Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Seeking sensations through sport : an interdisciplinary investigation of personality and genetics associated… Thomson, Cynthia J. 2013

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

Item Metadata

Download

Media
24-ubc_2013_fall_thomson_cynthia.pdf [ 22.59MB ]
Metadata
JSON: 24-1.0073919.json
JSON-LD: 24-1.0073919-ld.json
RDF/XML (Pretty): 24-1.0073919-rdf.xml
RDF/JSON: 24-1.0073919-rdf.json
Turtle: 24-1.0073919-turtle.txt
N-Triples: 24-1.0073919-rdf-ntriples.txt
Original Record: 24-1.0073919-source.json
Full Text
24-1.0073919-fulltext.txt
Citation
24-1.0073919.ris

Full Text

SEEKING SENSATIONS THROUGH SPORT: AN INTERDISCIPLINARY INVESTIGATION OF PERSONALITY AND GENETICS ASSOCIATED WITH HIGH-RISK SPORT by Cynthia J. Thomson B.Sc., Queen’s University, 2003 B.P.H.E., Queen’s University, 2003 M.Sc., The University of British Columbia, 2008 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES (Kinesiology) THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) June 2013 Cynthia J. Thomson, 2013  Abstract Sensation seeking involves a desire to seek out thrilling experiences and a willingness to take risks in exchange for rewards. Sensation seekers are drawn to risky activities and high-risk sports represent potentially positive outlets for such individuals. Sensation seeking is moderately heritable and variants in genes involved in dopaminergic transmission have been associated with sensation-seeking phenotypes, although no studies have investigated personality and genetic variants in high-risk sport practitioners. This interdisciplinary dissertation explores personality (general sensation seeking and contextual sensation seeking in sport) and genetic variables (polymorphisms in monoamine pathway genes) in proficient high-risk sport practitioners. In the first series of projects two independent cohorts (n = 220, n = 668) of skiers/snowboarders (risky, yet popular sports) completed questionnaires and provided DNA samples. Data derived from questionnaires were used to evaluate the reliability and predictive validity of a new sensationseeking tool for downhill sports that was developed as part of this study. The questionnaire showed strong psychometric properties and significantly predicted injury (! = .358, p < .001) in skiers and was used to define phenotypes in subsequent genetic studies. Using designs that employed independent replications, the newly defined phenotype was significantly associated (p < .001) with a functional variant (-521C/T) in the dopamine-4-receptor gene (DRD4), and an association between general sensation seeking and a variant (rs167771, intronic G/A) in the dopamine-3-receptor gene (DRD3) was also observed in the ski cohorts (p = .004). Personality and genetic variables were then compared using a quasi case-control design between practitioners of very high-risk (e.g., paragliding, ski-mountaineering, n = 141) and low-risk sports (e.g., running, n =132). The high-risk group scored higher than low-risk athletes on sensation seeking (p < .05), but not impulsivity, a trait commonly associated with deviant risk-  ii  taking and there were marginal associations between sport group and genetic variants in the stathmin (p = .004) and brain-derived neurotrophic factor (p = .03) genes, but the associations did not survive correction for multiple testing. The finding that risk-taking through sport may be, in part, predicted by genetic background provides a novel insight into the potential antecedents of performance.  iii  Preface Almost all of the work described in this dissertation was conducted during my doctoral studies with collaborations from the co-authors listed below. A portion of the work described in Chapter 3 was conducted during my master’s thesis with guidance from Dr. Mark Beauchamp (acknowledged in the published version) and a version of the work has been published (Thomson, Morton, Carlson, & Rupert, 2012, shown below). I conducted all of the testing and wrote the manuscript. I consulted with the co-authors on research methods and analyses, and all authors contributed by assisting with editing. C.J. Thomson, K.L. Morton, S.R. Carlson, J.L. Rupert. (2012). The contextual sensation seeking questionnaire for skiing and snowboarding (CSSQ-S): Development of a sport specific scale. International Journal of Sport Psychology, 43(6): 503-521. Chapter 4 is based on work conducted in part during my master’s thesis (74 genotypes from the pilot sample (n = 117) used at stage 1 of the analyses), the result was then replicated during my doctoral studies in an independent study (n = 386). A version of the work has been published (Thomson, Hanna, Carlson, & Rupert, 2013, shown below). I conducted all of the recruitment, testing, analyses, and wrote the manuscript. I consulted with the co-authors on research methods, and C. Hanna assisted with the genetic analysis. All co-authors helped edit the manuscript. C.J. Thomson, C.W. Hanna, S.R. Carlson, J.L. Rupert. (2013). The -521 C/T variant in the dopamine-4-receptor (DRD4) is associated with skiing and snowboarding behavior. Scandinavian Journal of Medicine & Science in Sports, 23(2): e108-113. doi: 10.1111/sms.12031 iv  Chapter 5: A version of the work has been published (Thomson, Carlson, & Rupert, 2013, shown below). I conducted all of the recruitment and analyses, and a majority of the sample preparation. A. Rajala assisted with DNA isolation and an external facility (Genome Quebec) performed the genetic analysis. I wrote the manuscript and the co-authors assisted with editing the manuscript. C.J. Thomson, S.R. Carlson, J.L. Rupert. (2013). Association of a common D3 dopamine receptor gene variant is associated with sensation seeking in skiers and snowboarders. Journal of Research in Personality, 47: 153-158. doi: 10.1016/j.jrp.2012.11.004 The data described in Chapters 6, 7, and 8 have not been published.  Ethics certificate Numbers: Reliability CSSQ: H10-00669 (2011) Impulsivity & Daring Behaviour (women, men): H10-00003, H09-02451 (2010) The genetics of sport behaviours: H09-02005 (2009) The genetics of sport behaviours: H07-00207 France Recruit: H10-02040 (2012)  v  Table of Contents Abstract.................................................................................................................................... ii! Preface..................................................................................................................................... iv! Table of Contents ................................................................................................................... vi! List of Tables ........................................................................................................................ xvi! List of Figures..................................................................................................................... xviii! List of Abbreviations ........................................................................................................... xix! List of Symbols ................................................................................................................... xxiii! Glossary .............................................................................................................................. xxiv! Acknowledgements ........................................................................................................... xxvii! Dedication ........................................................................................................................... xxix! Chapter 1: Introduction ........................................................................................................ 1! 1.1! Overview...................................................................................................................... 1! 1.2! Why study sport? ......................................................................................................... 2! 1.3! Sensation seeking......................................................................................................... 4! 1.3.1! Sensation seeking and other personality models: Approach traits ....................... 5! 1.3.2! Impulsivity ............................................................................................................ 7! 1.3.3! Laboratory correlates of sensation seeking........................................................... 8! 1.3.4! Sensation-seeking instruments............................................................................ 11! 1.3.5! Demographic trends in sensation seeking........................................................... 16! 1.4! Lifestyle correlates of sensation seeking ................................................................... 17! 1.4.1! Deviant high-risk behaviours.............................................................................. 17! 1.4.2! Sensation seeking and sport ................................................................................ 20!  vi  1.5! Neurobiology of sensation seeking............................................................................ 24! 1.5.1! Dopamine synthesis and transport ...................................................................... 28! 1.5.2! Dopamine and sensation seeking ........................................................................ 31! 1.6! Sensation-seeking genetics: A review of “candidate genes” ..................................... 32! 1.6.1! Dopaminergic pathway genes ............................................................................. 33! 1.6.1.1! The DRD4 gene ........................................................................................... 34! 1.6.1.2! The DRD2 gene ........................................................................................... 35! 1.6.1.3! Other dopamine receptors ............................................................................ 36! 1.6.1.4! The dopamine transporter ............................................................................ 37! 1.6.1.5! Dopamine metabolism ................................................................................. 38! 1.6.2! Serotonergic pathway genes ............................................................................... 42! 1.6.2.1! Serotonin transporter.................................................................................... 42! 1.6.2.2! Serotonin receptors ...................................................................................... 43! 1.6.3! Other potential candidate genes for sensation seeking ....................................... 44! 1.6.4! Interactions between candidate genes................................................................. 46! 1.7! Genetics of risk-inclined behaviours ......................................................................... 47! Chapter 2: Research overview............................................................................................ 49! 2.1! Objectives .................................................................................................................. 53! 2.2! Hypotheses................................................................................................................. 54! 2.2.1! General hypotheses ............................................................................................. 54! 2.2.2! Specific hypotheses............................................................................................. 54! 2.3! Methodology overview .............................................................................................. 56! 2.4! Overview of participants and recruitment ................................................................. 57!  vii  2.4.1! Sample 1 (Pilot sample) ...................................................................................... 59! 2.4.2! Sample 2 (Festival sample)................................................................................. 59! 2.4.3! Peer samples ....................................................................................................... 61! 2.4.4! Reliability sample ............................................................................................... 61! 2.4.5! High-risk/low-risk samples................................................................................. 63! 2.4.6! Participant exclusions ......................................................................................... 63! 2.5! Procedures.................................................................................................................. 64! 2.6! Methods ..................................................................................................................... 65! 2.6.1! Buccal DNA preparation .................................................................................... 66! Chapter 3: The Contextual Sensation Seeking Questionnaire for skiing and snowboarding: Development of a sport specific scale ....................................................... 69! 3.1! Summary .................................................................................................................... 69! 3.2! Introduction................................................................................................................ 69! 3.3! Study 1: Item generation and preliminary evidence for validity ............................... 72! 3.3.1! Participants and procedures ................................................................................ 72! 3.3.2! Measures ............................................................................................................. 75! 3.3.3! Results................................................................................................................. 77! 3.4! Study 2: Structural aspects of construct validity and reliability ................................ 79! 3.4.1! Participants and procedures ................................................................................ 80! 3.4.2! Measures ............................................................................................................. 81! 3.4.3! Results................................................................................................................. 81! 3.5! Study 3: Sensation seeking and injury prevalence..................................................... 83! 3.5.1! Participants and procedures ................................................................................ 84!  viii  3.5.2! Measures ............................................................................................................. 85! 3.5.3! Results................................................................................................................. 85! 3.6! Discussion .................................................................................................................. 87! Chapter 4: The -521 C/T polymorphism in the dopamine-4-receptor gene (DRD4) is associated with skiing and snowboarding behaviour ........................................................ 92! 4.1! Summary .................................................................................................................... 92! 4.2! Introduction................................................................................................................ 92! 4.3! Methods ..................................................................................................................... 96! 4.3.1! Participants.......................................................................................................... 96! 4.3.2! Measures ............................................................................................................. 98! 4.3.3! Genotyping.......................................................................................................... 98! 4.3.4! Statistical analyses ............................................................................................ 101! 4.4! Results...................................................................................................................... 102! 4.4.1! Stage 1............................................................................................................... 102! 4.4.2! Stage 2............................................................................................................... 103! 4.5! Discussion ................................................................................................................ 105! Chapter 5: Association of a common D3 dopamine receptor gene variant is associated with sensation seeking in skiers and snowboarders......................................................... 109! 5.1! Summary .................................................................................................................. 109! 5.2! Introduction.............................................................................................................. 109! 5.3! Methods ................................................................................................................... 112! 5.3.1! Participants........................................................................................................ 112! 5.3.2! Measures ........................................................................................................... 113!  ix  5.3.3! Genetic analysis ................................................................................................ 113! 5.3.4! Statistical analysis............................................................................................. 114! 5.4! Results...................................................................................................................... 115! 5.5! Discussion ................................................................................................................ 119! Chapter 6: No association between promoter variants of the dopamine-4 receptor gene and sensation seeking in skiers and snowboarders.......................................................... 123! 6.1! Summary .................................................................................................................. 123! 6.2! Introduction.............................................................................................................. 123! 6.3! Methods ................................................................................................................... 126! 6.3.1! Participants........................................................................................................ 126! 6.3.2! Measures ........................................................................................................... 128! 6.3.3! Genotyping........................................................................................................ 128! 6.3.4! Statistical analyses ............................................................................................ 129! 6.4! Results...................................................................................................................... 130! 6.5! Discussion ................................................................................................................ 133! Chapter 7: Interaction between allelic variants in the dopamine-4-receptor gene is associated with patterns of downhill sport behaviour..................................................... 136! 7.1! Summary .................................................................................................................. 136! 7.2! Introduction.............................................................................................................. 136! 7.3! Methods ................................................................................................................... 138! 7.3.1! Participants........................................................................................................ 138! 7.3.2! Measures ........................................................................................................... 139! 7.3.3! Genotyping........................................................................................................ 140!  x  7.3.4! Statistical analysis............................................................................................. 141! 7.4! Results...................................................................................................................... 142! 7.5! Discussion ................................................................................................................ 147! Chapter 8: Exploratory analysis of personality and genetic variables in high- and lowrisk sport participants ........................................................................................................ 151! 8.1! Summary .................................................................................................................. 151! 8.2! Introduction.............................................................................................................. 151! 8.3! Methods ................................................................................................................... 158! 8.3.1! Participants........................................................................................................ 158! 8.3.2! Procedures......................................................................................................... 158! 8.3.3! Measures ........................................................................................................... 160! 8.3.3.1! Demographic variables .............................................................................. 160! 8.3.3.2! Psychotropic medication............................................................................ 160! 8.3.3.3! Substance use inventories .......................................................................... 161! 8.3.3.4! Sport inventories ........................................................................................ 162! 8.3.3.5! Adult Self-Report Symptom Checklist 1.1 (ASRS-V1.1) ......................... 163! 8.3.3.6! ZKPQ ImpSS ............................................................................................. 164! 8.3.4! Genetic analysis ................................................................................................ 164! 8.3.5! Analyses............................................................................................................ 165! 8.4! Results...................................................................................................................... 167! 8.4.1! Participant exclusions ....................................................................................... 167! 8.4.2! Results for personality analyses........................................................................ 171! 8.4.3! Sex differences.................................................................................................. 173!  xi  8.4.4! Results from genetic analyses........................................................................... 175! 8.4.4.1! Preliminary tests of genotype distributions between recruitment sites...... 175! 8.4.4.2! Genetic analyses of high- and low-risk athletes ........................................ 176! 8.4.5! Results for consumption patterns...................................................................... 181! 8.4.5.1! Substance use ............................................................................................. 181! 8.4.5.2! Smoking ..................................................................................................... 182! 8.4.5.3! Alcohol....................................................................................................... 182! 8.5! Discussion ................................................................................................................ 183! 8.5.1! Personality trait differences between and within sport group........................... 183! 8.5.2! Genetic differences between sport groups ........................................................ 185! 8.5.3! Sex differences.................................................................................................. 188! 8.5.4! ADHD ............................................................................................................... 189! 8.5.5! Limitations and conclusions ............................................................................. 190! Chapter 9: General discussion.......................................................................................... 193! 9.1! Review of project findings....................................................................................... 193! 9.2! Key findings............................................................................................................. 195! 9.3! Evolutionary mechanisms........................................................................................ 198! 9.4! Sensation seeking in athletes ................................................................................... 200! 9.4.1! Sensation seeking and performance.................................................................. 200! 9.4.2! Gender differences in sensation seeking........................................................... 202! 9.5! Proposed mechanisms.............................................................................................. 203! 9.6! Potential factors influencing high-risk sport participation ...................................... 206! 9.6.1! Physiological factors......................................................................................... 206!  xii  9.6.2! Environmental factors....................................................................................... 207! 9.6.3! Gene-environment interplay ............................................................................. 208! 9.7! Limitations of the current research .......................................................................... 210! 9.8! Future directions ...................................................................................................... 212! 9.8.1! Epigenetics........................................................................................................ 212! 9.8.2! Personality and disinhibited behaviours in risky recreation ............................. 214! 9.9! Conclusions.............................................................................................................. 215! References............................................................................................................................ 217! Appendices........................................................................................................................... 289! Appendix A Descriptions of high-risk sports ................................................................... 289! Appendix B Relationships between impulsive sensation seeking and other approach measures............................................................................................................................ 291! Appendix C Zuckerman Kuhlman Personality Questionnaire (full version, Zuckerman et al., 1993) ........................................................................................................................... 293! Appendix D ImpSS subscale from the ZKPQ, from Zuckerman et al., 1993 .................. 296! Appendix E Summary of variants previously studied in association with approach- or avoidance phenotypes ....................................................................................................... 297! Appendix F Contextual Sensation Seeking Questionnaire for skiing and snowboarding 299! Appendix G Contextual Sensation Seeking Questionnaire for skiing: Peer version ........ 300! Appendix H Sample consent form.................................................................................... 302! Appendix I Correlations between CSSQ-S and ZKPQ subscales .................................... 304! Appendix J DNA concentrations obtained using various techniques and cell types........ 305! Appendix K Buccal swab instructions.............................................................................. 