effect in adolescents (38). Some studies adopted social cognitive or social action theory (350) or the transtheoretical model of behaviour change, while others used non-theoretical models. As adolescents transition to adulthood they seek to establish their independence. Thus, some studies suggested that self-determination or intrinsic motivator models may positively influence PA (351–353). Lonsdale and colleagues (354) reported that Grade 10 equivalent boys and girls who had higher self-determination performed more PA (steps/min) than students with low self-determination, regardless of whether they were in structured or free choice PE class (354). It may be that perceived competence, autonomy and sense of relatedness or social belonging (embedded within self-determination theory) (355) are factors that drive adolescents’ degree of choice and empowerment related to becoming more physically active. Prescribed PA regimens provided as a part of structured interventions as in most RCTs may not appeal to adolescents. Furthermore, new findings from studies that used self-determination theory (SDT) encouraged the application within a whole school model to increase adolescent PA (356,357).  To date, only a handful of school-based interventions used 3-D imaging tools to assess the effect of PA on bone strength and structure (32,34,358). None included older adolescents (>14 to 17 years). Our research group designed and implemented Action Schools! BC – a whole school-based intervention model that sought to provide students (average age 10.1 years at baseline) with 150 minutes of PA/week across 6 Action Zones [www.actionschoolsbc.ca]. The cornerstone of the model -- ‘classroom action’ -- provided students 15 minutes/day, 5 days/week of classroom based PA (32,113). Within classroom action, children attending intervention schools (n= 281) participated in a progressive jumping program called Bounce at Chapter 1 – Introduction, Lit Rewiew &Research Hypotheses 80  the Bell. Students performed high-impact jumps 3 times/day (at the morning, noon and end of the day school bell), about 1 min each time, 4 days a week. Based in part on its simplicity, compliance with Bounce at the Bell was high; implementation rate by intervention teachers was 74%. After an intervention period of 16-months, investigators assessed changes in bone structure and strength using pQCT. They reported significant increases in BSI (4%) and total BMD (2%) at the 8% site of the tibia in pre-pubertal intervention boys (n=143), compared with controls (n=64) (32). Change in the maximum second moment of inertia (Imax) at the tibia midshaft in boys, was also significantly higher (3%) in the intervention group compared with controls (113).  Two studies in the 6 to 12 years age group found no differences in bone outcomes between intervention and control groups following 28 (34) or 36 (358) weeks of PA intervention. Specifically, Greene and colleagues used MRI to examine changes in mid-third femoral bone structure and estimated bone strength in elementary school children who participated in a three times a week, 28-week high impact loading intervention (34). Students performed 10 sets of single-leg drop landing exercises from different drop-heights (High or Low) off a step bench. There was no difference for any bone values between the exercised leg and the non-exercised leg at the mid-third of the femur. This finding might be due in part, to the short time period for the intervention and/or the small sample size (n= 13 in each of High, Low intervention and control groups) (34). Anliker and colleagues conducted a 10 min, twice a week, circuit that comprised of five jumping and sprinting exercises for 36 weeks in 8 to 12-year old boys and girls (Intervention: boys, n=12, girls, n=18; Control: boys, n=16, girls, n=14) (358). They found no differences in bone strength and structure between intervention and control groups at the end of the intervention period (358). It appears that frequency of the intervention rather than intensity and duration of PA may be key to eliciting a positive effect in bone in this age group but each study had a different follow-up period – the Chapter 1 – Introduction, Lit Rewiew &Research Hypotheses 81  shortest was 28 weeks (34), followed by 36 weeks (358) followed by 16 months (32,113). It might also be due to a longer intervention duration. Bounce at the Bell was performed 4 day/week, three times each day (32,113). Thus, weight-bearing bones received a short bout (1 min/bout; 3 bouts/day), high impact stimulus with ground reaction force equivalent to 5 times BW (359). Other interventions delivered a ‘dose’ comprised of continuous 10 to 15-min PA bouts of stretching and jumping exercises, two (358) to three (34) times per week. Overall impact of jumps were similar across the three interventions as jumps were either single- or two-legged jumps (Greene and colleagues used a single-leg landing protocol). These studies were conducted in elementary schools. To my knowledge there are no PA interventions in a secondary school setting that assessed bone strength and structure (using 3-D bone imaging tools) in older adolescents.  