306!  xiii  Appendix L DNA Isolation from buccal cells .................................................................. 307! Appendix M Laboratory recipes ....................................................................................... 308! Appendix N Sample plate layout for Genome Quebec..................................................... 309! Appendix O Primers sequences for PCR amplification of various regions...................... 310! Appendix P Skiers ZKPQ scores shown with “norms” from Zuckerman et al., 1993 ..... 311! Appendix Q Path diagram for confirmation factor analysis of the CSSQ........................ 312! Appendix R Sample PyroQ result for DRD4 -521 C/T .................................................... 313! Appendix S Fisher’s method for combining probabilities................................................ 316! Appendix T Genome Quebec marker list and project report for Festival sample ............ 317! Appendix U Descriptive statistics for analyses at Stages 1 and 2 (chapter 5).................. 319! Appendix V Sample gel photographs for DRD4 120-bp tandem duplication .................. 323! Appendix W Additional analyses for impulsivity and sensation seeking as separate scales (Chapter 6) ........................................................................................................................ 325! Appendix X Psychiatric screening based on medications ................................................ 326! Appendix Y CSSQ generalized for “downhill” sports ..................................................... 328! Appendix Z CSSQ in French ............................................................................................ 329! Appendix AA Sample gel photographs for DRD4 exon III VNTR.................................. 330! Appendix BB DRD4 120-bp tandem duplication data for high- and low-risk athletic samples.............................................................................................................................. 332! Appendix CC Questionnaire used in Project 3 ................................................................. 333! Appendix DD Sport questionnaire used in Project 3 (high-risk sport version) ................ 336! Appendix EE Adult Self Report Scale V1.1 (Kessler et al., 2005) .................................. 338!  xiv  Appendix FF Genome Quebec marker list and project report for high- and low-risk sport samples.............................................................................................................................. 339! Appendix GG Genotype distributions that differed between samples recruited in France and Canada........................................................................................................................ 341! Appendix HH Genotype and allele frequencies for BDNF and STMN1 .......................... 342!  xv  List of Tables  Table 1-1 Correlations between novelty and sensation-seeking instruments ......................... 13! Table 1-2 Summary of commonly employed measures of novelty and sensation seeking .... 14! Table 1-3 Estimated fatality rates for a selection of sports..................................................... 21! Table 1-4 Studies comparing sensation seeking between high- and low-risk athletes or controls.................................................................................................................................... 23! Table 2-1 A summary of reportedly “functional” variants based on the literature................. 52! Table 2-2 Festival sample participant characteristics, pre-exclusions.................................... 60! Table 2-3 Participant characteristics (post-exclusions) for Projects 1 and 2 .......................... 62! Table 2-4 Participant characteristics (pre-exclusions) for Project 3....................................... 63! Table 3-1 Factor loadings from the EFA (n = 198) and CFA (n = 530) for the CSSQ-S....... 73! Table 3-2 Participant statistics for each sample...................................................................... 75! Table 3-3 Results from parallel analyses for 50 randomly generated samples ...................... 78! Table 3-4 Intercorrelations for sensation-seeking score and injury rate................................. 86! Table 3-5 Hierarchical multiple regression analyses predicting injury rate from sensation seeking .................................................................................................................................... 87! Table 4-1 Genetic association studies on -521C/T and approach traits.................................. 94! Table 4-2 Participant characteristics....................................................................................... 97! Table 4-3 Stage 1: A summary of sensation-seeking scores by genotype............................ 103! Table 4-4 Differences in sensation-seeking scores between males and females .................. 104! Table 4-5 Stage 2: A summary of contextual sensation seeking (CSSQ-S) scores by genotype .............................................................................................................................................. 105! Table 5-1 List of single nucleotide polymorphisms chosen for analysis.............................. 112! xvi  Table 5-2 Descriptive statistics and results comparing the samples at stages 1 and 2 ......... 116! Table 5-3 ANCOVA results from stage 1 and 2................................................................... 118! Table 5-4 Descriptive statistics for ImpSS scale and subscales grouped by DRD3 rs167771 .............................................................................................................................................. 119! Table 6-1 A summary of DRD4 promoter polymorphisms .................................................. 127! Table 6-2 Descriptive statistics for demographic and personality variables ........................ 131! Table 6-3 Descriptive statistics and ANOVA results for DRD4 promoter polymorphisms 132! Table 7-1 Comparison of demographic variables between factor levels.............................. 143! Table 7-2 Tests of Hardy Weinberg Equilibrium (HWE) and genotype frequencies........... 144! Table 7-3 CSSQ scores by genotype for the -521C/T and 48-bp VNTR polymorphisms ... 147! Table 8-1 List of SNPS chosen for analysis ......................................................................... 156! Table 8-2 Participant’s characteristics pre- and post-exclusions.......................................... 168! Table 8-3 Participation counts per high- and low-risk sports ............................................... 170! Table 8-4 Bootstrap analysis of personality measures between high- and low-risk participants............................................................................................................................ 173! Table 8-5 Descriptive statistics for personality measures in males and females................. 174! Table 8-6 Comparison of allele frequencies between high- and low-risk athletic groups.... 178! Table 8-7 Personality in high-risk sport males reporting problematic substance use .......... 182!  xvii  List of Figures Figure 1-1. Simplified version of Zuckerman’s psychobiological for impulsive sensation seeking (ImpSS)...................................................................................................................... 26! Figure 1-2. Inverted U-shape model of arousal. .................................................................... 27! Figure 1-3. Dopaminergic neuron.......................................................................................... 30! Figure 2-1. Sample Overview. ............................................................................................... 58! Figure 4-1. Sample gel electrophoresis photograph. ............................................................. 99! Figure 4-2. Sample PyroQ result. ........................................................................................ 100! Figure 7-1. CSSQ scores grouped by genotypes for the 48-bp VNTR (SS vs. SL & LL) and the -521 C/T (CC & CT vs. TT). .......................................................................................... 145! Figure 7-2. CSSQ scores grouped by genotypes for the 48-bp VNTR (SS vs. LL & LS) and the -521 C/T (CC vs. CT vs. TT). ......................................................................................... 146!  xviii  List of Abbreviations 5-HT: serotonin (5-hydroxytryptophan) A: adenine ADHD: attention deficit hyperactivity disorder AGFI: adjusted goodness-of-fit index ANKK1: ankyrin-repeat and kinase-domain-containing-1 ANOVA: analysis of variance ANCOVA: analysis of variance (with covariates) ASD: autism spectrum disorder ASRS: Adult Self Report Symptom checklist BART: balloon analogue risk task BAS: behavioural activation system BASE jump: building, antenna, span, earth (the sport involves parachuting off any of these) BDNF: brain-derived neurotrophic factor BIS: behavioural inhibition system bp: nucleotide base pair BS: boredom susceptibility (subscale of the SSS) C: cytosine cAMP: cyclic adenosine monophosphate CAST: cannabis abuse screening test CD: conduct disorder CFA: confirmatory factor analysis CFI: comparative fit index xix  COMT: catechol-O-methyltransferase CSSQ: Contextual Sensation Seeking Questionnaire DA: dopamine DAT1: dopamine transporter DBH: dopamine-beta-hydroxylase df: degrees of freedom Dis: disinhibition (subscale of the SSS) DNA: deoxyribonucleic acid DRD1, DRD2, DRD3, DRD4: dopamine receptors types 1, 2, 3, or 4 DV: dependent variable EFA: exploratory factor analysis ES: experience seeking (subscale of the SSS) fMRI: functional magnetic resonance imaging g: gram G: guanine GABA: gamma-amino butyric acid GWAS: genome-wide association study H: heterozygosity HA: harm avoidance HTR1A, HTR2A: serotonin receptors types 1A or 2A HWE: Hardy Weinberg Equilibrium Imp: impulsivity L: litre  xx  M: mean or molar MAF: minor allele frequency MAO: monoamine oxidase n: nano unit or sample size NEO-FFI: NEO Five-Factor Personality Inventory NS: novelty seeking PAGE: polyacrylamide gel electrophoresis PCR: polymerase chain reaction PET: positron emission tomography r: correlation coefficient RFLP: restriction fragment length polymorphism RMSEA: root mean square error of approximation RST: Reinforcement Sensitivity Theory s: seconds SD: standard deviation SLC6A3: dopamine transporter SLC6A4: serotonin transporter SNP: single nucleotide polymorphism SPSRQ: Sensitivity to Punishment/Sensitivity to Reward Questionnaire SS: sensation seeking SSS: Sensation Seeking Scales STMN1: stathmin T: thymine  xxi  Taq: thermos aquaticus polymerase TAS: thrill and adventure seeking (subscale of the SSS) TH: tyrosine hydroxylase TCI: Temperament and Character Inventory U: unit UBC: University of British Columbia UTR: untranslated region V: volt VNTR: variable number of tandem repeats ZKPQ: Zuckerman-Kuhlman Personality Questionnaire  xxii  List of Symbols !: alpha, used for Cronbach alpha reliability statistic or threshold for significance ß: beta, regression coefficient  !: Chi statistic ": delta, representing a change #: eigenvalue ": micro unit  xxiii  Glossary -521C/T: a single nucleotide polymorphism (C to T transition 521 bases upstream from the start codon) in the DRD4 promoter region Allele: one version of a genetic variant at a particular polymorphic locus (e.g. ‘C’ or ‘T’ at a ‘C/T’ polymorphism). Approach-related traits: a constellation of personality traits that involve strong motivation towards reward. Personality traits that have been grouped under this broader term include: extraversion, novelty seeking, sensation seeking, reward sensitivity, behavioural activation, and, in some cases, impulsivity. Avoidance-related traits: personality traits that involve sensitivity to punishment. Personality traits that have been grouped under this broader term include: anxiety, neuroticism, behavioural inhibition, and punishment sensitivity. Autoreceptor: a ligand receptor located on the pre-synaptic membrane. Behavioural measures: in psychology, behavioural measures involve assessing behavioural responses using laboratory tasks/paradigms or through observation. These differ from trait measures, which often rely on self-report questionnaires. Disinhibition: a term used in psychology to describe a lack of restraint, this may involve a disregard of social conventions and/or impulsivity. A number of traits are grouped under the broader term of disinhibition, including impulsivity, sensation seeking, and aggression. Disinhibited behaviours: these can range from violence, promiscuity, gambling, substance use, binge drinking, and other socially deviant acts. Dopamine: a neurotransmitter in the brain that has both excitatory and inhibitory functions related to motor control, motivation and reward pathways. Externalizing disorders: a cluster of syndromes generally involving behaviours directed outward (towards others, as opposed to inward, towards self). Disorders grouped under this broad term include: attention deficit hyperactivity disorder, conduct disorder, operational defiant disorder, substance abuse, alcoholism, etc. Factor: items that describe different components of the same larger dimension (e.g. a personality trait) comprise a factor. A factor is a single dimension that is independent, but is composed of highly related items. Factor Analysis: a statistical method used to group items (e.g. in a questionnaire) according to their relatedness.  xxiv  Joint-analysis: a two-stage method of analysis in which p-values from the stages are combined. Firstly, the analysis is carried out in a “discovery” sample, followed by the same analysis in an independent “replication” sample. Gene: a protein-encoding segment of DNA. Genes are located on chromosomes. Genotype: the combination of alleles at a particular locus (e.g. CC, CT or TT at a ‘C/T’ polymorphism). Haplotype: a block of alleles located on the same chromosomal region that are inherited together. Heterozygosity: a measure of the amount of genetic variability at a locus. HPLC: high performance liquid chromatography, a purification method. Knockout: a mutant organism (e.g., mouse) in which the function of a gene has been eliminated. Knockdown: a mutant organism (e.g., mouse) in which the function of a gene has been downregulated. Linkage disequilibrium: when alleles at two or more loci are sorting in a non-random fashion. Combinations of alleles at multiple loci on the same chromosome are present in a greater frequency than would be expected due to chance (if the alleles had been randomly assorting during meiosis). Maximum likelihood rotation: a rotation of eigenvalues by 90 degrees, used in factor analysis. NEO-FFI: NEO Five-Factor model measures five basic dimensions of personality traits including neuroticism, extraversion, openness to experience, agreeableness, and conscientiousness. Neurotransmitter: a brain chemical that transmits signals between neurons. PCR: polymerase chain reaction, a method that allows amplification of a specific DNA segment via varied cycled temperatures. Personality: the consistent pattern of behaviours that is characteristic of an individual. Personality is often organized in conceptual taxonomic structures. Phenotype: an observable characteristic of an individual that is influenced by genetic and environmental factors. Polymorphism: a common variation in DNA in which alternate sequences occur in xxv  populations. Promoter: a segment of DNA that lies before the start codon of a gene that commonly acts as a regulator of gene expression. RFLP: restriction fragment length polymorphism, a method used to identify polymorphisms using an enzyme that either recognizes a specific allele and cuts at the specific site in the sequence, or does not recognize the allele resulting in an un-cut strand. SNP: single nucleotide polymorphism, a variation in the gene sequence that occurs at a single locus (one nucleotide base). Trait: a behavioural characteristic of a person that is stable over time. Varimax rotation: a type of rotation in space that maximizes the variance captured by the items in a factor, used in factor analysis. SSS-IV, V: Sensation Seeking Scales forms IV and V measures four subscales of sensation seeking: thrill and adventure seeking, experience seeking, boredom susceptibility and disinhibition. Tag SNP: a SNP that is representative of other SNPs in the region due to the presence of high linkage disequilibrium. TCI: Temperament and Character Inventory, which was developed by Cloninger to measure four temperaments traits: novelty seeking, harm avoidance, reward dependence and persistence. ZKPQ: Zuckerman-Kuhlman Personality Questionnaire measures five traits: impulsive sensation seeking, aggression-hostility, neurotism-anxiety, sociability and activity. VNTR: variable number of tandem repeats occurs when a segment of the gene sequence is repeated a variable number of times.  xxvi  Acknowledgements I feel that it is important to acknowledge how the idea for this thesis came about. After making a research plan to study exercise-induced asthma during the fall term of my MSc, my supervisor, Dr. Jim Rupert asked me: “why do you and your friends spend all of your free time skiing/climbing? Do you think there is something genetic that drives this urge?” I subsequently spent the winter break diving into the research, because in my opinion this was the coolest thing I could imagine for a thesis project. Thank you, Jim, for inspiring me to delve deeper into this question, and for understanding that my love for recreation made this project the best fit for my graduate research. Jim has been an incredibly supportive supervisor, always willing to provide thoughtful insight and advice. I am especially thankful for Jim’s willingness to join me in entering the world of personality genetics, a field that was new to both of us. I have great respect for Jim’s never-ending curiosity about a vast range of topics, his expansive knowledge, and his ability to maintain balance despite the demands of academic life. I would also like to thank the members of my supervisory committee. A big thank you to Dr. Scott Carlson (U. Minnesota) for being my mentor in the field of psychology. I could not have done this project without Scott’s valuable insight and his patience for teaching me basic psychological concepts. Scott was incredibly generous with his time, and treated me as if I were one of his students. Dr. Wendy Robinson provided valuable insight and incredible attention to detail. Wendy has also been very generous in her laboratory space and equipment and helped in fostering collaborations with her graduate students. I am grateful for her support. Dr. Robert McMahon joined my committee late in my PhD and I am thankful that he was willing to invest time getting up to speed. He provided valuable input in the final stages of analysis and I thank Bob for his wise words and feedback. I met many incredible people through my research and I owe a big thanks to the athletes for being interested and for taking the time to participate in my studies. Thank you to Lisa Richardson for being a champion of the project and for getting me a kiosk at a festival in Whistler. I am very grateful to CIHR (Canadian Institute for Health Research) for believing in my project and generously funding me throughout my doctoral degree. I also thank BC Mental Health and Addictions Research Network and the UBC Faculty of Education for their support xxvii  towards my thesis project. Finally the Michael Smith Foreign Supplement Grant allowed me to carry out a portion of my research in France. The international experience was unforgettable and invaluable to my project, and I appreciate the support. I would like to thank my French colleagues. Prof. Grégory Michel (U. Bordeaux) has been very supportive and enthusiastic throughout our collaboration. He shared his valuable expertise on sensation seeking and I look forward to continuing our collaborations in the future. Julie Salla, a PhD student, was also incredibly supportive during my time in France. I feel lucky that I was able to work with such bright, kind, helpful students during my time in the Rupert Lab. Dr. Pei Wang taught me everything I know about laboratory procedures, and I owe her a great debt of gratitude for her time. Martin MacInnis is a statistics whiz and has a keen eye for detail, and all of the other students have been supportive in helping me prepare for presentations, or for brainstorming ideas, I am so thankful. Thank you also to Courtney Hanna, a master in the lab, for teaching me new techniques and to Katie Morton for assisting with questionnaire development. A special thanks to my awesome summer/volunteer students: Amelia Rajala, Nora Hase (vielen Dank), and Becky Power. You ladies were amazing to work with and helped keep me sane. Diving head first into a project can sometimes result in periods of “drowning”, and I will admit to having had a few of these moments. I honestly could not have completed this thesis without the support of my friends and family. My parents and siblings are so loving and supportive. I am lucky to have such a wonderful family, always ready to listen, or to tell me to “get over it” (whichever is necessary at the time!). Most importantly, though, was the support of my husband, Chris. I owe Chris the biggest thanks of all. He literally kept me going. I have a tendency to disappear into my work, but Chris made sure to force me to take breaks and he fed me amazing meals. He listened to me nerd-out about genetics and psychology, he helped recruit participants, he even assisted me in the laboratory when I was injured… but most importantly, he kept me laughing. He has been understanding of my work habits, and has kept me sane throughout the last 5 years. He is an adventurer, and was invaluable for bouncing off ideas, and for discussions about the unique world of high-risk sports. He is my life partner, my ski partner, and my climbing partner, and I cherish all of the time that we spend doing the things that we love to do. Thank you for keeping things in perspective and for being my number one supporter.  xxviii  Dedication  I dedicate this work to my husband, Chris, his passion for adventure, and joie de vivre!  xxix  Chapter 1: Introduction 1.1  Overview Personality is a combination of qualities or traits that can influence many aspects of our  lives, including choice of friends, partners, career, and even what we do in our spare time. This dissertation focuses on personality traits related to approach motivation, namely sensation seeking, and how traits influence choice of sport and patterns of behaviours within certain sports. It is generally accepted that personality traits are influenced by complex interactions and relationships between genetics and the environment, and genetic association studies are one common method to investigate the relationships between genotype and phenotype. Sport provides a domain in which context-specific tools can be used alongside general trait measures to quantify phenotypes for genetic association. There exists a substantial body of literature that has explored associations between approach-related traits and genetic variants in both healthy and clinical populations; and similarly, the evidence linking sensation seeking and sport participation is overwhelming, but prior to carrying out the current research program, no studies had used self-reported patterns of athletic behaviours and participation to define a prosocial phenotype for approach-related traits. This dissertation explores psychological and genetic characteristics of athletes who participate in high-risk sports. The first series of projects involve developing and testing a tool to measure patterns of sensation seeking in multiple samples of skiers and snowboarders. The second project involves investigating associations between genetic variants and the newly developed domain-specific phenotype (along with established trait measures) in skiers and snowboarders. Finally, personality and genetic variants are investigated in independent samples of high- and low-risk sport participants. 1  1.2  Why study sport? To an uninformed bystander, soloing (climbing without protection) a rock face may seem  as, if not more, reckless than taking drugs or gambling. Often times, what the onlooker does not realize is that the climber is highly practiced in his/her sport, has likely climbed the route dozens of time (with protection), memorizing each move, and has full confidence in his/her abilities to complete the task unharmed. Regardless, there is still an inherent risk to climbing a high wall without protection. So why take the risk? Many athletes claim that the risk is not the goal, instead, they seek an inner peace or connectedness to the natural world (Brymer & Oades, 2009), although risk-taking, escape from boredom, and seeking challenge and excitement are other common themes that emerge from the qualitative research data (Hallin & Mykletun, 2006; J.H. Kerr & Mackenzie, 2012). There is an extensive body of literature in which researchers have examined characteristics of elite practitioners of high-risk sports in an attempt to understand the motivation for participation, the personalities most likely to pursue such sports, along with the perceived risks and benefits of the activities. Fewer studies have explored the underlying physiological mechanisms driving motivation, and few-to-none have explored the potential of a genetic underpinning. High-risk sports carry a potential for severe injury or death as an inherent part of the activity (Willig, 2008). They are commonly referred to as “adventure” or “extreme” sports. The latter term was picked up by marketing companies to promote competitions like the X-games or energy drinks (e.g., Xtreme), and many high-risk sport participants do not identify with this term (personal communications). In this dissertation the terms “high-risk sports” or “adventure sports”, will be used rather than “extreme sports”.  2  High-risks recreation is not new, people have been practicing parachuting, paragliding, alpine skiing, and surfing (and many other sports involving risk) for over a century, but participation rates saw drastic increases over the last 30 years (reviewed in Celsi, Rose, & Leigh, 1993). In the United States participation levels in these types of sports increased 244% from 1978 to 2000 (Puchan, 2004). With the rise in high-risk sport participation, industry has capitalized on the market and there has been an increase in media devoted to the sports (including magazines, films, and TV series, e.g., Fear Factor), further increasing the popularity of these once fringe sports (Creyer, Creyer, Ross Jr, & Evers, 2003). With the surge in participation rates there has also been a corresponding surge in injuries and deaths (Creyer et al., 2003; Le Breton, 2000) and these have been documented by reviews on adventure sports (e.g., Celsi et al., 1993). The repertoire of high-risk sports has increased over the years as well, with the evolution of skydiving (with roots in the military) to BASE-jumping (jumping off a building, antenna, span, or the earth (i.e., a cliff) in 1978 (Hallin & Mykletun, 2006), and sports that are just gaining momentum now include high-lining (walking across a slack line of nylon webbing spanning a high gap) and speed flying (skiing with a kite). An extensive list of high-risk sports, including descriptions of each, is included in Appendix A. Understanding the motivations that drive people to participate in high-risk sports may help injury prevention research, but may also be useful for treatment and prevention of other deviant risk-practices. There are similarities in personality profiles of high-risk athletes and deviant risk takers (e.g., Franques et al., 2003), but there are also interesting differences between these risk-inclined populations (e.g., Goma-I-Freixanet, 1995; Goma-i-Freixanet, 2001). Prosocial and antisocial risk takers often report high levels of personality traits associated with seeking rewards, but athletes may differ from deviant risk-takers on traits involved in planning  3  and premeditation (discussed in detail below). Research into personality, emotional regulation, neurophysiology, and genetics are all avenues for understanding motivation for high-risk recreation.  1.3  Sensation seeking While motivation to participate in high-risk sport may vary between individuals, high  sensation seeking is a common trait among these athletes (Goma-i-Freixanet, Martha, & Muro, 2012). Sensation seeking, involves, “… the seeking of varied, novel, complex, and intense sensations and experiences, and the willingness to take physical, social, legal, and financial risks for the sake of such experience” (p. 27, Zuckerman, 1994).  The willingness to take risks is not the goal per se, but an important correlate of the trait (Zuckerman, 1994). Sensation seekers often underestimate the risk associated with an experience, or place such a high value on the reward that they are willing to accept the risk, but it is important to note that few prosocial sensation seekers seek the risk itself (Zuckerman, 1994). Not surprisingly, sensation seeking has been associated with a variety of high-risk social activities including unprotected, casual sex (Kalichman, Cain, Zweben, & Swain, 2003; Zuckerman & Kuhlman, 2000), illicit drug use (M.T. Bardo et al., 2007), gambling (Rosenbloom, 2003; Verdejo-Garcia, Lawrence, & Clark, 2008) and crime, as well as with highrisk sport participation (reviewed in Goma-i-Freixanet et al., 2012; for review of all domains, see: Roberti, 2004). Marvin Zuckerman coined the term “sensation seeking” in the 1970s, following a series of sensory deprivation experiments, in which he noticed shared characteristics among study volunteers. Most of the volunteers were “free-spirited”, young men who dressed in an alternative fashion (long hair, ripped jeans) and were drawn to the study by rumours of euphoric 4  side-effects. Zuckerman would later classify these study volunteers as “high sensation seekers” (Zuckerman, 1979). Soon thereafter, C. Robert Cloninger defined a similar trait, “novelty seeking” (NS) as a dimension in his tri-dimensional personality questionnaire (TPQ) (Cloninger, 1987). Novelty seeking is defined as “… a tendency toward intense exhilaration or excitement in response to novel stimuli or cues for potential rewards or potential relief of punishment” (p. 574, Cloninger, 1987). The two traits, novelty and sensation seeking, are often used interchangeably in behavioural genetics and psychology literature today.  1.3.1  Sensation seeking and other personality models: Approach traits Novelty and sensation seeking are moderately correlated (r = 0.4 to 0.7) (McCourt,  Gurrera, & Cutter, 1993; M. R. Munafo, Yalcin, Willis-Owen, & Flint, 2008; Zuckerman & Cloninger, 1996) and are members of a larger group of “approach-related traits’” which also include extraversion, sociability, and impulsivity (M. R. Munafo et al., 2008; Zuckerman, 2005a). Motivational tendencies can broadly be grouped as “approach” and “avoidance” orientations, and approach traits involve an increased sensitivity to positive stimuli (or removal of negative stimuli) (Elliot & Thrash, 2002). The Reinforcement Sensitivity Theory (RST) (Gray & McNaugton, 2000) is a contemporary theory of personality that influenced Zuckerman’s neurobiological model of sensation seeking (Zuckerman, 2007a), also pertaining to sensitivity to stimuli. A brief description of RST is provided below in order to provide context for discussing sensation seeking as an approach trait. For full review of RST as it relates to personality, please refer to Corr (2004). The RST proposes that individual differences exist in sensitivity to unconditioned and conditioned signals of reward and punishment (Gray & McNaugton, 2000). Under the RST,  5  there are systems governing behavioural activation (BAS) and inhibition (BIS): the former is thought to respond in a goal-directed manner to appetitive stimuli and to the removal of aversive stimuli (i.e., positive and negative reinforcement), while the latter is involved in resolving goal conflicts that can arise between BAS activation and activation of a third system, the Fight-FlightFreeze System (FFFS) which is activated upon presentation of aversive stimuli and cues predicting positive or negative punishment (Corr, 2004). BIS influences how an individual values a reward over the risk of punishment, e.g., the pleasure of speeding down a mountain versus the risk of losing control by going too fast. Self-report measures that putatively relate to the RST, including Carver and White’s (1994) BIS/BAS scale, and the Sensitivity to Punishment and Sensitivity to Reward Questionnaire (SPSRQ) by Torrubia, Avila, Moto and Caseras (2001), have been used in a few high-risk-sport studies, but are less commonly employed compared to the sensation-seeking measures (e.g., Goma-i-Freixanet et al., 2012). Other self-report measures may relate to the RST, as Elliot and Thrash (2002) proposed that several personality traits could be grouped into the broad temperaments of approach and avoidance orientation. Using factor analytic processes they found measures of extraversion, positive affect, and behavioural activation loaded together as approach traits, sharing a sensitivity to positive stimuli (i.e., reward); while measures of neuroticism, negative emotionality, and behavioural inhibition loaded together as avoidance, sharing a sensitivity to negative stimuli (i.e. punishment) (Elliot & Thrash, 2002). Approach-traits tend to reflect heightened reward sensitivity and strong BAS, whereas avoidance-traits are influenced jointly by activation of the BIS and FFFS and reflect punishment sensitivity (Corr, 2004). The trait most commonly associated with strong BAS is impulsivity (discussed in section 1.3.2), but sensation seeking also putatively relates to strong BAS (Corr, 2004; Zuckerman, 1994). An overview of sensation  6  seeking as it relates to reward sensitivity, behavioural activation, and approach is provided in order to understand neuro- and psychophysiological paradigms and to understand how metaanalytic studies of personality genetics have grouped multiple traits under the broader term “approach”.  1.3.2  Impulsivity Impulsivity is a multidimensional trait that involves acting without forethought (Evenden,  1999). Depending on the instrument, impulsivity sometimes appears as a subfactor within instruments that measure either novelty seeking and sensation seeking (Cloninger, 1987; Zuckerman, Kuhlman, Joireman, Teta, & Kraft, 1993), or vice versa, i.e., measures of impulsivity include sensation seeking as a subfactor. Zuckerman’s latest personality instrument (1993) places sensation seeking within the broader trait “impulsive-sensation seeking”, where sensation seeking represents the tendency to approach novel stimuli, and impulsivity governs the decision-making style of whether or not to approach, and although they represent distinct facets, they load together onto a single factor (Zuckerman, 1994, 2005b; Zuckerman et al., 1993). Various measures of impulsivity correlate both with Zuckerman’s combined impulsive sensation seeking scale and with each component (impulsivity and sensation seeking) separately (see Appendix B for results from an unpublished study); however, many studies demonstrate that they are dissociable traits (Cross, Copping, & Campbell, 2011; Flory et al., 2006; Magid, MacLean, & Colder, 2007). Impulsivity is not considered a unitary construct (Evenden, 1999), and commonly at least three dimensions of impulsivity have emerged from factor analysis of data derived from various impulsivity questionnaires, these include: reward sensitivity (and approach motivation), punishment sensitivity (and avoidance), and higher order cognitive impulsivity  7  (sometimes called rash impulsivity) (Cross et al., 2011; I. H. A. Franken & Muris, 2006; Miller, Joseph, & Tudway, 2004; Quilty & Oakman, 2004; L. D. Smillie, Jackson, & Dalgleish, 2006). A person can be impulsive without being high in sensation seeking, and vice versa. For example, a sensation-seeking skier might be inclined to jump off a small cliff, but whether or not the skier scopes out the landing ahead of time may depend on his/her trait impulsivity. Participation in many high-risk sport activities requires planning and expertise and is more likely related to the approach motivation aspects of impulsivity. Not surprisingly, participation in highrisk sports is more commonly associated with sensation seeking (Goma-i-Freixanet et al., 2012), whereas associations with impulsivity independent from sensation seeking are rare (i.e., no associations found with high risk sport participation (Goma-I-Freixanet, 1991, 1995; Goma-iFreixanet, 2001; Jack, Jack, & Ronan, 1998; J. H. Kerr & Svebak, 1989). A variety of nonathletic risk-taking behaviours are associated with high scores on the combined impulsive sensation seeking scale (Zuckerman & Kuhlman, 2000), but sensation seeking and impulsivity are also independently associated with high-risk behaviours, many of which are discussed below in section 1.4. The following sections focus mainly on correlates of sensation seeking, the trait most commonly associated with sport participation, with less focus on impulsivity as an independent factor.  1.3.3  Laboratory correlates of sensation seeking There are laboratory paradigms that measure behavioural inhibition and impulsivity  (Cross et al., 2011; Verdejo-Garcia et al., 2008), and there exist laboratory tasks that measure risk taking, but there are fewer tasks designed specifically to measure sensation seeking. There are, however, tasks that are moderately correlated with sensation seeking. Two such risk-taking  8  tasks include the Balloon Analogue Risk Task (BART) and a balance beam activity. The BART involves pushing buttons on a computer to inflate a balloon and monetary gains are incurred with each pump, at the risk of the balloon exploding due to over-inflation (maximum balloon circumference randomly altered per trial) (Lejuez et al., 2002). Risk taking as assessed using the BART was moderately correlated with sensation seeking scores (Lejuez et al., 2002). Performance on the BART has also been related to impulsive sensation seeking, in that low scorers were more risk averse when there was a greater potential for loss, whereas high scorers did not adjust their behaviour despite increased risk for loss (Bornovalova et al., 2009). Another risk-taking activity for children designed by Morongiello and colleagues (2006) involves crossing a balance beam set to variable heights and they found that the degree of risk taking was correlated with sensation seeking. Brocke and colleagues (1999) attempted to move beyond behavioural correlates of sensation seeking by testing three paradigms: continuous performance task, delayed reaction time task, and the augmenting-reducing paradigm in an attempt to validate behavioural paradigms that would reliably predict (or act as indicators of) sensation seeking. The chosen tasks had been used in attention deficit hyperactivity disorder (ADHD) research and the authors suggested that ADHD is an extreme manifestation of sensation seeking. They observed significant correlations between the tasks and sensation seeking, but only the delayed reaction time task was able to predict a small portion of variance in sensation seeking (Brocke et al., 1999). None of the above-mentioned tasks were designed specifically for sensation seeking, and although risk-taking and attentional tasks correlate with self-report measures of sensation seeking, they are only capturing a facet of the complex trait. Other behavioural correlates of the sensation-seeking trait involve measuring orienting responses to novel stimuli (either visual or auditory). Zuckerman (1994, 2005b, 2007a) reviews  9  studies finding that high sensation seekers have stronger orienting responses, measured by skin conductance and heart rate deceleration, to novel, non-aversive, intense stimuli than low sensation seekers. Researchers can vary the novelty or intensity of a stimulus in order to elicit orienting responses, startle responses, or defensive responses. Orienting responses are typical of low to moderate intensity stimuli, whereas high intensity stimuli may result in a defensive response, or a startle response if the stimulus is unexpected. High and low sensation seekers differ in their response to the same intensity stimulus, and a moderate tone can elicit an orienting reflex in high scorers and a defensive reflex in low scorers (Zuckerman, 1994). High sensation seekers also show a positive correlation between cortical arousal responses as measured by visual and auditory evoked potentials and stimulus intensity, whereas the converse is seen in low sensation seekers. These patterns of responses, which are referred to as either “augmenting” or “reducing”, have been observed in animal models (including cats and rats). Augmenter cats (those showing increasing evoked potentials in response to intense stimuli) displayed more exploratory behaviours (analogous to sensation seeking in humans), while reducer cats showed more withdrawal (reviewed by Siegel, 1997). Differences between high- and low-sensation seekers are observed not only in cortical responses, but hormonal responses too. Both animal and human experiments have shown that when exposed to aversive stimuli or stressors, high sensation seekers show a blunted cortisol response (Zuckerman, 1994, 2007a). Another task that has been shown to differentiate high- and low-sensation seekers involves measuring acoustic startle reflex after presentation of images differing in their affective valence. The International Affective Picture System is a set of normative emotional stimuli often used to measure arousal (P. J. Lang, Bradley, & Cuthbert, 1997); and images are grouped by affective valence: positive, neutral, or threatening. Startle responses in low sensation seekers  10  were increased after the presentation of threatening images compared to neutral images, but no differences were observed in high sensation seekers (Lissek & Powers, 2003). The data from this study suggest that high sensation seekers have weaker avoidance systems, which is further supported by studies described below. More recently, imaging studies have shed light on regions of the brain that are activated during laboratory tasks that measure phenotypes related to sensation seeking (i.e., risk-taking, arousal). Joseph and colleagues (2009) used functional magnetic resonance imaging (fMRI) to show that brain regions are differentially activated in high- and low-sensation seekers in response to intense visual stimuli from the International Affective Picture System, and the results supported the framework for strong approach and weak avoidance systems in the high scorers. Similarly, Kruschwitz and colleagues (2012) observed differences in magnitude and location of neural activation (recorded using fMRI) between approach and avoidance related responses in a Risky Gains Task1 in high and low sensation seekers. High sensation seekers show greater activation to reward compared to low sensation seekers, and low scorers showed greater activation to punishment compared to high scoring individuals (Kruschwitz et al., 2012). Taken together, these findings provide support for strongly activated approach systems and weak avoidance systems in high sensation seekers.  1.3.4  Sensation-seeking instruments There are a number of laboratory tasks that measure impulsivity, and we have seen that  there are tasks that correlate with sensation seeking, but many researchers measure sensation  1  Numbers are displayed on a screen and participants can choose to either accept the choice or to wait for a higher value number (at the risk of being presented a negative number). The numbers each represent a monetary reward (e.g., 20 = $0.20), but some numbers represent negative values. Participants can select the lowest (safe) value, or can wait for higher values at the risk of a negative (punishment) value. 11  seeking using self-report questionnaires. These include the Sensation Seeking Scale (SSS) versions IV or V (Zuckerman, 2006); the Brief Sensation Seeking Scale (BSS) (Hoyle, Stephenson, Palmgreen, Lorch, & Donohew, 2002); the Arnett Inventory of Sensation Seeking (AISS) (Arnett 1994); and questionnaires that contain sensation seeking as a factor within a broader questionnaire: the UPPS (Whiteside & Lynam, 2001) and the Zuckerman-Kuhlman Personality Questionnaire (ZKPQ, Appendix C) (Zuckerman et al., 1993). There is considerable overlap of items between the BSS, UPPS and ZKPQ, however the SSS, BSS, and UPPS all contain items with specific references to sports (e.g., water skiing, parachute jumping, skiing, and bungee jumping), while the ZKPQ Impulsive Sensation Seeking scale (ImpSS) makes no such references (Zuckerman, 2007a). The BSS, UPPS and ZKPQ also use modern language compared to the SSS. For example, the SSS uses the words/phrases “hippie”, “surf-board riding”, “far out friends”, and is not reflective of some social norms of today (e.g., SSS item: “I stay away from anyone I suspect of being “gay” or “lesbian”). Taking these into consideration, the ZKPQ is better suited for sport populations containing a large proportion of young adults, and because it lacks potentially confounding items relating to sport. Zuckerman’s SSS contains four subscales, thrill and adventure seeking (TAS), experience seeking (ES), boredom susceptibility (BS), and disinhibition (Dis). Items from each subscale relate to different expressions of sensation seeking; for example TAS reflects a desire for exciting physical activities, ES reflects a desire for exciting cultural experiences, BS is an aversion to monotony, and Dis reflects seeking excitement through social expressions (Zuckerman, 2005b). The ZKPQ ImpSS scale emerged from factor analysis of the SSS items, and therefore contains elements from each of the four expressions (Zuckerman et al., 1993). See  12  Table 1-1 for correlations between subscales of novelty- and sensation-seeking measures, and Table 1-2 provides a summary of novelty and sensation-seeking instruments and subscales.  Table 1-1 Correlations between novelty and sensation-seeking instruments Correlations between novelty and sensation-seeking instruments Instrument-Subscale Correlation with ZKPQ ImpSS SSS-total .66a SSS-TAS .43a SSS-Dis .43a SSS-BS .37a SSS-ES .43a TCI-NS-total .68b a Note. Data obtained from the following sources: (Zuckerman, 2007a); b(Zuckerman & Cloninger, 1996).  13  Table 1-2 Summary of commonly employed measures of novelty and sensation seeking Summary of commonly employed measures of novelty and sensation seeking  Brief description Approach measure  Instrument Zuckerman SSS Trait measure of total sensation seeking, divided into four subscales. Total sensation seeking (SSSTot) Boredom susceptibility (BS) Disinhibition (Dis) Experience seeking (ES) Thrill and adventure seeking (TAS)  Other traits  Zuckerman-Kuhlman (ZKPQ) Personality measure, five dimensions (traits). Impulsive sensation seeking (ImpSS) Impulsivity (Imp) Sensation seeking (SS)  Cloninger TCI Personality measure, three dimensions (temperaments), further subdivided into subscales. Novelty seeking (NS) Exploratory excitability vs. stoic rigidity (NS1) Impulsiveness vs. reflection (NS2) Extravagance vs. reserve (NS3) Disorderliness vs. regimentation (NS4) Harm avoidance (HA) Reward dependence (RD)  Sociability (Soc) Aggression-hostility(AggHos) Neuroticism-anxiety (NeuAnx) Activity (Act) Note. SSS = Sensation Seeking Scale, ZKPQ = Zuckerman-Kuhlman Personality Questionnaire, TCI = Temperament and Character Inventory.  The ZKPQ measures five personality traits, but its ImpSS (the instrument used for this dissertation) has been used in isolation. The ImpSS contains a total of 19 items, but may also be divided into its two factors: impulsivity (Imp, eight items) and sensation seeking (SS, 11 items), (Appendix D). The impulsivity factor measures lack of planning and forethought; while the sensation-seeking factor measures the desire to seek out new and thrilling experiences and the willingness to take risks (Zuckerman et al., 1993). The two-component scale allows for the consideration of impulsivity and sensation seeking as dissociable traits (e.g., Cross et al., 2011). The full ZKPQ has shown high re-test reliability (.82 to .87, retest interval of 2 months), and internal consistencies for the ImpSS subscale were .77 and .81 for males and females respectively (Zuckerman & Kuhlman, 2000), and .83 for ImpSS in an unpublished sample 14  (Thomson, 2008). The ImpSS scale used independently has also shown acceptable internal consistency (.84 and .87 in two samples, McDaniel & Mahan, 2008; .77 in one sample, Robbins & Bryan, 2004). Although the SSS remains the most commonly employed tool for assessing sensation seeking, there is support for the ImpSS as a valid, reliable alternative (McDaniel & Mahan, 2008). The ImpSS demonstrates concurrent validity with the SSS (McDaniel & Mahan, 2008), and based on a meta-analysis the reliability estimates of the SSS subscales are less than .70 (except for the TAS which is .75) (Deditius-Island & Caruso, 2002), which is generally lower than reliability estimates for the ImpSS (McDaniel & Mahan, 2008; Zuckerman, 2007a). Finally, the ImpSS is brief enough to administer in the field. While a majority of psychology and sociological studies continue to measure sensation seeking using the SSS (and ImpSS), it appears that a majority of studies in the fields of neuroscience and behaviour genetics measure approach using Cloninger’s Temperament and Character Inventory (TCI; Cloninger, 1987) novelty seeking subscale (perhaps because of language used to describe the traits in rat and mouse models). The disparity between the fields with respect to phenotype definition is further exemplified in the sport literature, where, to my knowledge, there exist only a handful of studies that have investigated approach in athletes using a tool to measure novelty seeking, since most sport studies measure sensation seeking.2 I have chosen to measure sensation seeking to be consistent with the sport psychology literature and the suitability of the sensation-seeking content for a sport population. Although the research described in this thesis focuses on impulsive sensation seeking, when reviewing the neuroscience  2  A “web of science” search of the terms “sport” + “novelty seeking” resulted in two articles; whereas a search for the terms “sport” + “sensation seeking” resulted in 157 articles. When searching the terms “genes” + “novelty seeking” as opposed to “genes” + “sensation seeking”, the novelty seeking search resulted in seven times (total > 700) more articles, and of this fraction of studies that mention sensation seeking, many actually measured novelty seeking as the phenotype of interest. 15  and genetics literature, literature from both novelty seeking and sensation seeking measures are reviewed and the overall results are discussed as they relate to “sensation seeking” from herein.  1.3.5  Demographic trends in sensation seeking Sensation seeking varies between the sexes and with age. Sensation seeking is positively  associated with “masculine” interests and characteristics and negatively associated with “feminine” interests and characteristics in North American culture (Daitzman & Zuckerman, 1980; Kish, 1971), and has been related to lower levels of felt gender compatibility among women (Saxvik & Joireman, 2005). In line with these findings, men consistently score higher on all sensation-seeking subscales except experience seeking (Zuckerman, 2007a). The largest differences between men and women are on the thrill and adventure seeking and disinhibition subscales from the SSS (Zuckerman, 2007a). Men also score higher on ImpSS than women (Thomson, 2008; Zuckerman & Kuhlman, 2000), though the differences in ImpSS scores between the sexes are less pronounced than sensation seeking measured using the SSS (Cross et al., 2011). Sensation seeking is lower in children, and scores generally peak in adolescence, and decline thereafter (Zuckerman, 1979, 2007a), although longitudinal research on sensation seeking in children is lacking. Significant age-related decline in all subscales except boredom susceptibility have been reported (Zuckerman, 2007a). Even after removing “age-dependent” items, such as those that require a certain level of physical fitness (e.g., skiing, mountaineering), the same post-adolescence inverse relationship between sensation seeking and age was observed (Roth, Hammelstein, & Braehler, 2007). Despite this overall decline, individuals maintain a relatively stable rank within a cohort as they age, in that high sensation seekers in adolescence  16  remain at the higher end of their cohort into adulthood (M.T. Bardo et al., 2007). Sensationseeking scores also purportedly vary between self-reported “ethnicities”3. Individuals reporting African American descent score lower on SSS subscales, with the exception of disinhibition compared to individuals of European descent (Zuckerman, 1994), and Western cultures score higher than Asian cultures, although these cultural differences have not been consistently reported across all sensation-seeking subscales (M.T. Bardo et al., 2007). Other demographic variables (e.g., marital status, education, religion) have also been associated with sensation-seeking scores. Divorced males scored significantly higher than single and married males and there was a similar trend in females, although the finding was not significant (Zuckerman & Neeb, 1980). Sensation seeking appears to play a role in mate selection, in that high sensation seekers tend to marry individuals similarly high in the trait (Bratko & Butkovic, 2003; Zuckerman, 2007a), and sensation seeking is an exception to the usually small-to-zero spousal correlations for most personality traits (e.g., Donnellan, Conger, & Bryant, 2004).  1.4 1.4.1  Lifestyle correlates of sensation seeking Deviant high-risk behaviours Sensation seeking and many related disinhibited traits have been associated with a  number of high-risk behaviours. Mind-altering drugs, gambling, risky sex, and crime are common outlets for satisfying exploratory urges (reviewed in M.T. Bardo et al., 2007; Zuckerman, 2007a). Numerous studies have shown that individuals who partake in these deviant  3  Depending on the self-report choices, the term “race” is sometimes used instead of “ethnicity”. For example, the term Hispanic does not refer to a race, but is an ethnic group, but researchers that assess ancestry based on categories like “white” or “black” are referring to race. There are inconsistencies in the literature, but most researchers use the term ethnicity as it applies to a population sharing common ancestry. 17  activities report higher levels of sensation seeking, and the trait is considered a reliable predictor of drug use and abuse (M.T. Bardo et al., 2007). One study found that sensation seeking was a stronger predictor of drug-use amongst adolescents than socio-economic status, self-esteem, and mental health (W. Pedersen, Clausen, & Lavik, 1989). Similarly, sensation seeking predicted tendencies for both occasional and frequent risk-taking across multiple contexts (including substance use, driving, sexual relations), whereas demographic variables (i.e., age) were only associated with frequent risk-taking (Desrichard & Denarie, 2005). Disinhibited traits, including sensation seeking have also been correlated with alcohol use (Sher, Grekin, & Williams, 2005). A meta-analysis found low to moderate effect sizes for associations between sensation seeking and alcohol use, but the size of the effect decreased in studies that included demographic covariates (e.g., age, sex, socio-economic status) (Hittner & Swickert, 2006). Another study found that the relationship between sensation seeking and alcohol use disappeared with the inclusion of additional variables, such as high scores on an inventory designed to measure antisocial features (Whiteside & Lynam, 2003). While other personality and demographic variables are likely to contribute to the aetiology of alcohol use and substance use disorders, sensation seeking is considered an important indicator of liability to develop patterns of disordered use (Kelly et al., 2006; Stacy & Newcomb, 1999). Disinhibition4, in particular, is the sensation-seeking facet (from the SSS) most strongly associated with alcohol use (Hittner & Swickert, 2006); however, the subscale includes a number of items that contain potentially confounding references to substance and alcohol use. Carlson and colleagues (2010) included only the thrill-and-adventure-seeking and boredom-susceptibility  4  Note: the term “disinhibition” does not refer to the lack of inhibitions a person might experience while under the influence of alcohol, but refers to the personality dimension within the SSS, which is reflective of seeking excitement through social situations. 