1.2.9 Summary  Conducting studies during adolescence – when biological and physiological systems are developing rapidly at different tempos and at different times and when youth are transitioning into adulthood – is a challenge. Given the tremendous variability in maturity level, studies in this age group must account for sex and maturity levels of participants in the study design and the statistical approach. Thus, it is not surprising that few studies examined the effects of whole school-based PA interventions on bone strength and structure in adolescents. I address this gap in my thesis and incorporate a novel peer-to-peer intervention in a secondary school setting to evaluate the effect of this choice based model on adolescent bone health (strength, structure, density). I also extend the literature by assessing PA objectively using accelerometry, in conjunction with a validated PA questionnaire. Another novel component is that I assess SED by accelerometry to evaluate its independent effect on bone density, structure and strength. Given the central role of muscle in bone’s response to exercise, I will examine the contribution of muscle strength to bone strength and structure at non-weight bearing bone sites. Taken Chapter 1 – Introduction, Lit Rewiew &Research Hypotheses 82  together, these novel components comprising my thesis represent the first whole school-based study of secondary school students that aimed to enhance PA and evaluate its influence on bone strength, structure and density.  1.3 Rationale, Objectives and Hypotheses 1.3.1 Part I: A systematic review of the effects of physical activity on bone strength and structure in children and adolescents Rationale: In the past two decades, studies of the pediatric skeleton shifted from assessing the acquisition of bone mineral to assessing the acquisition of bone strength and the contribution of bone density and structure to bone strength. This is possible through the advent of safe, precise and accurate 3-D bone imaging instruments such as pQCT and HR-pQCT. However, we still know surprisingly little about how bone strength adapts in the rapidly changing adolescent skeleton and even less about the role of PA in those adaptations. Therefore, I extend the literature by conducting a systematic review (including RCTs and observational studies) that evaluates the effect of PA on bone strength and structure in adolescents.  Objective: To systematically evaluate the effect of PA on bone strength and structure (i.e. bone density, total area, cortical area, cortical thickness) in children and adolescents.  Contribution: This systematic review (360) is a comprehensive and thorough examination of RCTs and observational studies that evaluated the effects of PA on bone strength and structure in children and adolescents. This is the first systematic review of PA studies in children and youth that focused on bone   Chapter 1 – Introduction, Lit Review & Research Questions 83  structure and estimated bone strength as primary outcomes. I extend the scope of previous reviews to include observational studies with recreational PA and athletic populations.  1.3.2 Part II: Determinants of bone strength and structure in adolescent boys and girls Rationale: Several factors are known to influence bone strength in children – PA is among them. However, we know relatively little about factors that influence bone strength accrual in adolescents. Thus, I aim to identify the role of modifiable factors (with a focus on PA) on bone outcomes (bone strength, density and structure) in adolescents. To do so I will account for non-modifiable traits such as ethnicity, sex and maturity in my analyses. I also examine relationships between bone outcomes and muscle strength, MVPA, SED and VPA.  Objectives:  1. To identify PA factors (with a focus on intensity) that predict estimated bone strength in adolescent boys and girls (accounting for sex, maturation, ethnicity, limb length and muscle mass).  2. To identify PA factors (with a focus on intensity) that predict bone density and structure in adolescent boys and girls (accounting for sex, maturation, ethnicity, limb length and muscle mass differences).  Hypotheses  H1: PA is positively associated with bone strength (adjusting for sex, maturation, ethnicity, limb length and muscle mass).   Chapter 1 – Introduction, Lit Review & Research Questions 84  H2: PA is positively associated with bone structure (total area, cortical area and medullary area) but not bone density (adjusting for sex, maturation, ethnicity, limb length and muscle mass) Contribution: This study is novel from several aspects other than the less studied 15-year old cohort. First, this is the first study on the association of objectively measured PA and SED on bone strength, structure and density assessed by a 3-D bone imaging tool (pQCT) at the distal and shaft sites of the tibia. Second, my investigations on the association of muscle function (i.e. grip strength) on bone strength, structure and density (by pQCT) at the distal and shaft sites of the radius, while controlling for muscle mass, will provide insights on the muscle-bone unit interaction. Overall, my findings will inform intervention studies on the determinants of bone strength, structure and density from PA-related factors in 15-year old boys and girls. 1.3.3 Part III: Effect of a secondary school physical activity intervention on bone strength and structure in adolescents. Rationale: School-based PA intervention programs may be the best means to access youth across a range of ethnicities and socioeconomic strata. Historically, schools represent a setting where strong compliance and an effective delivery system hold promise for PA interventions (361,362). As adolescents transition into adulthood, choice-based models may represent the best (and a novel) method to encourage PA