18  subscales from the SSS (to avoid subscales containing references to substance-related behaviours) in a model to predict binge drinking, and found that both SSS subscales were significant predictors. Other common outlets for sensation seeking include gambling, risky sexual behaviours, and criminal activities. Higher sensation-seeking scores have been reported in studies of pathological gambling (Potenza et al., 2003), but the findings are not ubiquitous, and some studies report no differences in sensation-seeking scores between pathological and nonpathological gamblers or between scores on gambling questionnaires in non-clinical populations (reviewed by Hammelstein, 2004; Zuckerman & Kuhlman, 2000). The findings for high sensation seeking among individuals who take sexual risks are more consistent (Hoyle, Fejfar, & Miller, 2000; Zuckerman, 2007a). One-night-stands, sexual acts under the influence of alcohol and drugs, unprotected sex, and promiscuity have been related to sensation seeking and impulsive sensation seeking (Donohew et al., 2000; Zuckerman & Kuhlman, 2000). Finally, a range of unlawful activities have been linked to sensation seeking. These include speeding, drinking and driving, vandalism, and theft (Horvath & Zuckerman, 1993; reviewed in Zuckerman, 2007b). Sensation seeking is moderately correlated with general deviance (e.g., attitudes towards law abidance, sexual events, illicit substance use) in adolescence and young adulthood, but a longitudinal study following youth from grades 10-12 until their early 20s found that the trait was not predictive of general deviance over time. Instead, it was found that sensation seeking predicted specific deviance related to substance use, and use of substances then predicted general deviance (Newcomb & McGee, 1991). Similarly, high sensation seeking is seen in schizophrenic patients, but more commonly when individuals have a comorbid substance use disorder or alcoholism (e.g., Dervaux et al., 2010; Zhornitsky et al., 2012). While sensation  19  seeking is most commonly associated with substance use disorder and alcoholism, the trait has been associated with characteristics common to other externalizing disorders, including conduct disorder (reviewed in J. S. Kotler & McMahon, 2005) and ADHD (Faraone et al., 1999).  1.4.2  Sensation seeking and sport Sensation seeking has been associated with disinhibited behaviours described above, but  also to non-deviant, prosocial outlets like travel and entrepreneurship (Lepp & Gibson, 2008; Nicolaou, Shane, Cherkas, & Spector, 2008). High-risk sports are another prosocial outlet for sensation seeking. High-risk sports were once fringe activities, but are now gaining mainstream popularity with increased media exposure and accessibility. Sports are considered “high-risk” when there is a high chance of severe injury or death if something goes wrong during the activity (i.e., falling, equipment failure, weather change, etc.) (Willig, 2008; Zuckerman, 2007b), yet people are drawn to such sports, even with full disclosure of the inherent dangers they present. Table 1-3 shows estimated fatality rates for select sports. Examples of high-risk sports include paragliding, skydiving, scuba diving, downhill mountain biking, mountaineering, big mountain skiing and snowboarding, surfing, high-lining, and more high-risk sports are invented each year (for an expansive list, please see Appendix A).  20  Table 1-3 Estimated fatality rates for a selection of sports Estimated fatality rates for a selection of sports Sport Fatality rate Year of census Reference BASE 1/2317 jumps 1995-2005 1 Skydiving 1/18333 jumps 1979-1983 2 Mountaineering 1/1 000 1983 3 >7000 m peaks 1/5 expeditions 1986 4 Hang gliding 1/1250 flights 1983 3 Scuba diving 1/100 000 dives 1983 3 Skiing 1/150000a skier days 1996-2006 5 Professional Football 1/200000 1983 3 Note. apermanent injury or death (Alberta & British Columbia). 1) (Soreide, Ellingsen, & Knutson, 2007); 2) (Ellitsgaard, 1987); 3) (Celsi et al., 1993); 4) (Pollard & Clarke, 1988); 5) (McBeth, Ball, Mulloy, & Kirkpatrick, 2009).  Athletes who participate in high-risk sports might share a number of qualities, e.g., a passion for the outdoors (Brymer & Oades, 2009), nonchalance towards heights (or an attempt to conquer a fear (J.H. Kerr & Mackenzie, 2012)), and perhaps a need for thrill and excitement. Numerous studies have shown that high-risk sport practitioners report higher levels of sensation seeking than low-risk sport practitioners (Diehm & Armatas, 2004; I. H. Franken, Zijlstra, & Muris, 2006; Franques et al., 2003; Gelernter et al., 1997; Kelly et al., 2006; Michel, Cazenave, Delpouve, Purper-Ouakil, & LeScanff, 2009; Zuckerman & Kuhlman, 2000). A selection of studies is shown in Table 1-4, for a recent review see Goma-i-Freixanet et al., (2012). To my knowledge there are no studies reporting contrary findings. Studies that have looked at other measures related to disinhibition, such as behavioural activation measured using the BIS/BAS scale (C. S. Carver & White, 1994), and reward sensitivity measured using the Sensitivity to Punishment and Sensitivity to Reward Questionnaire (SPSRQ) (Torrubia et al., 2001) similarly find higher “fun seeking” in skydivers compared to rowers (I. H. Franken et al., 2006). Likewise, mountaineers and other high-risk sportsmen scored higher on measures relating to  21  reward sensitivity (e.g., sensitivity to reward, and thrill and adventure seeking) (Castanier, Le Scanff, & Woodman, 2010a, 2010b; Goma-I-Freixanet, 1991). Many of the high-risk sport studies that exist have looked at sports that carry a very high degree of risk that are practiced by a select group of individuals. For example, Michel and colleagues (2009) studied BASE jumping which is practiced by only a few thousand people world wide and for which many clubs require that jumpers have skydived over 250 times before admission (e.g., Hallin & Mykletun, 2006) Franques et al. (2003) studied paragliders, a sport that requires a license to fly solo, and practiced by less than 1% of people reporting participation in adventure recreation (Ewert & Hollenhorst, 1997). Other sports that carry some physical risk, and are therefore considered high-risk, but are commonly practiced by the general public have been less thoroughly explored. In particular, skiing and snowboarding are popular, mainstream sports that are considered high-risk activities (Malkin & Rabinowitz, 1998) with a risk for injury of approximately 2 to 4 per 1000 participant days and that accounts for 28% of all injuries related to ice and snow sports (Warda & Yanchar, 2012). Another common trend in high-risk sport literature is that many studies include largely male (or male-only) samples (as shown below in Table 1-4) since there are more males that participate in many of the very high-risk sports (Jensen & Guthrie, 2006). Choosing commonly practiced downhill sports, such as skiing and snowboarding, allows for the recruitment of large samples with almost an equal representation of females.  22  Table 1-4 Studies comparing sensation seeking between high- and low-risk athletes or controls Studies comparing sensation seeking between high- and low-risk athletes or controls High-risk (n) Downhill skiing (219), M + F Bungee jumping (80), M + F(29) “Extreme risk takers” (20) rock climbing, skiing, stunt flying, kayaking, males  Controls and low-risk (n) Matched reference (299), M + F General public, M + F(45) High risk-takers (20) (safer version of same sports), males Low-risk athletes (20), males Swimming, marathon runners, aerobics, golf (73), M + F(37) Parachuting (novice) (14), males  Tool SSS SSS V (Fr) SSS V  Findings Skiers scored higher on TAS Bungee jumpers scored higher on TAS TAS higher in both risk groups (high and extreme) compared to controls. No differences between risk groups  Ref. 1 2 3  SSS V  SS higher in high-risk group  4  SSS V (Nor)  5  Risky sports (52), females Paragliding (34), M + F Surfing (41), M + F(11) Recreational (53) and professional (37) risk sports athletes, females  Controls (58), females Matched reference group (34), M + F Golfing (44), M + F(15) Low-risk athletes (90), females  SSS V (Sp) SSS IV (Fr) SSS V SSS (Fr)  Rock climbers (118), M + F(28)  No control group, measured risky behaviours Controls (11), males  ImpSS  No controls group, inquired about risky behaviours Non-athletic controls (54), males  SSS V (D)  Both groups scored higher on SSS than controls, experts scored higher than novices on ES Risk group scored higher on TAS, ES, and Dis Paragliders scored higher on BS, TAS, and Dis Surfers scored higher on TAS and ES Recreational athletes scored higher on SSS (and TAS and ES) than both controls and pros, and pros scored higher than controls Risky behaviours were not associated with sensation seeking, impulsivity, or ImpSS BASE jumpers scored higher on SSS total and TAS SS score was associated with risky behaviours  Hand-gliding, mountaineering, skydiving, auto racing (93), M + F(10) Parachuting (experts) (21), males  BASE jumpers (11), males Skiers, M + F (683)  SSS IV (Fr)  6 7 8 9  10 11 12  Parachuting, wakeboarding, SSS IV (P) Risk group scored higher on all SSS subscales 13 snowboarding, scuba, alpinism, paragliding (217), males Note. 1(Bouter, Knipschild, Feij, & Volovics, 1988); 2(Michel, Carton, & Jouvent, 1997); 3 (Slanger & Rudestam, 1997); 4(Jack et al., 1998); 5(Breivik, Roth, & Jorgensen, 1998); 6(Goma-i-Freixanet, 2001); 7(Franques et al., 2003); 8(Diehm & Armatas, 2004); 9(Cazenave, Le Scanff, & Woodman, 2007); 10(D.J. Llewellyn & Sanchez, 2008); 11(Michel et al., 2009); 12(Ruedl et al., 2010); 13(Guszkowska & Boldak, 2010). M = male, F = female, Fr = French, Nor = Norwegian, Sp = Spanish, D = German, P = Polish.  23  It is important to note that although the risk sports mentioned above and in Appendix A carry a risk, most participants report that risk is not the goal of the activity (Brymer, 2010). High-risk athletes often describe themselves as self-aware, disciplined, controlled, and their sport offers them a state of relaxation and emotional clarity (Brymer, 2010). Athletes interviewed in Brymer’s (2010) study were not carelessly taking risks within these sports, and most of the above-listed activities require great skill and planning. While sensation seeking is the trait most commonly associated with participation in high-risk sports (reviewed in Goma-i-Freixanet et al., 2012), many participants report high self-efficacy (D.J. Llewellyn, Sanchez, Asghar, & Jones, 2008; Slanger & Rudestam, 1997) and low anxiety (Goma-I-Freixanet, 1991). Another trait that has been associated with prolonged participation in high-risk sport is alexithymia; studies have found that high-risk athletes have a difficulty in identifying and expressing emotions (Cazenave et al., 2007; Woodman, Hardy, Barlow, & Le Scanff, 2010; Woodman, Huggins, Le Scanff, & Cazenave, 2009), which perhaps explains why sport offers “emotional clarity” (described in Brymer, 2010) for some. While multiple traits likely work together to motivate high-risk sport participation, high sensation seeking and/or strong reward sensitivity are the most consistently reported. In order to explore whether genetic influences motivate sport participation, we must first understand the neurobiological pathways that underlie these approach traits.  1.5  Neurobiology of sensation seeking Researchers categorize individuals as high or low sensation seekers based on scores on  questionnaires, and variations in the functioning of our neurotransmitter systems putatively contribute to these individual differences. Cloninger (1987) hypothesized that three  24  neurotransmitters underlie broad temperaments and he created the Tri-dimensional Personality Questionnaire (TPQ, now modified and expanded to the Temperament and Character Inventory, TCI) under this premise. According to Cloninger (1987), dopamine, serotonin and norepinephrine underlie novelty seeking, harm avoidance, and reward dependence, respectively. Similarly, Zuckerman proposed that sensation seeking, which involves behavioural mechanisms of strong approach, weak avoidance, and weak arousal, is associated with strong dopamine reactivities, weak serotoninergic, and weak norepinephrine reactivities, respectively (Zuckerman, 2007a). Zuckerman (2007a) notes that with regards to reactivities of neurotransmitter systems, it is the sensitivity of the receptors in response to stimuli and not the levels of circulating neurotransmitters that is important. These three neurotransmitter systems interact, and similarly, behavioural mechanisms involving approach and avoidance interact (see Figure 1-1 for a simplified version of Zuckerman’s psychobiological model; Zuckerman, 1994). Zuckerman’s model is in line with the RST (section 1.3.1), which proposes that brain structures underlying BAS (approach) involve mesolimbic dopaminergic pathways, whereas the structures underlying BIS (avoidance) involve serotonergic pathways arising from the septo-hippocamppal system (C. S. Carver, 2004; Gray & McNaugton, 2000; Zuckerman, 2005a).  25  Figure 1-1. Simplified version of Zuckerman’s psychobiological for impulsive sensation seeking (ImpSS). Model shows behavioural mechanisms (approach and inhibition) that underlie ImpSS, their interactions, and how the neurotransmitters interact to influence the mechanisms. + represents an excitatory pathway and – represents an inhibitory pathway (Adapted from Figure 14.2; Zuckerman, 1994).  Prior to his psychobiological model of ImpSS, Zuckerman proposed a model for sensation seeking that was based on optimal levels of catecholamine system activity (CSA) (Zuckerman, 2007a). The CSA model suggested that under basal arousal conditions, high sensation seekers have low levels of dopamine and norepinephrine activity and are therefore are under-aroused at rest with a susceptibility to boredom. In basal conditions low sensation seekers maintain optimal levels of catecholamine activity but additional stimulation may produce adverse effects, whereas an identical level of stimulation in high sensation seekers might be optimal (Zuckerman, 2007a). Many researchers still refer to optimal arousal and/or catecholamine system activity models when disseminating biological and genetic findings relating to sensation seeking and anhedonia5 (I. H. Franken et al., 2006). The arousal model is also called the “inverted U-shape” (e.g., Cross et al., 2011), in which an increase in the hedonic  5  Anhedonia refers to the inability to experience pleasure (or hedonic feelings). 26  valence of an activity (“pleasant” vs. “unpleasant”) is associated with increasing arousal/stimulation up until a maximum point, after which increased arousal becomes aversive and reduces the enjoyment of the activity. Individuals differ in their starting point along the arousal axis (see Figure 1-2).  Figure 1-2. Inverted U-shape model of arousal. As arousal increases, enjoyment of an activity will increase up until a maximum point at which any increase in arousal may induce anxiety and/or be viewed as unpleasant. Individuals purportedly differ in their optimal levels of arousal.  While a complex network of neurotransmitters appear to underlie mechanisms for approach and avoidance, a prominent role for dopamine in sensation seeking is widely supported by neuro-imaging studies, animal models, and genetics (Ebstein, 2006; Ebstein & Israel, 2009; reviewed in Hariri, 2009). All drugs of abuse increase extracellular levels of dopamine, particularly in the mesocorticolimbic system (Koob, Sanna, & Bloom, 1998), and natural rewards like sex and food stimulate the release of extracellular dopamine (Hernandez & Hoebel, 1988).  27  Dopamine has been most widely cited for its involvement in “reward processes”, but it is much more complex in reality and participates in a wide range of cognitive and sensorimotor functions (Salamone, Correa, Farrar, & Mingote, 2007; Salamone, Correa, Mingote, & Weber, 2005). Dopamine blockade has been shown to impair response habits in mice (reviewed in Wise, 2004), supporting a role in learning. Agonists for dopamine are sometimes used to treat ADHD, suggesting a role for dopamine in sustained attention (Avale et al., 2004). Animal models provide additional support for dopamine’s role in memory encoding and learning; for example: in mice, dopaminergic neurons are activated when an unexpected reward is presented (Pecina, Cagniard, Berridge, Aldridge, & Zhuang, 2003) and in rats, exposure to novel environments elicit increases in dopamine release (Rebec, Christensen, Guerra, & Bardo, 1997). It is important to note that the reward-related dopamine release is not associated with an increase in the “liking” of the reward, but rather an increase in the “wanting” of the reward (Pecina et al., 2003); a finding that supports a model in which dopamine is important for modulating functions related to motivation towards goal-directed behaviours, but not the consummatory reward enjoyment itself (Wise, 2004).  1.5.1  Dopamine synthesis and transport Dopamine is produced in the cell bodies in dopaminergic neurons and then stored in  vesicles at the axonal termini until signalled for release into the synapse. Once released, dopamine either binds to G-protein coupled dopaminergic receptors on the post-synaptic cell, or to autoreceptors on the pre-synaptic neuron (Zuckerman, 2005a). Dopaminergic neurons fire tonically releasing a small amount of dopamine into the synaptic cleft, or phasically (when activated by excitatory stimuli) due to a burst of firing resulting in a spike-dependent release of a  28  large amount of dopamine into the cleft (Grace, 2000). Reward-related mechanisms have been linked to the phasic release of dopamine, but dopaminergic tone can influence the phasic release via negative feedback (i.e., autoreceptors on pre-synpatic neuron are inhibitory) (Grace, 2000). Currently, there are five known types of dopamine receptors, and the genes that encode them are named DRD16 to DRD5. The D2-like receptors (which include D2, D3 and D4) behave both as standard post-synaptic receptors projecting a signal to neighbouring neurons, and as autoreceptors (D2 and D3) sending negative feedback signals to decrease dopamine release. In general, D2-like receptors are considered inhibitory, decreasing the downstream release of cyclic AMP (cAMP); whereas D1-like receptors (D1 and D5) are considered excitatory, increasing cAMP. D1-like receptors exclusively act post-synaptically on target cells, whereas D2-like receptors are expressed both post- and pre-synaptically (autoreceptors) (reviewed by Beaulieu, 2010). Once released, the dopamine transporter (DAT1) recycles dopamine back into the presynaptic cell (Hong, Cheng, Shu, Yang, & Tsai, 2003); therefore the combined effects of the transporters and receptors regulate the amount of dopamine in the synapse (see Figure 1-3).  6  By convention, human gene names are capitalised and italicized e.g., DRD1 (a gene) encodes the D1-receptor (a protein). 29  Figure 1-3. Dopaminergic neuron. Tyrosine is converted to TH (tyrosine hydroxylase), which is converted to L-DOPA (dihydroxphenylalanine) and finally to dopamine (DA). Upon arrival of an action potential, dopamine is released from the pre-synaptic cell, and may bind to five receptor types: D1-like excitatory (D1,D5) or D2-like inhibitory (D2,D3,D4). Alternatively, dopamine may be metabolized by MAO (monoamine oxidase) or COMT (catechol-o-methyltransferase) into HVA (homovillac acid), or recycled back into the neuron by DAT1 (dopamine tansporter). D2 and D3 act as both pre and post-synaptic receptors. Dopamine may also be converted to norepinephrine, by DBH (dopamine-!-hydroxylase). (adapted from Nernoda, Szekely, & Sasvari-Szekely, 2011; adapted from Tavernarakis website "Sensory transduction and integration," n.d.).  30  1.5.2  Dopamine and sensation seeking Dopamine participates in a wide range of functions, but there is support for its specific  involvement in novelty or sensation seeking. Dopamine agonists used in the treatment of Parkinson’s disease increased reward processing and novelty seeking in young patients (Bodi et al., 2009). Marginal positive correlations have been observed between novelty seeking and dopamine receptor (D2/D3) availability (H. Y. Huang et al., 2010). These results are in accordance with another study that found high novelty seeking scores were associated with lower midbrain D2-like receptor availability (Zald et al., 2008). A decrease in autoreceptor density purportedly contributes to vulnerability for disinhibited behaviours. This vulnerability hypothesis has been supported by multiple studies including positron emission tomography (PET), which showed that trait impulsivity is positively correlated with amphetamine-induced dopaminergic release and negatively correlated with D2/D3 autoreceptor availability. In other words, reduced autoreceptor availability results in less inhibition of dopaminergic firing and release, thus resulting in increased neuronal firing and dopaminergic release in response to novel/rewarding stimuli (Buckholtz et al., 2010). Low D2 receptor availability has also been observed in susceptibility to alcoholism based on the finding that non-alcoholic members of families with alcoholic subjects had higher D2 receptor densities in caudate and ventral striatum (Volkow et al., 2006). Imaging techniques further support a role for dopamine in sensation seeking, specifically, finding positive correlations between activation in dopaminergic brain regions (e.g., nucleus accumbens) and thrill and adventure seeking (from SSS) and exploratory excitability (NS subscale) (Abler, Walter, Erk, Kammerer, & Spitzer, 2006). Data from knockout models in mice also support involvement of dopamine receptors and transporters in behavioural approach. For example, mice lacking D2 autoreceptors show  31  increases in dopamine release upon stimulation, are hyperactive, hypersensitive to the effects of cocaine, and showed enhanced motivation towards food reward (Bello et al., 2011). Similarly, Drd47- and Drd3-knockout mice are hyperactive, hypersensitive to ethanol and amphetamines, and display exploratory behaviours (Accili et al., 1996; Rubinstein et al., 1997). A link between D4 and ADHD was demonstrated by knocking out the Drd4 in mice that had been lesioned with 6-hydroxydopamine (a compound shown to alter central dopaminergic pathways) to exhibit hyperactive (or ADHD-like) symptoms. Mice lacking Drd4 did not exhibit the hyperactive behaviours typical of the lesioned model, suggesting the involvement of D4 in hyperactivity and behavioural inhibition (Avale et al., 2004). Dopamine neurotransmission is affected not only by receptors but also by the dopamine transporter. Dat1 knockout mice exhibited behaviours analogous to symptoms of ADHD compared to wild-types (Giros, Jaber, Jones, Wightman, & Caron, 1996). While the involvement of dopamine in ADHD does not directly relate to sensation seeking, people high in this trait display impulsive, excitable, and exploratory tendencies, some of which are features of ADHD (Faraone et al., 1999; Zuckerman, 2007a). In summary, manipulating dopaminergic systems in the brains of mice and humans have resulted in altered reward-seeking and exploratory behaviours. These findings provide support for dopamine as a candidate pathway for sensation-seeking behaviours.  1.6  Sensation-seeking genetics: A review of “candidate genes” Sensation seeking is moderately heritable, with approximately 60% of trait variation due  to genetic factors (Hur & Bouchard, 1997; Koopmans, Boomsma, Heath, & van Doornen, 1995; Stoel, De Geus, & Boomsma, 2006). Many genetic association studies attempt to identify the  7  By convention, mouse genes are italicized with the first letter capitalised. 32  genetic variants that underlie the novelty-seeking and sensation-seeking traits, and genes that encode proteins involved in each stage of dopamine and dopamine-related pathways (including synthesis, transport and metabolism) are potential candidates for such studies. Serotonin is a neurotransmitter most commonly cited for its involvement in anxiety-related disorders and avoidance-related traits (reviewed in Hariri, 2009; M. R. Munafo et al., 2003), but has also been implicated in impulsivity (C.S. Carver & Miller, 2006). Weak avoidance and strong approach are the motivational tendencies that are thought to underlie sensation seeking (see sections 1.3.3 and 1.5); therefore, serotonin, which has been implicated in avoidance-related traits, may be another potential candidate system for sensation seeking.  1.6.1  Dopaminergic pathway genes As discussed earlier, imaging, pharmaceutical manipulations in humans and animals, and  animal knockout models strongly support a role for the dopamine receptors’ involvement in approach-related processes, and therefore dopamine receptors genes are potential candidates for novelty- and sensation-seeking genetic studies. Polymorphisms within dopamine receptors have been investigated for associations with approach traits (e.g., sensation seeking, novelty seeking, extraversion), externalizing behavioural disorders (e.g., ADHD, conduct disorder (CD), operational defiant disorder (ODD)), and substance use (e.g., alcohol, amphetamine, nicotine) (see Appendix E for a list of polymorphisms and phenotypes). The D2-like receptors are more commonly studied than D1-like receptors, in particular, the DRD2 and DRD4 genes are the most frequently studied candidates for approach traits and/or disinhibited behaviours (M. R. Munafo et al., 2003).  33  1.6.1.1  The DRD4 gene DRD4 is located on chromosome 11, at 11p15.5. The promoter region of the DRD4 is  highly polymorphic (Okuyama et al., 2000) and multiple variants of the gene have been associated with novelty seeking, extraversion, schizophrenia, and externalizing disorders (Lai et al., 2010; Mitsuyasu et al., 2001; reviewed by M. R. Munafo et al., 2003; reviewed by M. R. Munafo et al., 2008). The D4 receptor is found in the entorhinal and prefrontal cortex, the dorsomedial thalamus, and parts of the limbic system including the lateral septal nucleus, the hypothalamus, and the hippocampus (Primus et al., 1997), all of which are regions of the brain thought to be involved in emotional regulation, attention, and motivation (Kreek, Nielsen, Butelman, & LaForge, 2005). Two hallmark studies in 1996 reporting associations between alleles of a 48-bp variable number tandem repeat (VNTR) polymorphism in exon III and novelty seeking (Benjamin et al., 1996; Ebstein et al., 1996) spurred a flurry of studies into the genetic underpinnings of personality traits in healthy populations (reviewed by Oak, Oldenhof, & Van Tol, 2000; Paterson, Sunohara, & Kennedy, 1999). Both studies found that individuals carrying at least one copy of the 7-repeat (7R) allele reported higher novelty seeking scores than those carrying the more common 4-repeat allele. Although the two most common alleles in Caucasian populations are the 4- and 7-repeat, allelic versions between 2 and 11 repeats have been reported, and the frequencies vary greatly between populations (Ding et al., 2002). High novelty seeking in carriers of the long (7R) allele have since been inconsistently replicated (reviewed by M. R. Munafo et al., 2008; Oak et al., 2000; Paterson et al., 1999), though less ambiguous associations with externalizing disorders and the 7R allele have been observed (e.g., ADHD; Gizer, Ficks, & Waldman, 2009). Other variants in DRD4 have similarly been explored in association with  34  approach-related traits and externalizing disorders, notably a single nucleotide in the promoter region, -521 C/T (db SNP rs1800955) and a 120-bp tandem duplication (db SNP rs4646984) (Nernoda et al., 2011). Purported functional differences between alleles at all three of these loci have been reported and will be discussed in detail in the chapters that follow (and shown in Table 2-1).  1.6.1.2  The DRD2 gene DRD2 is located on a different arm of chromosome 11, at 11q23.1. The TaqIA  polymorphism (dbSNP rs1800497), also located near DRD2, has been associated with rewardseeking behaviours (including substance use, binge eating, gambling, and sensation seeking) (Blum, Sheridan, et al., 1996) and extraversion (L.D. Smillie, Cooper, Proitsi, Powell, & Pickering, 2010). Although the findings are not ubiquitous, a meta-analysis found a significant association with alcoholism (M. R. Munafo, Matheson, & Flint, 2007). The TaqIA (resulting in a thymine to cytosine transition, referred to as A1 and A2 alleles, respectively) is located near the termination codon of the DRD2 gene, but the polymorphism was later mapped to lie within a downstream neighbouring gene called ankyrin-repeat and kinase-domain-containing-1 gene (ANKKI), and it is still sometimes referred to as a DRD2 variant (Ponce et al., 2009). The A1 allele exhibits reduced expression and brain autopsies revealed that individuals carrying this allele have 30% fewer D2 receptors than those carrying the A2 allele (Noble, Blum, Ritchie, Montgomery, & Sheridan, 1991). PET scans on healthy participants provide additional support for reduced striatal density of D2 receptors in the presence of the A1 allele (Jonsson et al., 1999). Blum and colleagues (1996) proposed the “reward deficiency theory” as the mechanism underlying reward/stimulus-seeking behaviours associated with DRD2. Specifically, individuals  35  with fewer D2 receptors would exhibit lower dopaminergic activity in the reward areas of the brain. In theory, A1 carriers would experience less reward in response to a stimulus than A2 carries, leading the A1 carriers to seek more stimuli (i.e., stimulus seeking) (Blum, Sheridan, et al., 1996). A handful of other functional SNPs within the DRD2 have been identified. Intronic SNPs, A/C rs2283265 and G/T rs1076560 (also known as Taq1B), were associated with decreased expression and increased striatal activity and poor performance during attentional tasks (Y. Zhang et al., 2007) and both intronic SNPs have been implicated in amphetamine abuse (Moyer et al., 2011).  1.6.1.3  Other dopamine receptors D3 and D1-like receptors (D1, D5) are less commonly studied in association with  approach traits and disnhibited behaviours. DRD3 maps to chromosome 3q13.31. The most commonly studied variant in DRD3 is a functional polymorphism in exon I, Ser9Gly (312 C/T, dbSNP rs6280) which has been investigated in association with a wide range of phenotypes, from migraine-risk to personality, addiction, schizophrenia, and ADHD (migraine: GarciaMartin et al., 2010; ADHD: Gizer et al., 2009; smoking: Novak et al., 2010; schizophrenia: F. Zhang et al., 2011). A meta-analysis found that the association between the Ser9Gly polymorphism and schizophrenia was not significant (Jonsson, Kaiser, Brockmoller, Nimgaonkar, & Crocq, 2004), and although the SNP has been associated with sensation seeking and novelty seeking (Duaux et al., 1998; Staner et al., 1998), there are several personality studies reporting non-significant findings (Jonsson et al., 2003; Schosser et al., 2010). More recently a handful of studies have found associations between an intronic DRD3 SNP (dbSNP rs1677771) and various psychiatric phenotypes related to autism spectrum disorder (de Krom et al., 2009;  36  Staal, de Krom, & de Jonge, 2012). Variants in DRD3 and relevant phenotypes will be discussed in Chapter 5. D1 receptors are the primary target for dopamine in the pre-frontal cortex and have been shown to be involved in working memory and attention based on mouse models (Rinaldi, Mandillo, Oliverio, & Mele, 2007). DRD1 maps to chromosome 5q35.2 and is intronless. Two SNPs, the DdeI RFLP8 (-48 A/G; dbSNP rs4532) and 1403 T/C (dbSNP rs686), in the DRD1 have been associated with schizophrenia (Zhu et al., 2011), alcoholism (Batel et al., 2008), nicotine dependence (H. Y. Huang et al., 2010), and addictive behaviours (Comings et al., 1997; Liu, Chen, Leu, Wu, & Lin, 2006). Though there are fewer studies that have explored DRD1 variants in association with personality traits, alleles at the DdeI RFLP have been associated with sensation seeking in alcoholic males (Limosin, Loze, Rouillon, Ades, & Gorwood, 2003).  1.6.1.4  The dopamine transporter The dopamine transporter gene (DAT1), sometimes referred to as solute carrier family 6  (SLC6A3), has also been investigated as a candidate gene in studies of approach-related traits and ADHD. Stimulant medication widely prescribed to treat ADHD directly inhibits the action of the dopamine transporter, and is effective in reducing ADHD symptoms, suggesting a role for DAT1 in the aetiology of ADHD (reviewed by Turic, Swanson, & Sonuga-Barke, 2010). Of greater relevance to my research, the DAT1 has also been linked to risk-taking (Mata, Hau, Papassotiropoulos, & Hertwig, 2012) as measured using the Balloon Analogue Risk Task (see section 1.3.3). DAT1 is located on chromosome 5p15, and codes for the protein responsible for recycling dopamine back to the pre-synaptic neuron (see Figure 1-3). The most commonly  8  RFLP stands for restriction fragment length polymorphism. Polymorphisms are sometimes named after the restriction enzymes used in genotyping. 37  studied variant in DAT1 is a 40-bp VNTR (3 to 13 repeats) in the 3’ untranslated region (UTR) of the gene; however, the results are not consistent (Nernoda et al., 2011). Both the 9-repeat and the 10-repeat alleles at the VNTR have been associated with drug and alcohol use, ADHD, schizoid-avoidant behaviours, and other personality traits (Blum et al., 1997; Comings et al., 1996; Ujike et al., 2003), but there have been numerous null findings (Hong et al., 2003; Hou & Li, 2009; Jorm et al., 2000; D. W. Li, Sham, Owen, & He, 2006). Functional studies have shown reduced expression of DAT1 in carriers of the 9-repeat allele (versus the more common 10repeat) (VanNess, Owens, & Kilts, 2005). While most studies have focused on the DAT1 UTR VNTR, there are a few notable SNPs that have been associated with approach-related traits and externalizing disorders, including rs6347 and rs27072 (Feng et al., 2003; Ouellet-Morin et al., 2008).  1.6.1.5  Dopamine metabolism Dopamine is either recycled into the pre-synaptic cell by the dopamine transporter, or is  hydrolyzed or metabolized by various enzymes (catechol-0-methyltransferase, monoamineoxidase, or dopamine-!-hydroxylase; Figure 1-3). Changes in these enzyme levels can affect the amount of dopamine in the cell, and therefore enzymes in the dopamine pathway are candidates for personality and behavioural studies. Dopamine-!-hydroxylase (encoded by DBH) is an enzyme that hydrolyses dopamine to norepinephrine (Kamata et al., 2009). Dbh knockout mice and patients with a rare null DBH allele have higher levels of dopamine and its metabolite, 3,4Dihydroxyphenylacetic acid (DOPAC) (reviewed in Kamata et al., 2009). High sensation seeking has been related to low levels of dopamine-!-hydroxylase (reviewed in Zuckerman, 1994). A functional SNP in the DBH promoter region, -970 C/T (formerly called -1021 C/T,  38  dbSNP rs1611115) that results in a cytosine to thymidine substitution, may explain some of the variation that exists in dopamine-!-hydroxylase activity. Carrying the T allele results in reduced expression, and therefore a lower level of the enzyme and reduced dopamine-!-hydroxylase activity (Zabetian et al., 2001). The -970 C/T SNP has been associated with harm avoidance in healthy females (Kamata et al., 2009) a trait that has been negatively correlated with sensation seeking (McCourt et al., 1993). This SNP has also been associated with alcohol dependence and withdrawal (reviewed in Koehnke, 2008). A non-synonymous SNP located in exon 11 of DBH, 16549 C/T (Arg549Cys, dbSNP rs6271) and an intronic SNP, intron 5 TaqI (dbSNP rs2519152) also appear to influence dopamine-!-hydroxylase enzyme activity (Tang et al., 2006; Zabetian et al., 2001). The two aforementioned SNPs (rs6271, rs161115) in combination with intronic SNP rs1611122 significantly contributed to the linkage signal observed in a study of schizophrenics (Cubells et al., 2011). DBH has also been studied in association with other externalizing disorders such as ADHD. Strong correlations between enzyme activity and rs161115 and rs2519152 were observed in both ADHD cases and matched controls, and there was over – transmission of the rs2519152 G allele to ADHD probands (Bhaduri, Sarkar, Sinha, Chattopadhyay, & Mukhopadhyay, 2010). Catechol-O-methyltransferase (encoded by the gene COMT) is an enzyme that metabolizes dopamine in the synaptic cleft (Reuter & Hennig, 2005). COMT genetic association studies have focused on a functional variation in the gene, where a G to A (at base 472) transition causes a valine to methionine substitution at amino acid 158 (Val158Met, dbSNP rs4680). The Met158 allele has been linked to lower catechol-O-methyltransferase enzyme activity than the Val158 allele, and is often called ‘COMT L’ (low activity) (Lotta et al., 1995).  9  1654 C/T is sometimes called 1603 C/T (Arg503Cys). 39  Similarly, in post-mortem brain tissue, the Va1158 variant has been shown to exhibit approximately 38% higher activity than the Met158 variant (Chen et al., 2004). In theory, a higher activity allele would result in decreased synaptic dopamine levels, and because catecholO-methyltransferase is highly expressed in the pre-frontal cortex, this may contribute to impaired cortical function (Chen et al., 2004). The COMT Val158Met polymorphism has been extensively studied in association with neuropsychiatric disorders and traits in healthy populations and there have been many conflicting findings (Lachman, 2008). The COMT Val158 allele has been associated with lower responses to reward measured by either surveys or fMRI techniques (Ettinger et al., 2008; Wichers et al., 2008). These results support the “reward deficiency” hypothesis for dopamine (discussed briefly in section 1.6.1.2), suggesting that Val158 homozygotes (who presumably have lower synaptic dopamine) might require additional stimulation to attain reward from everyday pleasures (Lachman, 2008). Differences in fear processing measured using the startle reflex have also been observed, and Val158 carriers showed a decreased startle reflex compared to Met158 homozygotes (Montag et al., 2008), and similarly Val158 carriers had lower sensitivity (measured by event-related potentials) to aversive stimuli than Met158 carriers (Herrmann et al., 2009). The high-activity Val158 allele has also been associated with sensation seeking and reward dependence in females (U. E. Lang, Bajbouj, Sander, & Gallinat, 2007; Tsai, Hong, Yu, & Chen, 2004) and extraversion in males and females (Reuter & Hennig, 2005). Conflicting findings associating the Met158 allele with approachrelated traits and disinhibited behaviours have also been reported (e.g., risky sex: Bousman et al., 2010; novelty seeking in methamphetamine users: Hosak, Libiger, Cizek, Beranek, & Cermakova, 2006; cocaine abuse: Lohoff et al., 2008); therefore there is no clear consensus on which allele confers a susceptibility to reward seeking (Lachman, 2008; Nernoda et al., 2011).  40  Bilder (2004) reconciles the reported inconsistencies by considering both tonic and phasic release of dopamine (one allele affecting each type of dopaminergic release); however, distinguishing which allele is associated with a given phenotype is difficult because true replication studies employing identical phenotype characterization are rare. Two other COMT SNPs, a synonymous coding SNP (C/T rs4633) and promoter SNP (A/G rs6269) have also been studied in association with approach-related phenotypes (e.g., Choudhry et al., 2012; Halleland, Lundervold, Halmoy, Haavik, & Johansson, 2009; Roe et al., 2009), and a four-SNP-haplotype including the three above-mentioned SNPs along with synonymous coding SNP (C/G; rs4818) together alter the secondary mRNA structure and affect protein expression (Nackley et al., 2006). Monoamine-oxidase types A and B (encoded by the genes MAO-A and MAO-B) are also enzymes that catabolize endogenous monoamines (including neurotransmitters), but monoamineoxidase B has been the focus of many personality associations since its preferred substrate is dopamine (Oreland, 2004). The enzyme is found in blood platelets, and low levels of platelet monoamine-oxidase B have been correlated with high sensation seeking (reviewed in Zuckerman, 1994). Production of this enzyme increases with age, and is increased in females compared to males (N. L. Pedersen, Oreland, Reynolds, & McClearn, 1993; Zuckerman, 2005b). Theoretically, dopamine levels would decrease with age and be lower in females - a similar trend to that seen in sensation-seeking scores, making MAO-B a potential candidate for sensationseeking genetics research (Zuckerman & Kuhlman, 2000). Mao-B knockouts (mice) display reduced habituation to novel environments (based on locomotor activity) (M. Lee, Chen, Shih, & Hiroi, 2004), and therefore lower levels of monoamine-oxidase (and theoretically higher levels of dopamine) may influence novelty/sensation seeking (M.T. Bardo et al., 2007). Monoamineoxidase activity has a high heritability (h2 = 0.76) (N. L. Pedersen et al., 1993) and has been  41  linked to variations in the MAO-B gene, located on the X chromosome (Garpenstrand, Ekblom, Forslund, Rylander, & Oreland, 2000). The intron 13 A/G SNP (rs1799836) is the most commonly studied variant, and has been studied in association with MAO enzyme levels, but the results are mixed (Garpenstrand et al., 2000; Pivac et al., 2006).  1.6.2  Serotonergic pathway genes The neurotransmitters do not act in isolation, and Cloninger’s (1987) model is somewhat  oversimplified in that novelty seeking may not be influenced by dopamine alone. The serotonin (also called 5-hydroxytryptamine (5-HT)) and dopamine systems interact, exerting regulatory control over each other (Malmberg, Wargelius, Lichtenstein, Oreland, & Larsson, 2008; Zuckerman, 2005a). As previously described in section 1.5, dopamine is involved in mediating “approach” behaviours and serotonin, in mediating “avoidance”. Rats with low levels of serotonin are more impulsive and have reduced harm avoidance (Winstanley, Dailey, Theobald, & Robbins, 2004). Other animal (monkey) studies support a role for serotonin in impulsive behaviours and risk-taking (Fairbanks, Melega, Jorgensen, Kaplan, & McGuire, 2001; Long, Kuhn, & Platt, 2009), though most personality genetic studies that investigate serotonergic genes focus on internalizing disorders (e.g., anxiety and depression) (reviewed in Hariri, 2009).  1.6.2.1  Serotonin transporter The serotonin transporter gene (SLC6A4, located at 17q.11) is involved in regulating the  magnitude and duration of serotonin action in serotonergic neurons (Lesch et al., 1996), and it is the most extensively studied gene in the field of personality research (M. R. Munafo et al., 2003; M.R. Munafo et al., 2009). A repeat polymorphism in the regulatory region of SLC6A4, the 5-  42  HTTLPR, has been associated with a range of conditions and traits (see Appendix E), but it is most commonly linked to avoidance traits (M. R. Munafo et al., 2003; M.R. Munafo et al., 2009). Lesch and colleagues’ (1996) study demonstrated functional differences in the transcriptional efficiency, the “S” (short, 14 repeats) allele showing reduced activity and acting in a dominant-recessive fashion (replicated by Bradley, Dodelzon, Sandhu, & Philibert, 2005). This study also linked the 5-HTTLPR S allele to harm avoidance, neuroticism, and anxiety measured using three different personality scales (Lesch et al., 1996). More recently, however, a meta-analysis indicated that the 5-HTTLPR polymorphism was not associated with harm avoidance, and while there was support for an association with one measure of neuroticism (from Costa & McCrae’s 1997 five-factor model of personality), there was no association with neuroticism defined by a different instrument (Eysenck’s Personality Questionnaire) (M.R. Munafo et al., 2009). Some externalizing disorders and approach-related traits (especially impulsivity) have been studied in association with the 5-HTTLPR variant (Aluja, Garcia, Blanch, De Lorenzo, & Fibla, 2009). A few studies included sensation seeking, and one found that the S allele was associated with items from the ZKPQ sociability subscale and there was a trend for higher ZKPQ ImpSS in a sample of borderline personality disorder patients (Pascual et al., 2007), though others report no association (Patkar et al., 2002).  1.6.2.2  Serotonin receptors Similar to dopamine receptors, the serotonin receptors are G-protein coupled receptors  (with the exception of subtype 3) that have either inhibitory or excitatory effects via the second messenger, cAMP, and as many as 14 receptor subtypes have been identified (Nichols & Nichols, 2008). Several variants in 5-HT receptor 2A located on chromosome 13 (HTR2A) have  43  been studied in association with approach-phenotypes, including -1438 A/G (dbSNP rs6311), -783 A/G (dbSNP rs6312), 1354 C/T (His452Tyr: dbSNP rs6314) and 102 T/C (Ser34Ser: dbSNP rs6313), though the findings have been inconsistent (reviewed by Gizer et al., 2009; Nomura et al., 2006). A functional SNP in the gene that encodes 5-HT receptor subtype 1A (HTR1A), the -1019 C/T (dbSNP rs6295) has been associated with internalizing psychiatric disorders (Lemonde et al., 2003), but also with impulsivity (Benko et al., 2010). While serotonergic and dopaminergic systems interact (see Figure 1-1), and serotonin has been implicated in impulsivity and aggression, there is less support for an independent role in sensation seeking. Studies that have found significant associations with approach-related traits have mostly looked at interactions between genes encoding serotonin and dopamine receptors and transporters (see section 1.6.4).  1.6.3  Other potential candidate genes for sensation seeking Neural plasticity is the capacity of the brain to remodel networks in response to  environmental changes, and neuronal growth-associated proteins regulate these changes. One particular protein, stathmin, plays a role in neural plasticity (Ehlis et al., 2011). Stathmin (encoded by the gene STMN1) is of potential interest in the study of high-risk sports and sensation seeking because knockout mice were shown to be “fearless” (Shumyatsky et al., 2005). Although Zuckerman suggests that sensation seeking is not related to fearfulness, fear accompanies many high-risk sport activities (e.g., skydiving), and a person experiencing more pleasure and less fear and anxiety, may be more likely to repeat the activity (Zuckerman, 2007c). Many risk sports involve heights (i.e., mountain and gravity sports), an innate fear in humans, and individuals with reduced anxiety associated with such innate fears might be more likely to  44  participate in such activities. Fearlessness in mice was measured by a task involving a raised platform to test for fear of heights, and mice lacking the gene encoding stathmin spent more time on the raised platform (Shumyatsky et al., 2005). A role for stathmin in fear processes in human was supported by a study finding associations between a tag10 SNP (dbSNP rs182455) and startle responses and cortisol release following a stressor (Brocke et al., 2010). The same SNP was associated with errors on the Stroop test (a measure of behavioural inhibition) (Ehlis et al., 2011). Another protein implicated in neuronal plasticity is the brain-derived neurotrophic factor (encoded by BDNF located on chromosome 11). Brain-derived neurotrophic factor is involved in memory and learning (reviewed by Tyler, Alonso, Bramham, & Pozzo-Miller, 2002). In mice, Bdnf knockouts showed abnormal Drd3 expression, suggesting an involvement for brain-derived neurotrophic factor in controlling DRD3 expression (Guillin et al., 2003). A non-synonymous variant at codon 66 in BDNF (Val66Met, dbSNP rs6265) has been studied in association with anxiety disorders and neuroticism. A meta-analysis found that individuals carrying at least one Met66 allele report lower neuroticism scores (Frustaci, Pozzi, Gianfagna, Manzoli, & Boccia, 2008). BDNF has also been proposed as a candidate for sensation seeking (Kang, Song, Namkoong, & Kim, 2010), and a genome-wide association study found an association between Val66Met and the approach-related trait extraversion (Terracciano, Sanna, et al., 2010). This was replicated in a study finding that BDNF Met66 homozygotes scored lower than Val66 carriers on extraversion (Terracciano, Tanaka, et al., 2010).  10  Tag SNPs are SNPs that are chosen to be representative of a block of DNA because the surrounding polymorphisms pass from one generation to the next with little “re-shuffling” between generations. 45  1.6.4  Interactions between candidate genes Sensation seeking is a complex trait likely influenced by many genes, with each  polymorphism explaining only a small proportion of phenotypic variance (Ebstein, 2006; Nernoda et al., 2011). As discussed in section 1.5, Zuckerman’s psychobiological model of sensation seeking suggests that there are interactions between dopamine and serotonin systems. Similarly, data from candidate gene association studies support intergenic interactions; for example, interactions between three major “candidate polymorphisms” (DRD4 VNTR, SLC6A4 5-HTTLPR, COMT Val158Met) and novelty seeking have been reported (Benjamin et al., 2000; Strobel, Lesch, Jatzke, Paetzold, & Brocke, 2003). In the Benjamin et al. (2000) study, novelty seeking and harm avoidance were negatively correlated and there were interactions between the genes associated with each trait. Furthermore, one study on financial risk-taking and another study measuring harm avoidance both found similar interactions between the DRD4 VNTR and the SLC6A4 5-HTTLPR (Kuhnen & Chiao, 2009; Szekely et al., 2004). Intergenic interactions have also been observed between other genes encoding proteins within the dopamine pathway, including enzymes, transporters, and neurotrophic factors. Yacubian and colleagues (2007) observed an interaction between polymorphisms in genes encoding proteins involved in dopamine re-uptake (DAT1) or catabolism (COMT) and striatal activity (measured using fMRI) during a guessing task. The study provided support for COMT’s involvement in regulating basal (or tonic) secretions of dopamine, and DAT1’s involvement in regulating phasic dopamine secretions (released in response to drugs, novelty, reward) (Bilder et al., 2004); suggesting that both genes are involved in inter-individual variations in sensitivity to reward. COMT has also been explored in association with BDNF and DRD4. An interaction between coding SNPs in BDNF and COMT was significantly associated with the boredom  46  susceptibility subscale of the SSS (Kang et al., 2010), and Li and colleagues (2004) found an interaction between polymorphisms in DRD4 (120-bp duplication) and COMT (Val158Met) in methamphetamine users. Finally, the two most commonly studied candidate genes for ADHD are DRD4 and DAT1, and a recent meta-analysis reported an interaction between the DRD4 120bp duplication and two downstream variants in DAT1 (3’ UTR VNTR and intron 8 VNTR) (Sanchez-Mora et al., 2011). Significant gene-gene interactions illustrate the complexity of the genetic background and suggest a role for non-additive or epistatic contributions to approach phenotypes (Ebstein, 2006; Ebstein & Israel, 2009). While the presence of these complex molecular interactions might explain some inconsistencies plaguing single gene association studies, few interactions have been convincingly replicated and many studies lack a priori functional hypotheses linking the polygenic loci.  1.7  Genetics of risk-inclined behaviours There have been numerous genetic studies on other risk-inclined populations including  substance users, alcoholics, and gamblers (reviewed in Dick, Prescott, & McGue, 2009; Goodman, 2008; and in section 1.6), but to my knowledge there has only been one genetic association study on personality traits in high-risk sport populations (Cam et al., 2010). The physiological mechanisms that underlie the motivation to participate in antisocial pastimes may be similar to those that attract people to high-risk sports (Zuckerman, 1983), but high-risk sports participants may represent true “arousal seekers” (Fjell et al., 2007), potentially unconfounded by other variables like impulsivity. Cam and colleagues (2010) compared genotype frequencies at three loci (DRD4 VNTR, DAT1 VNTR, and HTR2A 102 T/C) between university students who reported participation in high-risk (n = 60) and low-risk (n = 133) sports, and they also  47  compared personality scores obtained from the Five Factor Personality Inventory between genotype groups. There were no significant differences in genotype frequencies between the sport groups and although they observed small differences between HTR2A genotype groups on sub-dimensions of neuroticism and “openness to experience” the association would not have survived correction for multiple tests. Other than the Cam et al. (2010) study, researchers have yet to examine polymorphisms in populations actively involved in seeking out, and experiencing physical risks in sports, and no studies have characterized a phenotype by quantifying patterns of sport-specific behaviours.  48  Chapter 2: Research overview Chapter 1 defined the importance of sensation seeking within the context of high-risk behaviours, from prosocial sports to antisocial activities including drug and alcohol use. Individuals who partake in high-risk activities generally score higher on sensation seeking than their low-risk counterparts, and these findings are especially consistent among high-risk athletes. Sensation seeking was placed in a broader theoretical framework, grouping it with other traits like novelty seeking, reward seeking, extraversion, and behavioural activation – all sharing a sensitivity towards reinforcing stimuli. Neurotransmitters have been implicated in reward and punishment sensitivity, and the main neurotransmitter implicated in motivation towards reward is dopamine, while serotonin is commonly implicated in punishment sensitivity. Animal models and imaging studies provide support for these purported roles of dopamine and serotonin in motivational tendencies and personality traits. Genetic association studies on a range of personality traits and psychiatric phenotypes were reviewed, as were findings that support genes encoding proteins within the dopaminergic and serotonergic pathways as potential candidate genes for sensation seeking. Since the 1990s, behavioural geneticists have tested associations between personality traits and variants in numerous candidate genes; some associations have been replicated, while numerous others have failed. Inconsistent reports may stem from weak phenotype measures, improper screening of controls, small sample sizes, failure to correct for multiple tests, and failure to consider interactions (Attia et al., 2009; Ebstein & Israel, 2009; Lusher, Chandler, & Ball, 2001). With these potential faults in mind, when designing and executing the studies described in this thesis, attempts were made to recruit relatively large, homogeneous samples of athletes, multiple variants were tested and stringent corrections were applied, and when the  49  sample was large enough joint-analysis (two-stage) designs were used to replicate significant findings. Studying sensation seeking in athletic samples allowed phenotypes to be measured in a novel way: one based on patterns of sports behaviours, using questionnaires with strong psychometric properties in order to characterize “approach-related” phenotypes. Sports provide a context where patterns of sensation-seeking behaviours can be quantified through self-report, and unlike other disinhibted expressions of sensation seeking, sensation seeking in sport may be potentially unconfounded with impulsivity. Psychiatric studies have similarly recruited groups exhibiting common disinhibited behaviours (e.g., gambling problems, alcohol misuse) and have quantified patterns of behaviours specific to that population (e.g., quantity-frequency indices). One advantage of investigating genetic associations in cohorts that have distinctive, shared characteristics is that there exist domain-specific tools to measure phenotypes (e.g., Dick et al., 2009) and in some cases, studies have found genetic associations with domain-specific measures but not when comparing genotypes across groups defined by broad diagnostic criteria or personality traits (e.g., J. E. McGeary, Esposito-Smythers, Spirito, & Monti, 2007). Sensation seeking, rather than novelty seeking, was the phenotype selected as being a more appropriate construct for assessing personality in a sports population. Although sensation seeking is moderate-to-highly correlated with novelty seeking, tools that measure sensation seeking are more reflective of the intense and complex sensations (Zuckerman, 2005b) sought after by highrisk athletes. Exploring the biological (using genetics) and psychological (using personality questionnaires) characteristics among risk-inclined athletic populations may shed some light on the factors that underlie a predisposition towards risk-taking, leading to more personality-tailored  50  prevention programs for other deviant forms of risk-taking. If genetic variants commonly associated with addiction phenotypes are also overrepresented in a cohort of high-risk athletes, then the high-risk sport could potentially act as an outlet for an innate predisposition to risky behaviours. In a document published by the United Nations (United UNODCCP, 2002) sport in general has been cited as a strategy for drug prevention, but little research has focused on high-risk sports, which may be better suited to the personalities (and potentially innate needs) of riskseeking individuals. This dissertation includes multiple studies involving comprehensive investigation of personality traits and candidate genes in high-risk sport populations. The first series investigate associations in cohorts of skiers and snowboarders as practitioners of potentially hazardous, yet popular sports where patterns of both low and high sensation seeking are commonplace. Most alpine ski resorts offer a variety of runs that, combined with the option of going “out-of-bounds”, span the gamut from relatively safe to very hazardous; therefore, a range of sensation-seeking behaviour should be present in the skiing and snowboarding community. These projects were followed up with an exploratory study on high- and low-risk sport participants, carefully selected for ability and sport practices. Exploratory analyses of personality and genetic variables are carried out as a quasi case-control study. Variants from candidate genes were chosen based on purported functional differences between alleles (a summary of functional candidates is shown in Table 2-1, and based on previous associations with approach-related traits; see Appendix E).  51  Table 2-1 A summary of reportedly “functional” variants based on the literature A summary of reportedly “functional” variants based on the literature Gene BDNF COMT DAT1 DBH  DRD1 DRD2  Polymorphism Vall66Met Val158Met 3’ UTR VNTR 3’ UTR -1021 C/T (aka -970 C/T) Arg549(535)Cys Intron 5 TaqI 3’ UTR A/G  rs ID rs6265 rs4680 n/a rs27072 rs1611115 rs6271 rs2519152 rs686  Function Impaired memory and hippocampal function Differences in enzyme activity Differences in expression mRNA expression and translation Differences in plasma enzyme activity Plasma enzyme activity Plasma enzyme activity Differences in expression  Taq1A Intron 6 957 C/T -521 C/T  rs1800497 rs1076560 rs6277 rs1800955  Differences in expression Differences in expression Differences in expression Transcriptional activity (mixed results)  References (Egan et al., 2003) (Chen et al., 2004; Lotta et al., 1995) (VanNess et al., 2005) (Pinsonneault et al., 2011) (Zabetian et al., 2001) (Zabetian et al., 2001) (Tang et al., 2006) (W. H. Huang & Li, 2009; W. H. Huang, Payne, Ma, & Li, 2008)  (Noble et al., 1991) (Y. Zhang et al., 2007) DRD2 (Duan et al., 2003) DRD4 (Kereszturi et al., 2006; Okuyama et al., 2000) 120-bp duplication rs4646984 Transcriptional efficiency (D'Souza et al., 2004) Exon III VNTR n/a Antagonist binding profiles (Asghari et al., 1994; Van Tol et al., 1992) DRD3 Ser9Gly rs6280 Agonist binding affinity (Lundstrom & Turpin, 1996) HTR1A 1019 C/T rs6295 Differences in autoreceptor expression (Lemonde et al., 2003) HTR2A -1438 G/A rs6311 Differences in expression (mixed results) (Myers, Airey, Manier, Shelton, & SandersBush, 2007; Smith et al., 2012) HTR2A His452Tyr rs6314 Altered cell signalling (Hazelwood & Sanders-Bush, 2004) MAO-A Arg297Arg rs6323 Enzyme activity (Jansson et al., 2005) MAO-B Intron 13 A/G rs1799836 MAO enzyme levels (mixed results) (Garpenstrand et al., 2000; Pivac et al., 2006) SLC6A4 5HTTLPR n/a Transcriptional efficiency (Bradley et al., 2005; Lesch et al., 1996). SLC6A4 C/T rs25532 Differences in expression (Wendland et al., 2008) TH -824 C/T rs10770141 Differences in promoter activity (Rao et al., 2008) Note. BDNF = brain-derived neurotrophic factor, COMT = catechol-O-methytransferase, DAT1 = dopamine transporter, DBH = dopamine-betahydroxylase, DRD2, DRD3, DRD4 = dopamine receptors, HTR1A, HTR2A = serotonin receptors, MAO-A, MAO-B = monoamine oxidase, SLC6A4 = serotonin transporter, TH = tyrosine hydroxylase, UTR = untranslated region, VNTR = variable number of tandem repeats.  52  2.1  Objectives The work described in this dissertation can be broadly divided into three projects; all of  which were carried out with a common goal in mind: to gain a better understanding of the psychological and genetic characteristics associated with participation in high-risk sports. 1) Project 1: Evaluate the psychometric properties of the newly developed contextual sensation-seeking questionnaire (CSSQ-S) in multiple samples of skiers and snowboarders. 2) Project 2: a. Replicate findings from unpublished research (Thomson, 2008; association in DRD4 -521 C/T and sensation seeking in skiers/snowboarders). b. Analyze whether associations exist between sensation-seeking measures and other single nucleotide polymorphisms in dopaminergic and serotonergic pathways. These include genes involved in dopamine/serotonin transport, synaptic transmission and metabolism (see Appendix E and Table 2-1). c. Analyze whether associations exist between sensation-seeking measures and polymorphisms (both single nucleotide and repeat variants) in the DRD4 promoter region. 3) Project 3: a. Analyze whether polymorphisms involved in dopaminergic and serotonergic neurotransmission (chosen based on Project 2 results, functional support, and previous associations) are over-represented in high-risk sport participants when compared to a low-risk sport group.  53  b. Confirm that risk-inclined sport populations exhibit higher levels of sensation seeking (as measured from questionnaires) compared to low-risk athletes. 2.2 2.2.1  Hypotheses General hypotheses Twin studies have shown that approximately 60% of variability in the sensation-seeking  trait is due to genetic influences (Stoel et al., 2006). Dopamine has consistently been implicated in novelty and reward-seeking in mice and humans; therefore, it is generally accepted that dopamine is likely to be involved in other approach traits, including sensation seeking. The moderate heritability of sensation seeking combined with a proposed neural pathway lead me to hypothesize that variations within genes in the dopamine pathway will be associated with sensation-seeking scores. Sensation seeking is a complex trait and therefore a polygenic modeof-inheritance is likely, and any associations found are likely to be small.  2.2.2  Specific hypotheses Hypotheses for each project are shown below and are presented as null hypotheses (H0)  and alternative hypotheses (HA).  Project 1 H0: There will be no relationship between patterns of context-specific ski behaviours (measured using the CSSQ-S) and approach-related traits. HA: Contextual sensation-seeking scores will correlate moderately with established sensation-seeking scales in all samples.  54  Project 2 i) H0: Contextual sensation-seeking scores will not vary between DRD4 -521 C/T genotypes in a sample of skiers. HA: DRD4 -521 C/T will be associated with domain-specific sensation-seeking scores in a sample of skiers and snowboarders (replicating unpublished results; Thomson, 2008). Specifically, the CC genotype will be associated with high levels of sensation seeking. ii) H0: No SNPs within candidate genes investigated will be associated with sensation seeking. HA: SNPs within genes involved in dopamine transport, synaptic transmission, and/or metabolism will emerge as candidates for sensation seeking (see Appendix E). Project 3 i) H0: No SNPs within candidate genes will be associated with high-risk sport participation. HA: Alleles that have previously been associated with sensation seeking will be overrepresented in the risk-inclined athletic group compared to the low-risk sport group. ii) H0: There will be no significant differences in personality traits between athletes grouped by their participation in high- or low-risk sports. HA: Athletes who participate in high-risk sports will score higher on measures of sensation seeking than low-risk sport participants.  55  2.3  Methodology overview There are three projects described in this thesis, each including multiple participant  cohorts. They are briefly described below and then discussed in detail in the following chapters. The first project consists of studies to support the use of the contextual sensation-seeking questionnaire for skiing and snowboarding (CSSQ-S, Appendix F) that was developed as a reliable measure of sport-specific sensation seeking (development and validation of this tool is described in Chapter 3). Questionnaire development involved a series of studies to support item, content, and external validity of the CSSQ-S (and a generalized version of the questionnaire, the CSSQ, was later designed to be used for multiple downhill sports). The data described in Chapters 3 and 4 incorporate data gathered during my MSc research (Thomson, 2008) with data gathered during my tenure as a PhD student. The same samples recruited for Project 1 were used in the genetic analyses of Project 2 (described below), with the exception of a small sample recruited solely to obtain reliability data for the CSSQ-S. The second project involves analyzing variants in candidate genes for sensation seeking in two populations of skiers and snowboarders (“MSc” sample and new “Festival” sample, described below). The data are presented in Chapters 4, 5, and 6 and all analyses involve comparing quantitative measures of personality (dependent variables = sensation-seeking measures) between genotype groups (independent variables). The third project involves comparing frequencies of variants in candidate genes between high- and low-risk sport participants and psychological measures between groups (described in Chapter 8; employing a quasi case-control design). A separate analysis of variants that were associated with skiing and snowboarding phenotypes in Chapter 4 were analyzed in a subset of  56  athletes who reported participation in a downhill sport from the high- and low-risk sport cohorts (described in Chapter 7). All of the participants in the projects described in this thesis were athletes who were moderately proficient at their sport (based on self-report). Inclusion criteria for age varied depending on the project (details are provided in respective chapters). For example, age in skiers was initially limited from older-youth to middle-age adults (17 to 49 years)11, but the upper age was relaxed to 60 years for high- and low-risk sport participants because older adults are a significant demographic participating certain sports (e.g., mountaineering) (British Mountaineering Council, 2003). Although the same samples were used for genetic and psychology analyses, the sample sizes differ due to varying exclusion criteria (e.g., ethnic exclusions were felt to be necessary for the genetic analyses, but not for psychological analyses; exclusions for sport ability affect CSSQ-S data, but not personality traits) and/or missing data (e.g., incomplete questionnaires or failure to genotype). Specific rationale or explanations of exclusions for analyses are discussed in the corresponding chapters. A brief overview of each sample is provided below. Further details about recruitment and variables measured, along with descriptive statistics are included in respective chapters.  2.4  Overview of participants and recruitment All participants provided written, informed consent, and the studies were all reviewed  and approved by the UBC Clinical and/or Behavioural Research Ethics Boards. Both the UBC Clinical Research Ethics Board and the University of Bordeaux Sagalen approved the studies  11  Initially, age was limited to 40 years (see consent form, Appendix H), but due to the significant number of >40 year olds (over 10% of the sample) reporting sufficient skiing ability and frequency the age criteria was relaxed. 57  involving high- and low-risk participants recruited from France and Canada. Ethics certificate numbers are detailed in the Preface.  Figure 2-1. Sample Overview. Numbers represent sample sizes. Project 1 includes the entire MSc sample, a portion (n = 530) of the Festival sample (after exclusions), two peer samples, and a re-test reliability sample (RT). Project 2a includes portions of the MSc and Festival samples successfully genotyped at the -521 C/T locus (n = 117 and 386, respectively). Project 2b includes a portion of the MSc sample (n = 174) that consented to DNA banking and a portion of the Festival (n = 536) sample meeting exclusion criteria, of these a total of 599 amplified successfully. Project 3 includes athletes participating in either high- or low-risk sports.  58  2.4.1  Sample 1 (Pilot sample) The first sample included 223 skiers and snowboarders predominantly from Western  Canada, mostly recruited for my MSc thesis project. Due to low DNA yields and genotyping difficulties, only 74 genotypes for the DRD4 -521 C/T polymorphism (rs1800955) were confirmed at the time of my MSc thesis defense (Thomson, 2008). Many participants were willing to provide an additional sample of DNA and a large portion (~80%) of skiers and snowboarders (Caucasian males = 86, females = 88, n = 174 out of 223) consented to DNA banking, granting permission to analyze the DNA for other potential sensation-seeking candidate genes (discussed in Chapter 5). Additional samples collected for my MSc studies were genotyped for the -521C/T SNP, ultimately resulting in a total of 117 genotypes. These DNAs comprise the “Pilot sample” for Chapter 4, entitled, “The -521 C/T variant in the dopamine-4receptor gene (DRD4) is associated with skiing and snowboarding behaviour”. The exploratory factor analysis stage of the CSSQ-S validation (Chapter 3) included psychological data from Sample 1 (n = 198; analyses which were completed during my MSc thesis; Thomson, 2008). The sample sizes for any analyses carried out post-MSc research are slightly larger because after finalizing the MSc sample there was continued interest from research subjects. By the time the injury data were formally analyzed (Chapter 3: to support external aspect of validity for the CSSQ-S), the sample had increased to n = 220. None of the injury data were included in my MSc thesis.  2.4.2  Sample 2 (Festival sample) A total of 668 skiers and snowboarders (ages 17 to 49 years of age) visiting the Telus  World Ski and Snowboard Festival in Whistler, BC in April 2010 participated in the study. Table  59  2-2 provides detailed demographic information on the full sample. Chapters 3, 4, 5, and 6 include participants from the Festival sample, but sample sizes vary depending on exclusion criteria (i.e., exclusions of 69 to 138 subjects). For example, no missingness was tolerated in the data set that was used for questionnaire validation, and subjects had to meet the post-screening ability requirement (see section 2.4.6) and after exclusions the questionnaire validation study (Chapter 3) included 530 athletes from the Festival sample (Table 2-3). Because both personality and CSSQ-S scores were included in the genetic analyses, personality scores could be included (even if the participant did not meet the sport ability requirement for the CSSQ-S), hence the genetic studies include as many as 599 Festival sample participants (Table 2-3). Data from the Festival sample were used in the confirmatory factor analysis stage of questionnaire validation (Chapter 3, n = 530), stage 2 of -521 C/T study (Chapter 4, n = 386), in the multi-SNP study (Chapter 5 n = 599), and in the DRD4 promoter study (Chapter 6, n = 444).  Table 2-2 Festival sample participant characteristics, pre-exclusions Festival sample participant characteristics (pre-exclusions) Age (years) Sex Ethnicity  Pre-exclusions (n = 668) Mean = 27.42, SD = 7.34 58% male, 42% female 91% Caucasian descent, 9% other  Exclusion detailsa n = 10, < 16 or > 49 yrs n = 55 reporting non-Caucasian or mixed ethnicities  Education Marital Status Dependents Ability  69% post-secondary education 20% married or common-law 9% reported having dependents 24% intermediate, 28% advanced, 44% n = 24 reported less than intermediate expert ability Note. athere was overlap between exclusions, i.e., participant ID-839: a 14 year old reporting African descent. SD = standard deviation.  60  2.4.3  Peer samples Peers who had frequently skied/snowboarded with the participants completed a peer-  CSSQ (Appendix G) for samples 1 (Pilot, n = 67) and 2 (Festival, n = 78). The numbers are substantially lower than the full samples because the participant had to be with a peer at the time of recruitment. A mail-back option was not offered, as this method could compromise the anonymity of the peer’s responses leading to a greater chance of bias.  2.4.4  Reliability sample Skiers and snowboarders (n = 33) recruited through UBC psychology pool and online  postings participated in a reliability analysis of the CSSQ-S (described in Chapter 3 and in Table 2-3). Participants completed questionnaires only (no DNA analyses), and the data were not combined with any other samples for psychological analyses.  61  Table 2-3 Participant characteristics (post-exclusions) for Projects 1 and 2 Participant characteristics (post-exclusions) for Projects 1 and 2 Variable  Age (years) Sex Ethnicity Education Marital Status  Questionnaire validation (Chapter 3) Factor analysis (n = 530) 26.70 (6.02) 41% female 1% non-European 29% high school, 71% post-secondary  Reliability (n = 33) 26.4 (4.75) 33% female 9% non-European 100% post-secondary  19% married or common-law 8% reported having dependents 24% intermediate, 32% advanced, 44% expert  36% married or common-law 3% reported having dependents 3% intermediate, 42% advanced, 55% expert  Genetic associations (Chapters 4, 5, 6) Chapter 4 (n = 386) 26.3 (5.9) 42% female 100% European 30% high school, 67% post-secondary, 3% missing 17% married or common-law 8% reported having dependents 23% intermediate, 32% advanced, 45% expert  Chapter 5 (n = 599) 27.12 (6.45) 43% female 100% European 28% high school, 72% post-secondary  Chapter 6 (n = 444) 26.92 (6.27) 42% female 100% European 28% high school, 72% post-secondary  22% married or 22% married or common-law common-law Dependents 9% reported having 8% reported having dependents dependents Ability 24% intermediate, 31% 22% intermediate, 32% advanced, 42% expert, advanced, 42% expert, 3% beginner/novicea 4% beginner/novicea Sport 60% skiers, 40% 75% skiers, 25% 54% skiers, 40% 56% skiers, 38% 55% skiers, 39% snowboarders snowboarders snowboarder, 6% both snowboarders, 5% both, snowboarders, 5% both Note. All samples except the reliability sample are dependent (drawn from the sample larger Festival sample). Values for age represent mean (SD). a  CSSQ scores from individuals reporting less than intermediate ability were excluded from analyses.  62  2.4.5  High-risk/low-risk samples Athletes participating in high-risk sports were largely recruited in France (n = 141), while  low-risk athletes were recruited both from France and Canada (n = 132, Table 2-4). Genetic and psychological data from these samples are analyzed in Chapter 7 and 8. The high- and low-risk sport samples differ on more variables (i.e., sex, location of recruitment) than the selection variable (i.e., sport participation), so a number of additional screening analyses and exclusions were carried out in order to analyze these samples using a quasi case-control design (described in Chapter 8).  Table 2-4 Participant characteristics (pre-exclusions) for Project 3 Participant characteristics (pre-exclusions) for Project 3 Variable  High-risk sport group (n = 141)  Low-risk sport group (n = 132)  Age (years) Sex Ethnicity Education Marital Status Dependents Ability  29.2 (9.2) 19% female 96% European 21% high-school, 79% post-secondary 40% married or common-law 16% reported having dependents 8% intermediate, 19% advanced, 71% expert  25.8 (9.8) 48% female 83% European 30% high school, 70% post-secondary 25% married or common-law 8% reported having dependents 34% intermediate, 36 advanced, 19% expert  Note. Values for age represent mean (SD). Exclusions varied depending on the type of analysis, participants meeting sport and age criteria are shown in this table.  2.4.6  Participant exclusions The exclusions for age and ability have been chosen to avoid confounding results. Levels  of sensation seeking decrease after the age of 40 years (Zuckerman, 1979, p. 125), possibly due to an increase in levels of an enzyme that breaks down dopamine, monoamine oxidase type B (Zuckerman, 1979, p. 376). Although the genetic make-up of an individual does not change with age, lower levels of dopamine may have an effect on sensation-seeking, thus confounding studies  63  measuring quantitative traits if too broad an age range is included. In an attempt to be as inclusive as possible, minors (ages 17 and 18 yrs) were included in the Festival sample (Appendix H, consent form). A number of the festival competitors (and athletic festival goers) were underage and the study had ethical approval to treat these athletes as emancipated adults. Skiing/snowboarding ability was limited to intermediate or higher levels to ensure that all participants have the ability and experience to display so-called “sensation seeking” skibehaviours in the field. Participants from all ethnic background were included in the psychology-based analyses, but to minimize bio-geographical diversity, only participants self-reporting European descent were included in genetic analyses (for example see demographics, Table 2-2). Individuals reporting non-Caucasian ancestries were excluded in an attempt to control for population stratification (the presence of systematic differences in allele frequencies between experimental subgroups (e.g., case: control) that may be due to differential ancestry in the two groups (Attia 2009, B).  2.5  Procedures Participants were recruited through a variety of methods including posters, email, online  forums, word-of-mouth, and approached in person. Participants did not receive any financial compensation, but the Telus Festival offered a unique incentive, and details are provided below. The 2010 Telus World Ski and Snowboard Festival provided an ideal recruitment ground for the study. The festival included competitions for elite riders along with 10 days of free concerts, attracting a large number of ski/snowboard enthusiasts. The festival organizers provided a tent in the concert plaza for recruitment of spectators (who were largely recreational  64  skiers and snowboarders visiting Whistler Blackcomb Resort for the festival), and a table in the athlete’s registration building for recruitment of competitive athletes. In collaboration with a local ski company (Crown Skis Ltd.), we offered all kiosk visitors a chance to win a new pair of skis (no participation in the research necessary) and a voucher for a hot beverage at a local café. Riders filled out two questionnaires (CSSQ-S and ZKPQ ImpSS, Appendix D and F), provided demographic data, and two buccal samples by brushing the inner cheek with a cytobrush (Fisher Scientific, Canada). A peer of the subject (when present at the time of recruitment) was invited to fill out a third party questionnaire (optional component, a consensual validity check, shown in Appendix G). The entire process took between 10 and 15 minutes. During the pre-screening, all participants indicated that they were at least intermediate ability and were between the ages of 17 and 49 years.  2.6  Methods Participants from each project described above (section 2-4) completed evolving versions  of the same instruments. For example, the Pilot sample completed the very first template of the CSSQ-S, which was a 13-item questionnaire (described in Chapter 3). Participants recruited after my MSc defense and at the Telus Festival completed a 10-item version of the questionnaire since the factor structure of the 13-item CSSQ-S had been analyzed, and three items were removed because they appeared to represent a separate construct (loaded onto an independent factor). The instrument used to measure personality traits also varied between studies. After testing the correlations between the CSSQ-S and the ZKPQ subscales (to provide evidence for discriminant validity; Chapter 3) only the ImpSS subscale (Appendix D) was included in all other studies. See Appendix I for correlations between CSSQ and ZKPQ subscales. This  65  subscale was the most pertinent to the objectives (to investigate associations with sensation seeking) and the exclusion of the other four subscales greatly reduced the time commitment for the participants. The measures will be explicitly presented in the methods sections in the following chapters. All samples, with the exception of a sample recruited for a reliability study within the questionnaire development chapter (Chapter 3), completed a questionnaire portion and provided a buccal cell sample for genetic analysis.  2.6.1  Buccal DNA preparation After testing various methods of collection (saliva, blood, buccal) and isolation  (Invitrogen PureLink genomic DNA extraction kit, Oragene-DNA kit) we concluded that our laboratory’s standard alcohol-based isolation/purification technique produced the highest yields of DNA at the lowest cost, without significant impact on purity (estimated based on wavelength ratios 260/280) (see Appendix J; Mulot et al., 2005). In an attempt to standardize the buccal swabbing technique, all participants that provided buccal cell samples were given visual, verbal, and written instructions (see Appendix K), and all subjects provided two samples. After brushing, the cytobrushes were stored in paper envelopes at room temperature, allowing them to air dry and then were frozen at -20°C for longer-term storage. Buccal cell DNA was isolated from cytobrushes using a standard purification technique described by Saftlas et al. (2004) (Appendix L). The brushes were incubated at 55°C overnight (at least 8 hrs) in 700 µl lysis buffer (for recipes, see Appendix M) containing proteinease K (0.11 mg/ml) to breakdown cellular proteins and remove the cells from the cytobrush. After  66  incubation, the tubes were centrifuged for 2 minutes at 15 900 g at 4 °C. The brushes were discarded, and RNAse (0.03 mg/ml) was added to the supernatant which was then incubated for 60 minutes at 55°C to denature RNA. Subsequently, 320 µl of 5M potassium acetate precipitation buffer (KOAc) was added and the tubes were stored on ice for at least 20 minutes and then centrifuged (15 900 g) for 5 minutes. The supernatants were transferred to new tubes and the pellets of precipitate discarded. The DNA was then precipitated out of the remaining solution by the addition of glycogen (0.025 mg/ml) and 510 µl isopropanol followed by incubation on ice for 20 minutes. The tubes were centrifuged (15 900 g) for 10 minutes and the supernatants were discarded leaving DNA pellets, which were then rinsed with 70% ethanol (200 µl) followed by a 1-minute centrifuge to remove remaining salts. The ethanol was carefully discarded from each tube and the DNA pellets were air dried and re-suspended in 50 µl TE buffer (10 mM Tris/Cl, 1 mM EDTA pH 8.0) and stored at -20°C for future use. The protocol was modified slightly to include an extra alcohol rinse step, and lower elution volume (50 µl) to ensure the samples would meet the minimum purity and concentrations required by the genotyping facilities (laboratory protocol and recipes for reagents in are shown in Appendices K and L). Copy-number variant polymorphisms were genotyped in the Rupert or Robinson laboratories (details in Chapters 6 and 7), and all SNPs (except -521 C/T, described in Chapter 4) were genotyped by Genome Quebec at McGill University using the Sequenom iPlex® platform. Genome Quebec requires 30 µl of DNA at a concentration of 20 ng/µl. Samples were quantified and diluted accordingly (Nanodrop 2000c, Thermo Fisher Scientific Inc.). Genome Quebec analyzed a total of eight plates in the study described in Chapter 5, and three plates in the study  67  described in Chapter 8. All 96-well plates contained two wells reserved for negative controls, and two reserved for positive controls. A sample plate layout is shown in Appendix N. Genotyping of three copy number variants was attempted in the Rupert laboratory using polymerase chain reaction (PCR) and gel electrophoresis, but after numerous attempts failed optimization. These include the SLC6A4 5HTTLPR (the polymorphism most commonly associated with internalizing disorders, but that has also been associated with impulsivity and reward seeking); the DAT1 3’ UTR 40-bp VNTR (commonly associated with externalizing disorders, especially ADHD, and approach-related traits); the DRD4 48-bp VNTR (one of the most studied variants for approach-related traits). Details about phenotypes that have been studied in association with these variants are found in Appendix E. Details about primers and attempted protocols are shown in Appendix O. As the DRD4 VNTR was the most studied candidate for approach-traits, specifically, this variant was attempted again using a more sensitive genotyping technique in the Robinson laboratory with sizing of alleles using a capillary system (ABI Prism 310 Genetic analyzer) and genotyping was successful (details in Chapter 7).  68  Chapter 3: The Contextual Sensation Seeking Questionnaire for skiing and snowboarding: Development of a sport specific scale12  3.1  Summary The Contextual Sensation Seeking Questionnaire (CSSQ-S) was developed to measure  patterns of sensation-seeking behaviours in skiing and snowboarding. Three studies were conducted supporting several aspects of its validity. First, a focus group (n = 4) generated items representative of sensation seeking in skiing and the factor structure was explored in a sample of skiers (n1 = 220). Second, the factor structure was confirmed using data from an independent cohort (n2 = 530). Finally, evidence is provided for criterion-relevance and applied utility of CSSQ-S scores by demonstrating positive relationships between scores and self-reported injury. CSSQ-S scores explained greater variance (n1 = 217, ! = .358, p < .001) in injury prevalence than an established assessment tool (Zuckerman’s Impulsive-Sensation Seeking scale). In summary, the CSSQ-S represents a psychometrically promising measure of contextual sensation seeking and may be used to explore factors associated with risk-taking in skiing and snowboarding.  3.2  Introduction Sensation seeking is measured using a variety of trait-related questionnaires including the  Sensation Seeking Scale (SSS; Zuckerman, 1994), the UPPS (Urgency, Premeditation, Perseverance, and Sensation Seeking) Impulsive Behaviour Scale (Whiteside & Lynam, 2001)  12  The data in this chapter was published as, “The Contextual Sensation Seeking Questionnaire for skiing and snowboarding: Development of a sport specific scale” in the International Journal of Sport Psychology, 2012. 69  and the impulsive-sensation seeking scale (ImpSS) of the Zuckerman-Kuhlman Personality Questionnaire (ZKPQ; Zuckerman et al., 1993), all designed to measure sensation seeking across a range of life categories (e.g., social, career, physical, etc.). A significant limitation of these instruments for use in the study of narrow behavioural tendencies such as high-risk snow sport is that they measure sensation seeking in general as opposed to sensation seeking in specific activities. In addition, the items are typically worded in a hypothetical form (e.g., “I would like to…”) rather than reflecting actual sensation-seeking behaviours or experiences (e.g., “I like to…”). From a construct validity perspective (cf. Messick, 1995), such instances may result in a lack of concordance between what a person claims that he/she would like to do, and his or her actual behaviours. A lack of concordance between desires and actual behaviours make it difficult to characterize sensation seeking in studies that measure correlations between sensation seeking and other psychological factors (e.g., risk perception, self efficacy, disinhibition), or in studies that explore biological underpinnings for sensation-seeking behaviour using methods such as neuroimaging or genetic analysis. In a sport population, a context-specific measure of sensation seeking may help to identify psychological processes that are associated specifically with risky sport behaviour. For example, previous studies have examined psychological constructs such as self-efficacy (Bandura, 1997) and sensation seeking in a variety of high-risk sports and found that only self-efficacy was significantly associated with the level of risk-taking behaviour in the sport (D.J. Llewellyn et al., 2008; Slanger & Rudestam, 1997). However, it is important to note that differences in risk-taking behaviours were only observed when a domain-specific measure of self-efficacy was utilized. Specifically, no differences in risk-taking behaviour were observed when general self-efficacy was assessed (Slanger & Rudestam, 1997). Similarly, the lack of  70  association between risk-taking in sports and sensation seeking may be explained by the use of generalized measures of sensation seeking (e.g., ZKPQ ImpSS and SSS). Narrower traits (or facets) can predict variance in specific behaviours not accounted for by more general traits (e.g., Paunonen & Ashton, 2001) like sensation seeking. Arguably, a brief self-report instrument designed to assess sensation-seeking behavioural tendencies specific to a sport may provide stronger predictive utility in terms of an individuals’ propensity to engage in risky behaviours within that context. Such a tool may be useful to provide a more focused characterization of participants when exploring motivations for participation or to help inform injury prevention in downhill sports by identifying risk-seeking athletes. Three studies were conducted in order to develop a sport-specific questionnaire that measures patterns of sensation-seeking behaviours in skiers and snowboarders (called “The Contextual Sensation Seeking Questionnaire for Skiing and Snowboarding” (CSSQ-S)), and evidence for several aspects of construct validity are provided in this chapter (cf. Messick, 1989, 1995). Construct validity “comprises the evidence and rationales supporting the trustworthiness of score interpretation in terms of explanatory concepts that account for both test performance and score relationships with other variables’’ (Messick, 1995, p. 743). The studies presented in this chapter involve (a) developing items and establishing evidence for the content aspect of construct validity (e.g., the content relevance and representativeness of items; study 1), (b) establishing evidence for the structural aspect of construct validity (i.e., factorial validity) using exploratory (study 1) and confirmatory factor analytic procedures (study 2), and, (c) providing evidence for the external aspect of construct validity (evidence based on relations to other variables, i.e., criterion relevance and applied utility) through an examination of the relationships between sensation-seeking measures (CSSQ-S and ZKPQ ImpSS, studies 1 and 2) and peer  71  evaluations (studies 1 and 2), as well as through an evaluation of relative associations of scores on measures of sensation seeking (CSSQ-S and ImpSS) and injury prevalence (study 3).  3.3  Study 1: Item generation and preliminary evidence for validity The purpose of study 1 was to develop a battery of items that reflect the sensation-  seeking construct in skiing and snowboarding, then to investigate the factor structure using exploratory factor analysis (EFA), and finally to provide preliminary evidence for external aspects of construct validity.  3.3.1  Participants and procedures Item generation. The research literature on sensation seeking, risk-taking, and sensation  seeking in sports was reviewed in order to generate a detailed conceptualization of the specific content and range of sensation seeking, such as what specific behaviours sensation seeking encompasses. This initial review highlighted several “key words” (e.g., “new”, “thrilling”, and “risks”) associated with sensation seeking that were believed to be representative and relevant to context-specific sensation seeking for skiing and snowboarding. A group of proficient skiers and snowboarders were then recruited in order to develop the initial battery of items. Consulting members of a target population is an important step to test the comprehensiveness and relevance of items for a given construct (Vogt, King, & King, 2004). The group included two male and two female advanced and expert skiers and snowboarders, average age 27.5 (SD = 1.29) years. Two of the athletes (one expert skier, one expert snowboarder) had lived in a mountain town (in Alberta, Canada) for at least two ski seasons, while the other two participants (one advanced snowboarder, one expert skier) rode regularly but were “weekend riders” from British Columbia,  72  Canada. The group was provided with Zuckerman’s definition of sensation seeking (Zuckerman, 1979) and several key words (i.e., thrilling) that the initial literature review had shown to be important aspects of sensation seeking in general. They subsequently generated a list of “sensation-seeking” behaviours specific to skiing and snowboarding, creating an item pool of approximately 20 items. The initial list was qualitatively reviewed, grouping items under “novelty”, “speed/thrill”, “risk/impulsive”, and removed unrelated and/or redundant items (e.g., two items focused on “jumping”, both capturing aspects of thrill and risk, but were too similar to include both). The initial ski/snowboard-specific scale used for exploratory factor analysis contained 13 items (Table 3-1).  Table 3-1 Factor loadings from the EFA (n = 198) and CFA (n = 530) for the CSSQ-S Factor loadings from the EFA (n = 198) and CFA (n = 530) for the CSSQ-S NO  CSSQ Items  Factor loadings  1 2  I like to ski/ride fast. I like to ski/ride down runs that I have never been down before.  EFA .65 .64  CFA .81 .80  3  I like to start a run even if I cannot see what lies ahead (i.e., big cornice).  .55  .74  4  I like to ski/ride out of bounds.  .74  .76  5  I like to attempt jumps even if I’m not sure of the quality of the landing area.  .53  .70  6  I like to push my boundaries when I ski/ride.  .73  .80  7  If I lose control, I don’t try to immediately slow down, I just go with it.  .54  .66  8  If the only way down is a straight line through a narrow pass, I go for it without hesitation even if I know I will have to go fast.  .64  .78  9  I am always trying to find new and exciting ways down a run.  .73  .85  10  A 15-foot high drop off a cliff isn’t too high a jump for me.  .68  .82  11  I slow down on busy runs.  .70†  -  12  I don’t slow down on busy runs, instead I just dodge people.  .99†  -  13  If I see a “danger of avalanche” sign, I will usually try to find another safer route.  <.40  -  Note. The 13-item CSSQ-S was used in the EFA and the 10-item CSSQ-S was used in the CFA. Items 11, 12, and 13 were removed from the instrument for the CFA. EFA = exploratory factor analysis, CFA = confirmatory factor analysis. †Item loaded onto a second factor. 73  Factor analysis and external aspects of construct validity. An independent sample of skiers and snowboarders (which is referred to from herein as “sample 1”) was recruited to explore the factor structure and included 110 males and 110 females13, mean age = 27.1 (SD = 4.8) years, 63% of whom were skiers (the remainder were snowboarders). Further sample characteristics are shown in Table 3-2. Participants were recruited in ski resorts in British Columbia and Alberta, Canada, through posters, online advertisements, and selection in the cafeteria areas. Skiers and snowboarders in sample 1, and the samples described below, were excluded from the analysis for not falling within the age range (17 to 49 years) or skiing/snowboarding ability range (at least intermediate level). Further exclusions for each study are detailed below. The sample age was limited to 49 years and at least intermediate in skill to ensure that all participants had the physical ability to carry out the behaviours described in the CSSQ-S, furthermore sensation seeking is thought to decrease after the age of 40 (Zuckerman, 1979, p. 125). Despite the limited inclusion criteria, it was still anticipated that ski-behaviours might be, to some extent, age and ability influenced; therefore, correlations between the CSSQ-S and demographic variables were measured. All participants provided informed consent and subsequently completed a questionnaire package (described below). In psychological research, the “common method bias” can be problematic because the dependent variable is often measured from one source (e.g., self-report), which may be sensitive to the effects of impression management (Podsakoff, MacKenzie, Lee, & Podsakoff, 2003). With this in mind, a subset of individuals (peers who had frequently skied with the participant and were present at the time of sampling) was invited to fill out a peer-rating version of the  13  Subjects (n = 198) recruited during my MSc. thesis project comprised the majority of this sample. The additional 22 participants responded to the online advertisements and were included only in study 3 (investigating the applied utility of the CSSQ). 74  CSSQ-S in addition to their personal CSSQ-S (i.e., this subset acted both as a participant and an informant). A peer rating supplemented 67 subjects in sample 1.  Table 3-2 Participant statistics for each sample Participant statistics for each sample Sample 1 Exploratory Sample (n = 198) Age (years) Sex Ethnicity  Injury data (n = 217)a  Sample 2 Confirmatory Sample (n = 530) 26.70 (6.02) 41% female 1% non-European  Reliability (n = 33)  27.1 (4.8) 26.68 (4.49) 26.4 (4.75) 50% female 50% female 33% female 8% non-European 7% non-European 9% non-European (6% Asian, 2% (5% Asian, 2% (3% Asian, 6% other) other) other) Education 25% high school, 27% high school, 29% high school, 100% post58% post55% post71% post-secondary secondary secondary, 17% secondary, 18% post-graduate post-graduate Sport 64% skiers 63% skiers 60% skiers 75% skiers Ability 23% intermediate, 24% intermediate, 24% intermediate, 3% intermediate, 36% advanced, 40% 37% advanced, 39% 32% advanced, 44% 42% advanced, expert expert expert 55% expert Days 27.0 (16.3) 25.2 (15.5) 30.9 (15.4) days/year 36.1 (11.4) skied/year days/year days/year days/year Note. athe first two columns (exploratory and injury) are not independent samples. The numbers shown for variables “age” and “days” are in the following format: mean (standard deviation).  3.3.2  Measures Demographic and sport information. Participants completed a brief demographic  questionnaire that included age, sex, education, and ethnicity and questions about their sport participation that included ability and average number of days skied per year (Table 3-2). Ability was rated using a self-report scale: “beginner”, “novice”, “intermediate”, “advanced”, or “expert” (with a note defining “expert” as “any terrain, any condition”) (Appendix F). The Contextual Sensation Seeking Questionnaire for Skiing and Snowboarding (CSSQ-S). The preliminary measure of sensation seeking for skiing and snowboarding used in 75  the exploratory factor analysis had 13 items (see Table 3-1) scored using a 5-point Likert scale anchored by 1 (strongly disagree) to 5 (strongly agree). In addition to the self-rating of sensation seeking, the CSSQ-S was also used to obtain a peer rating of each participant’s perceived sensation-seeking behaviour. For the peer-CSSQ-S, items were rated using the same 5-point scale, with the only difference being that the pronoun was amended to reflect the third-person perspective (i.e., “he/she” rather than “I”). Zuckerman-Kuhlman Personality Questionnaire (ZKPQ). To examine global sensation seeking, participants completed the full ZKPQ (Appendix C). The ZKPQ is a 99-item true or false inventory that assesses five dimensions of personality that include impulsivesensation-seeking (ImpSS), aggression/hostility, sociability, neurotism/anxiety, and activity (Zuckerman et al., 1993). The full ZKPQ includes an infrequency scale, which contains items that might be socially desirable, but are unlikely to be true for anyone (e.g., a high infrequency score would be suspect). Sample 1 completed the entire ZKPQ both to determine the correlation between the CSSQ-S and the ImpSS subscale, and to establish discriminant validity by comparing the correlations between CSSQ-S and the four remaining ZKPQ subscales (i.e., the correlation between data derived from sensation-seeking measures (CSSQ-S and ImpSS) should be greater than each correlation between the CSSQ-S and other ZKPQ subscales (e.g., D. T. Campbell & Fiske, 1959)). The ZKPQ subscales demonstrated acceptable internal consistencies in sample 1, Cronbach alphas ranging from .71 to .99 (shown in Appendix P, comparable with “norms” from Zuckerman et al., 1993).  76  3.3.3  Results Exploratory factor analysis (EFA). EFA was used to examine the factor structure of  the data derived from the CSSQ-S. A total of 198 participants from sample 1 completed the initial 13-item instrument (five subjects were excluded for not meeting age or ability inclusion and 17 subjects had incomplete data for EFA). Generalized least squares estimation and varimax rotation procedures were used to determine the amount of variance accounted for by the factor(s). Using the parallel analysis method (PA) for factor retention (Hayton, Allen, & Scarpello, 2004), two factors had eigenvalues greater than what would be expected in a series of randomly generated parallel samples (Table 3-3). The ten items that loaded onto Factor 1, labelled “contextual sensation seeking”, accounted for the majority of the variance (Table 3-1). Two items loaded onto a second factor, and one item did not load sufficiently onto either factor (factor loading < .40; Table 3-1). These three items were deleted from the instrument. A second EFA was conducted on the 10 items retained from the initial EFA. This demonstrated that the items were best represented by a single factor (! = 4.94), namely “contextual sensation seeking”, that represented 49.94 % of the variance. The factor loadings for the 10-item CSSQ-S ranged from .54 to .79. The data derived from the 10-item instrument, herein referred to as the “Contextual Sensation Seeking Questionnaire for Skiing and Snowboarding (CSSQ-S)” (see Table 3-1; Appendix F), had a Cronbach alpha of .88, indicating a high internal consistency (Clark & Watson, 1995).  77  Table 3-3 Results from parallel analyses for 50 randomly generated samples Results from parallel analyses showing average and 95th percentile eigenvalues for 50 randomly generated samples Actual ! Average ! SD 95th percentile (N1, n = 198) (N50, n = 198) 1 5.50 1.44 0.071 1.46 2 1.40 1.33 0.051 1.34 3 0.98 1.24 0.031 1.25 4 0.85 1.17 0.029 1.18 5 0.79 1.10 0.028 1.11 6 0.64 1.04 0.033 1.05 7 0.57 0.98 0.033 0.99 8 0.57 0.93 0.027 0.93 9 0.47 0.87 0.023 0.88 10 0.40 0.82 0.027 0.83 Note. ! = eigenvalues, N1 = results from single study sample, N50 = results from 50 randomly generated samples. SD Factor  = standard deviation for mean eigenvalues.  External aspects of construct validity. The extent to which the CSSQ-S covaries with demographic variables was analyzed and data derived from the CSSQ-S were not correlated with age, education, or ethnicity (p > .05), but were significantly correlated with ability (r(198) = .65, p < .001). The data derived from the CSSQ-S and the ZKPQ ImpSS were compared using Pearson’s correlation coefficient to establish evidence of criterion validity. A moderate correlation (r = .35 to .60) would show that the new measure is related to an established inventory, while not being redundant. Sample 1 data were normally distributed with no multivariate outliers. Pearson’s correlation coefficient revealed a significant association between the data from global (ZKPQ ImpSS, M = 11.85, SD = 4.12) and context-specific sensationseeking (CSSQ-S, M = 33.61, SD = 7.16) measures (r (192) = .49, p < .001). The correlation falls within the moderate effect size range and the size of the correlations indicated that the measures of sensation seeking are related, but share no more than 25% of their variance.  78  Sample 1 completed the entire ZKPQ (five subscales), therefore correlations between the data from the CSSQ-S and from the other ZKPQ subscales were compared to establish discriminant validity (e.g., D. T. Campbell & Fiske, 1959). The correlations were then compared using Steiger’s (1980) t-test for dependent correlations. Scores from the aggression-hostility (Agg-Hos; M = 6.77, SD = 3.20) and activity (M = 10.04, SD = 3.54) subscales were correlated with CSSQ-S scores (r (192) = .23 and r (193) = .19, respectively, p < .01), and scores from the neuroticism-anxiety (Neur-Anx; M = 6.04, SD = 4.11) subscale were negatively correlated with CSSQ-S scores (r (193)= -.25, p < .001). There was no significant correlation between scores from the CSSQ-S and sociability subscale (M = 9.13, SD = 3.53). The correlation between scores from the CSSQ-S and the ImpSS subscale, however, was significantly greater than each of the correlations between scores from CSSQ-S and Agg-Hos, CSSQ-S and activity, and CSSQ-S and reverse-scored Neur-Anx (tAgg-Hos = 3.37, p = .001; tActivity = 3.87, p = .0001; tNeur-Anx = 2.92, p = .004). Finally, the CSSQ-S was compared to the peer-CSSQ-S version described above. The correlation between the total scores from the self-report and the peer-report was .81 (p < .01).  3.4  Study 2: Structural aspects of construct validity and reliability The purpose of the second study was to confirm the factor structure of the 10-item  CSSQ-S through confirmatory factor analysis and to measure the reliability of the questionnaire over a two-week interval. Additionally, for the samples in which the relevant data were collected, the analyses conducted in study 1 were repeated to provide further evidence for validity (i.e., with a new independent sample).  79  3.4.1  Participants and procedures Confirmatory factor analysis (CFA). A sample of skiers and snowboarders was  recruited at alpine recreation sites (herein referred to as “sample 2”) and included 313 males, 217 females, mean age = 26.7 (SD 6.0) years. Over 93% of the participants from all samples were Caucasian (self reporting European descent), and the majority (60%) of the participants from sample 2 were skiers (see Table 3-2 for additional details). Athletes were recruited through a display kiosk in Whistler Blackcomb ski resort during the Telus World Ski and Snowboard Festival.14 A peer-rating process (as described in study 1) supplemented data from 78 participants in sample 2 (when participants arrived at the kiosk accompanied by peers a peerCSSQ-S was included in the questionnaire package). No participants received monetary compensation. Reliability. The third sample (herein referred to as “sample 3”) was recruited through the UBC Psychology department subject pool (20 males, 13 females, mean age = 26.4 (SD 4.75) years (see Table 3-2 for additional details), and participants received a credit for their participation. The initial sample included 45 participants, but only 37 of them completed the survey at time 2 and of these four were excluded: three did not meet the skiing/snowboarding ability requirement and one was missing questionnaire data. A majority (75%) of the final sample participants were skiers. The size of sample 3, though small, is adequate (>80% power) based on the large effects (e.g., r = .8) that are common in short-interval reliability studies (e.g., Zuckerman & Kuhlman, 2000).  14  Telus Festival Sample (subset of the total, n = 668 described in Chapter 2). All participants with missing CSSQ data or not meeting ability requirement were excluded, leaving 530 participants. Details regarding recruitment are found in Chapter 2. 80  3.4.2  Measures Sensation-seeking measures. To examine global and contextual sensation seeking  participants in samples 2 and 3 completed the 19-item ZKPQ ImpSS subscale and the 10-item CSSQ-S. The other four subscales from the ZKPQ were omitted because although small-tomoderate correlations with other subscales were found in study 1, the ImpSS is the most frequently studied in association with risk behaviours and was significantly more correlated with the CSSQ-S when compared to the other ZKPQ subscales. Both sensation-seeking measures demonstrated acceptable internal consistency (ImpSS: sample 2: ! = .79; sample 3 ! = .68; CSSQ-S: sample 2 ! = .85; sample 3 ! = .87).  3.4.3  Results CFA. The factor structure of CSSQ-S data from the larger sample of skiers and  snowboarders (n = 530, sample 2) was assessed using LISREL 8.8. Specifically, the Weighted Least Squares estimation (WLS) method was employed (due to the ordinal nature of the data) which utilizes the polychoric correlation matrix (Flora & Curran, 2004). Multiple fit indices were considered to determine the quality of model-data fit. Comparative Fit Index (CFI) and Adjusted Goodness-of-Fit Index (AGFI) close to .95 and Root Mean Square Error of Approximation (RMSEA) close to .06 (but less than .08) are indicative of an acceptable fit (Hu & Bentler, 1999). Further, factor loadings greater than .63 provide support for a “very good” model-data fit for EFA studies, and may be considered as a rough guideline for CFA studies (DiStefano & Hess, 2005). The correlation matrix was first screened for bivariate normality; no assumptions were violated. A unidimensional model was tested based on the structure found in the EFA. The 81  initial fit statistic was significant (!2(35) = 140.43, p < .001), but !2 is sensitive to large sample sizes (Hair, Black, Babin, & Anderson, 2009), and therefore other fit statistics were considered. The results from the single factor model provided evidence for an acceptable fit; CFI = .93, AGFI = .97, RMSEA = .075. Factor loadings ranged from .66 to .85 (see Table 3-1), further supporting a unidimensional model. A path diagram is shown in Appendix Q. Reliability analysis. Reliability was assessed through a re-test study (sample 3), comparing the inter-item correlations between responses at times 1 and 2 (two-weeks apart). The re-test reliability between scores at times 1 and 2 for the CSSQ-S and ImpSS were r(31) = .94, p < .01 and r(30) =.70, p <.01, respectively. Means for the CSSQ and ImpSS were similar to the previous two projects, M = 37.50, SD = 6.70 and M = 10.18, SD = 3.08, respectively (at time 1). External aspects of construct validity. Similar to study 1, analyses were carried out for both the correlation between data derived from the two sensation-seeking measures (ImpSS and CSSQ-S), and when available, the correlation between participant and peer CSSQ-S data. Data sets were normally distributed (skewness and kurtosis statistics < |1|) with no multivariate outliers. The correlation between the data from global (ZKPQ ImpSS, M = 12.73, SD = 3.72) and context-specific SS (CSSQ-S, M = 36.77, SD = 6.97) measures in sample 2 and 3 were significant (r (530) = .37, p < .01; r(32) = .45, p < .01) as was the correlation in sample 2 between the total scores from the self-report and the peer-report (r(78) = .84, p < .01). The interitem correlations between peer and self-report were all significant at p < .01 ranging from .45 to .82.  82  3.5  Study 3: Sensation seeking and injury prevalence Risk-taking is a facet of the sensation-seeking construct (Zuckerman, 1979, 1994), and  the CSSQ-S is designed to assess both risky and sensation-seeking behaviours in skiing and snowboarding. Theoretically, sensation seekers differ in their optimal level of arousal and each individual has a “target level” of risk he/she deems acceptable when balanced by expected benefits (reviewed in Zuckerman, 2007a). For example, an individual might decide that his/her behaviour (e.g., skiing at high speeds) carries an acceptable risk depending on the perceived benefit of an activity (e.g., a rush of excitement). In line with these theories, an individual higher in sensation seeking might risk more to obtain his/her optimal level of arousal. Risk involves the possibility of a negative outcome, and in risky sports, injury is the most probable negative outcome. Risk-taking behaviours have been associated with injuries and trauma in high-risk sports and motor vehicle riders (including all-terrain vehicles and motorcycles) (Foley, Draus, Santos, & Franklin, 2009); and because risk-taking is a facet of sensation seeking, it was expected that the frequency of injuries would be higher in sensation seekers. Numerous epidemiological studies on factors affecting injury in skiers and snowboarders exist, but most have focused on environmental (e.g., visibility, snow conditions) and demographic variables (e.g., age and sex) (Girardi, Braggion, Sacco, De Giorgi, & Corra, 2010). The few studies that examined participant characteristics beyond demographics explored the frequency of ski injuries/accidents and used the SSS, and abbreviated versions, to assess sensation seeking (e.g., Bouter et al., 1988; Cherpitel, Meyers, & Perrine, 1998), and interestingly, found associations between low levels of sensation seeking and injury. Context specific measures of sensationseeking behaviours in sport may be able to explain additional variance in risk-taking behaviour more so than general sensation-seeking measures. Other studies that have investigated injury in  83  skiing and snowboarding have employed dichotomous measures of risk-taking, for example by simply asking participants whether they considered themselves “risky” or “cautious” (Ruedl, Abart, Ledochowski, Burtscher, & Kopp, 2012; Ruedl et al., 2010), or have created risk-taking and sensation seeking ski-measures, without providing evidence of validity for the instruments (Cooper, 2009). It was anticipated that scores derived from a multi-item scale (such as the CSSQ-S) specifically developed for skiing and snowboarding would be a better predictor of sensation-seeking-related outcomes than a single-item measure. In order to establish evidence for the external aspect of construct validity (i.e., the applied utility of the CSSQ-S), the relationship between self-reported injury counts and scores on the CSSQ-S and the ImpSS were assessed. It was hypothesized that sport-specific sensation seeking (CSSQ-S) would be associated with a higher injury rate over the course of multiple seasons. Furthermore, it was hypothesized that the CSSQ-S would explain more variance in injury rate than the more general ImpSS scale.  3.5.1  Participants and procedures Male and female skiers and snowboarders from sample 1 who had reported the number of  ski-related injuries sustained over the past three seasons were included in the analysis (n = 217, 50% female; see Table 3-2 for further details). In order to establish the incremental validity of data derived from the CSSQ-S relative to data from the more general ImpSS, a hierarchical regression was conducted with injury rate as the criterion.  84  3.5.2  Measures Sensation-seeking measures. The 10-item CSSQ-S and the 19-item ZKPQ ImpSS  described above (and in Appendices D and E). Ski related injuries. Injury prevalence was measured by two self-report items: (a) number of ski-related injuries sustained during the past season (choice format: 0, 1, 2, >3); and (b) number of ski-related injuries sustained over the last three seasons (open-ended format). Self-report data are comparable to data obtained by other means (e.g., emergency room reports) when interested in frequency (and not the nature and severity) of an injury (Gabbe, Finch, Bennell, & Wajswelner, 2003; Valuri, Stevenson, Finch, Hamer, & Elliott, 2005). Participants were instructed to include only injuries that impaired their sport ability for at least one day (a commonly used definition in injury epidemiology; Goldberg, Moroz, Smith, & Ganley, 2007; Nicholl, Coleman, & Williams, 1995).  3.5.3  Results Data derived from the CSSQ-S and ImpSS were normally distributed (CSSQ, M = 34.28,  SD = 7.41; ImpSS, M = 11.67. SD = 4.07), whereas the injury rate over one season (Mdn = 0, IQR = 1) and three seasons (Mdn = 1, IQR = 2) were positively skewed. To normalize the injury counts square-root transformations were applied. Correlations between injury rate and sensationseeking measures are shown in Table 3-4. A 3-step hierarchical regression was carried out for each: injury rate over one season and injury rate over three seasons as the criterion. Both criterion variables yielded similar final models (Table 3-5). Step 1 included age and sex and accounted for a significant amount of variance in injury over three seasons (p < .001), but not over one season (p > .05). On Step 2, ImpSS was added and this increased the variance in 85  injuries accounted for across three seasons (p < .05), but not over a single season. On Step 3, the CSSQ-S was added and this accounted for a significant increase in injury variance over the ImpSS for both criterion variables (one season, p < .01; three seasons, p < .001). In the final models, the CSSQ-S was significantly related to injuries (single season: standardized ! = .271, p < .01 and three seasons: standardized ! = .358, p < .001). The partial relationships between injury rate and sex, age, and ImpSS were all non-significant (p > .05, Table 3-5).  Table 3-4 Intercorrelations for sensation-seeking score and injury rate Intercorrelations for sensation seeking score and injury rate Intercorrelations Variable CSSQ-S ImpSS Injuries, 1 season Injuries, 3 season CSSQ-S .43** .19** .30** ImpSS .10 .15* Injuries, 1 season .36** Injuries, 3 seasons Note. n = 217. Correlations with injury data were measured using Kendall’s tau and the correlation between CSSQS and ImpSS was measured using Pearson’s r. CSSQ-S = contextual sensation seeking questionnaire for skiing and snowboarding, ImpSS = impulsive-sensation seeking. *p < .01. **p < .001.  86  Table 3-5 Hierarchical multiple regression analyses predicting injury rate from sensation seeking Hierarchical multiple regression analyses predicting injury rate from sensation seeking Injury rates †  Injury count over a single season Injury count over three seasons† Predictor !R2 df F " !R2 df F " Step 1 .016 2, 214 1.73 .050 2, 214 5.57** Sex -.091 -.199** Age -.102 -.134 Step 2 .026 1, 213 2.29 .022 1, 213 5.09* Sex -.071 -.169* Age -.095 -.124 ImpSS .104 .152* Step 3 .074 1, 212 10.96** .083 1, 212 20.91*** Sex .044 -.019 Age -.091 -.119 ImpSS .010 .027 CSSQ-S .271** .358*** Note. .n = 217. †A square root transformation of injury count over one and three seasons was applied. CSSQ-S = contextual sensation seeking questionnaire for skiing and snowboarding, ImpSS = impulsive-sensation seeking. *p < .05, **p < .01, ***p < .001.  3.6  Discussion Sensation seeking has been measured in a number of specialized populations (e.g.,  alcoholics, athletes, gamblers), but the questionnaires used were not specific to the populations being assessed, nor do they reflect actual behaviours that an individual engages in (i.e., they are hypothetical in nature). The purpose of the present study was to develop a self-report sensationseeking questionnaire that is specific to a sport (skiing and snowboarding). An analysis of the psychometric properties of the CSSQ-S was carried out in three samples of skiers and snowboarders, and the data provides evidence for several aspects of construct validity (cf. Messick, 1995).  87  Discussions with the focus group provided support for the content validity (cf. Haynes, Richard, & Kubany, 1995) of the instrument since all of the items in the initial domain-specific sensation-seeking measure were deemed relevant and representative of what might be considered sensation-seeking behaviours in skiing and snowboarding. The CSSQ-S displays strong psychometric properties, with a unidimensional model providing the best fit for the data. The final CSSQ-S is a 10-item, unidimensional scale, which is brief enough to administer to a sport population in the field and appears to measure a person’s tendency to seek out new, thrilling, or physically stimulating experiences while engaged in downhill snow sports, regardless of potential hazards. Although CSSQ-S scores are related to scores from the other ZKPQ subscales, the relationship between CSSQ-S and ImpSS scores was significantly stronger than the respective relationships to scores from the other subscales, thus demonstrating discriminant validity. It was not surprising that correlations between the CSSQ-S and other ZKPQ subscales were present, given that scores from the aggression-hostility subscale have been positively correlated with risk-taking behaviours, and high sensation seekers often have lower levels of anxiety (Zuckerman & Kuhlman, 2000). The moderate correlation between scores from the CSSQ-S and the ImpSS subscale of the ZKPQ provides support that the CSSQ-S is related to Zuckerman’s definition of sensation seeking, yet is not a redundant instrument. Fjell and colleagues (2007) found physiological differences (e.g., cortical habituation) between individuals that have high global sensation seeking (as measured by the ZKPQ) that do not participate in extreme sports compared to those who also score high on the ZKPQ, but engaged in extreme sports; they suggest that the extreme sport practitioners might be the true arousal seekers. Furthermore, Slanger & Rudestam (1997) suggested that those who engage in risky snow sports may be too closely clustered together in the high end of the SSS scale dimensions, and that this  88  global sensation seeking scale may lack the resolution to discriminate among individuals scoring in the very high end of the continuum, e.g., extremely high-risk sports (such as extreme skiing which involves cliff jumping into unknown terrain), and the moderately high-risk versions of the same sport (skiing in controlled terrain). The CSSQ-S represents a facet of Zuckerman’s sensation-seeking trait more closely related to risk-taking behaviours in downhill sports, and the use of both scales may improve the characterization of respondents. Another context-specific sensation-seeking measure created by Kalichman and colleagues (1994) to supplement general sensation-seeking scales relates more specifically to sexual sensation seeking, and has been used in numerous studies that investigate HIV and risky sexual behaviours. Narrow trait instruments can have a predictive advantage over broad traits instruments (Paunonen & Ashton, 2001). Many self-report questionnaires inquire about behaviours that people would like to do, but these may not be aligned with what the individuals actually do. Although the studies presented in this chapter did not measure actual behaviours by field observation, the CSSQ-S was designed to inquire about actual, rather than hypothetical behaviours. Studies have shown that hypothetical behaviours are often exaggerated compared to actual behaviours (e.g., Alpizar, Carlsson, & Johansson-Stenman, 2008). The CSSQ-S, which measures sport-specific, recent patterns of behavioural tendencies, explained significantly more of the variance in injury rates than the ImpSS subscale, a broader, hypothetical trait-measure. Similarly, a study that explored the relationship between optimism and risk perception in a sport population found that there was no relationship between risk perception measured within a specific context and optimism as a general personality trait (even though links between these two traits have been established in occupational and health psychology literatures (e.g., Fontaine, 1994)). The authors suggested  89  that domain- and population-specific psychometric measures are more appropriate for measuring risk-taking in high-risk sports (Martha, Sanchez, & Goma-i-Freixanet, 2009). While domain- or context-specific instruments might explain more variance in a given criterion, there are limitations to narrow psychometric instruments. For example, a limitation of a context-specific questionnaire like the CSSQ-S that inquires about actual behaviours is that it is to some extent age- and ability-influenced. The sample was restricted to include young- to middle-aged adults based on Zuckerman’s observation that sensation seeking declines after 40 years of age, and although age did not significantly correlate with CSSQ-S score in this study, there might be a decline in sensation seeking that occurs at an earlier age when measured in a sport context. A majority of CSSQ-S items describe behaviours that involve taking physical risks and may be influenced by a person’s physical fitness, which generally declines with age (Sallis, 2000). A minimum ability level was also imposed for study inclusion, but there was still a significant positive correlation between proficiency and CSSQ-S scores. Whether a skier is more likely to improve if he/she is high in sensation seeking and willing to take risks, or whether mastery of the sport makes a skier more likely to exhibit risky or sensation-seeking behaviours is not known, but links between ability and risk-taking have also been observed in other high-risk sports (D.J. Llewellyn et al., 2008). Future longitudinal studies might be useful to examine the association between ability levels and sensation seeking and also consider whether other variables such as physical fitness and/or previous injuries might be associated with sensation seeking in skiing and snowboarding. Establishing validity is an ongoing process, not limited to one or two studies. With this in mind, future research may include an extension of the CSSQ-S for use in other high-risk sports studies, such as those investigating injury in mountain biking or river kayaking. Both are  90  “downhill sports” that share terrain characteristics with skiing and snowboarding (e.g., varying chute-width and slope-grade affecting the thrill and risk of the activity) and items might be generalized to examine the applicability of the CSSQ to such other sports. While the CSSQ may be transferable to downhill sports, its generalizability to other sports is limited due to its narrow scope. “Gravity sports” for example, which include sky-diving, BASE jumping, and speedflying, would require entirely different items to operationalize sensation-seeking facets that pertain to novelty, thrill, and risk within the sport-context. As the popularity of downhill and adventure sports continue to increase (Hudson, 2004), recreation sites create interesting natural settings for observing sensation-seeking behaviours in the field. Furthermore, high-risk sports athletes consistently report higher sensation seeking than controls, and may ultimately represent a homogeneous group in which to study extreme manifestations of the sensation-seeking trait. The development of the CSSQ-S, a focused trait instrument that measures patterns of sensation seeking in skiing and snowboarding, may be useful in future research to investigate relationships between sport experience, ability, risktaking, injury, and a variety of psychological constructs. The CSSQ-S is used in the next chapters to explore associations between sensation seeking and dopaminergic genetic variants.  91  Chapter 4: The -521 C/T polymorphism in the dopamine-4-receptor gene (DRD4) is associated with skiing and snowboarding behaviour15  4.1  Summary A single nucleotide polymorphism, -521 C/T (rs1800955) in the promoter region of the  dopamine-4-receptor gene (DRD4), is associated with approach-related traits including novelty seeking and extraversion, in some, but not all studies. Using a joint-analysis approach, sensation seeking was measured in two cohorts of experienced male and female skiers and snowboarders (n = 503) using a sports-specific tool, the Contextual Sensation Seeking Questionnaire for Skiing and Snowboarding (CSSQ-S, Chapter 3), and a more general trait measure, the Zuckerman Kuhlman Personality Questionnaire (ZKPQ) impulsive sensation seeking subscale. A significant association between the DRD4 -521CC genotype and sports-specific sensation seeking as measured using the CSSQ-S was detected and replicated (p < .001). These data suggest that the DRD4 -521 C/T polymorphism contributes to a “sensation-seeking phenotype” in skiers and snowboarders, but the variant was not associated with impulsive sensation seeking (p = .9).  4.2  Introduction Dopamine has been implicated in behavioural activation, instrumental learning (involving  both positive and aversive motivations), appetitive approach, and emotional processing (Lauzon & Laviolette, 2010; Salamone et al., 2007; Wise, 2004), and is thought to contribute to the neurobiological basis for impulsive sensation seeking (Zuckerman, 2005b). The intensity of the  15  Data in this chapter was published as, “The -521 C/T polymorphism in the dopamine-4-receptor gene (DRD4) is associated with skiing and snowboarding behaviour” in the Scandinavian Journal of Medicine and Science in Sport, March 2013. 92  feeling of pleasure immediately before and during a risky situation (e.g., the “high” that a skydiver might feel during a jump (I. H. Franken et al., 2006)) is thought to be related to dopamine levels (Di Chiara et al., 2004), levels that may be affected by the density and/or function of dopamine receptors (Blum et al., 2009). More details about the purported role of dopamine in sensation seeking are reviewed in Chapter 1. The D4 receptor is encoded by the gene, DRD4, and is located on chromosome 11 (11p15.5). There are numerous polymorphisms of the 3.4 KB DRD4 in human populations (Okuyama et al., 2000); including the commonly studied 48 base-pair variable number tandem repeat (VNTR) in exon III and -521 C/T (a cytosine (C) to thymine (T) transition at base -521 in the upstream promoter region). The -521C allele is associated with a 40% increase in DRD4 transcription in cultured cells (Okuyama et al., 2000), a phenotype, which if expressed in vivo, could have an impact on neuromodulation (Cordell & Clayton, 2005) and the emotional processing of stimuli (Lauzon & Laviolette, 2010). A number of studies have investigated the impact of DRD4 variants (reviewed in Chapter 1, and Table 4-1 below) on approach-related traits. The results of these studies have been inconsistent (see reviews: Kluger, Siegfried, & Ebstein, 2002; Schinka, Letsch, & Crawford, 2002), although two meta-analyses found the -521 C/T to be a more promising candidate DRD4 polymorphism for approach-related trait studies than the VNTR (M. R. Munafo et al., 2008; Schinka et al., 2002).  93  Table 4-1 Genetic association studies on -521C/T and approach traits Genetic association studies on -521 C/T and approach traits or externalizing disorders Cohort Young male (Japanese)  N 86  Measure TCI  Extreme scorers (high and low) (European, Finland)  200  TCI  Schizophrenics and controls (Japanese) Students (European, Hungary)  173 99  TCI TCI  Healthy subjects (African-American)  71  Healthy subjects (European, Swedish) Community sample, including 95 children and their parents (European Hungary) Healthy subjects (European, German) Clinically depressed patients (European, New Zealand)  381 95  NEOFFI TCI N/A  276 156  TCI TCI and SCID  Adolescent females (Korean)  101  TCI  Schizophrenic patients (210) and healthy controls (206) (Han Chinese)  416  N/A  Healthy subjects (European, German)  104  Healthy subjects (Russian)  220  NEOFFI EPI  6 year old boys and girls (European, Hungary)  57  ERPa  Healthy subjects (Russian)  130  TCI, EPI  Schizophrenic patients (216) and healthy controls (243) (Japanese)  459  N/A  Findings CC highest NS scores, TT lowest, p < .001 (no mechanisms proposed) n.s. n.s. CC higher NS than CT and TT, especially in women, p = .008 (no mechanisms proposed) CC highest scores on extraversion compared to CT + TT (no mechanisms proposed) n.s. There was a significant interaction between -521C/T and the exon III VNTR associated with disorganized attachment n.s. TT associated with avoidant and obsessive symptoms. No association between -521C/T and TCI NS (no mechanisms proposed) No association for -521 C/T, but a significant interaction between -521C/T and exon III VNTR and NS (no mechanisms proposed) No association for -521C/T between groups, but significant difference in a haplotype containing -521C/T between groups (no mechanisms proposed) n.s. A joint contribution of -521 C/T and -809 A/G to levels of extraversion (no mechanisms proposed) Analyzed -521C/T only as haplotype with exon III VNTR: T-7 haplotype associated with “resistance to distraction” -521 C allele associated with extraversion and NS in women homozygous for the COMT Met allele (no mechanisms proposed) -521C allele frequency was marginally higher in females with schizophrenia (no mechanisms proposed)  Reference (Okuyama et al., 2000) (Ekelund, Suhonen, Jarvelin, Peltonen, & Lichtermann, 2001) (Mitsuyasu et al., 2001) (Ronai et al., 2001) (Bookman, Taylor, AdamsCampbell, & Kittles, 2002) (Jonsson et al., 2002) (Lakatos et al., 2002) (Strobel et al., 2002) (Joyce et al., 2003) (H. J. Lee et al., 2003) (Xing et al., 2003) (Eichhammer et al., 2005) (Golimbet, Gritsenko, Alfimova, & Ebstein, 2005) (Birkas et al., 2006) (Golimbet, Alfimova, Gritsenko, & Ebstein, 2007) (Mitsuyasu et al., 2007)  94  Cohort Extreme scorers (117 high and 192 low) (European, English) Heroin dependent (84) and controls (168) (Chinese males) Heroin users + healthy controls (Chinese)  N 309  600  Measure EPI  Findings n.s.  Reference (M. R. Munafo et al., 2008)  N/A  T allele frequency higher in heroin-dependent group  (Ho, Tang, Cheung, & Stadlin, 2008) (Lai et al., 2010)  n/a  -521 T allele frequency higher in heroin users (no mechanisms proposed) Community sample (99) (European, USA) two- N/A TT overrepresented in patients with borderline and (Nemoda et al., 2010) Clinical sample (136) (European, Hungary) stage antisocial symptoms in US sample, but finding not replicated (no mechanisms proposed) Note. n.s. = not significant, NS = novelty seeking, TCI = Temperament and Character Inventory (Cloninger), EPI = Eyesenck Personality Inventory, NEO-FFI = NEO Five-Factor model (Costa & McCrae), ERP = novelty elicited event-related potentials, SCID = structured clinical interview for DSM-III-R (Diagnostic and Statistical Manual of Mental Disorders). Most of the studies reporting significant findings do not propose a physiological mechanism and the few studies that do propose a mechanism looked at the -521 C/T in combination with other variants or report findings in the opposite direction to the Munafo et al. (2008) and Schinka et al., (2002) meta-analyses.  95  Previous associations have been reported between alleles at -521 C/T and novelty seeking, extraversion, and drug use (Lai et al., 2010; M. R. Munafo et al., 2008) (Table 4-1); however, at the time when data collection for the current chapter was underway, researchers had yet to examine the alleles in populations actively involved in seeking out and experiencing physical risks in sports. To address this gap, moderately- to highly-skilled skiers and snowboarders were recruited at mountain resorts in Western Canada to study whether there were associations between sensation seeking in snow sports and the DRD4 -521C/T. These sports were chosen because 1) there is a wide range of established risk levels inherent in the sports (i.e., runs of varying degrees of danger), 2) there is a large pool of potential subjects, heterogeneous in their tolerance for risk (from the “cautious” to the “high risk-taker”), and 3) because tools that quantify patterns of sensation-seeking behaviours in a sport context can be used alongside more general trait instruments for genetic studies. Athletes’ global sensation-seeking behaviour was assessed using the Zuckerman Kuhlman Personality Questionnaire (ZKPQ) impulsive sensation seeking subscale (ImpSS), while their context-specific (skiing/snowboarding) sensation-seeking behaviour was evaluated using the Contextual Sensation Seeking Questionnaire for Skiing and Snowboarding (CSSQ-S) which was developed and tested for this study (see Chapter 3).  4.3 4.3.1  Methods Participants Skiers and snowboarders between 17 and 49 years of age (joint-sample: n = 503: 287  males, 216 females; mean age = 26.8 years (SD = 6.0)) of intermediate ability or better  96  participated in this two-stage study. A pilot sample16 (n = 117) was recruited from Whistler and Vancouver, British Columbia, and Lake Louise and Banff, Alberta. A second sample17 (n = 386) was recruited at the 2010 Telus World Ski and Snowboard Festival in Whistler, British Columbia. Skiing and snowboarding ability was self-assessed by participants and was based on ability to ski/ride runs classified as a “blue square” (the standard North American notation for “mid-level”) or harder. For details about Festival recruitment, see Chapter 2. To minimize confounding effects of differing biogeographical backgrounds on allele frequencies, only participants self-reporting as “White/European descent” were included in the genetic analysis. A majority of participants (67%) had completed a post-secondary degree or diploma, 30% were in their final two years, or had completed secondary school (details Table 4-2).  Table 4-2 Participant characteristics Participant characteristics Demographic Variable Age (Mean (SD)) Sex Ability  Pilot, n = 117  Festival, n = 386  27.1 (4.5) years 26.3 (5.9) years 50% male, 50% female 58% male, 42% female 28% intermediate, 36% advanced, 23% intermediate, 32% advanced, 45% 37% expert expert Sport 65% skier, 35% snowboarder 54% skiers, 40% snowboarder, 6% both Education 27% high school, 60% post30% high school, 67% post-secondary, secondary, 3% missing 3% missing Note. SD = standard deviation. There were no significant differences in any of the demographic variables between samples (p > .13).  16 17  Pilot sample is from MSc sample. A portion of the “Festival sample” that was successfully genotyped. 97  4.3.2  Measures Impulsive sensation seeking. Global sensation seeking was assessed using the  impulsive sensation seeking scale (ImpSS) from the ZKPQ (Zuckerman et al., 1993). In this study, scores derived from the ImpSS demonstrated acceptable internal consistency (Cronbach alpha = .82). The Pilot sample completed the full ZKPQ, which contains five subscales plus an infrequency scale, which was included to support data screening. The second sample (Festival sample); however, only completed the ImpSS subscale (with an infrequency scale) due to its hypothesized relationship with skiing and snowboarding behaviour (Chapter 3). Contextual Sensation Seeking Scale for Skiing (CSSQ-S). Sensation seeking in the specific context of skiing and snowboarding was assessed using the 10-item CSSQ-S (questionnaire development is described in Chapter 3). In this study, CSSQ-S scores were significantly correlated with ImpSS scores, r(480) = .46, p < .01, and scores derived from the CSSQ-S demonstrated high internal consistency (Cronbach alpha = .86).  4.3.3  Genotyping Buccal cell DNA was isolated from cytobrushes (Fisher Scientific, Ottawa, ON, Canada)  using a standard alcohol purification technique (Saftlas et al., 2004) (Appendix L) and genotyped for the -521C/T polymorphism (dbSNP rs1800955). For the Pilot sample, DNA was genotyped initially using polymerase chain reaction (PCR) restriction fragment length polymorphism (RFLP) analysis; however, the assay was difficult to optimize and subsequently a more rapid pyrosequencing strategy was employed. The majority of the genotypes were ascertained using pyrosequencing.  98  RFLP-based genotyping. Primers used were DRD4-F: 5’-GAT CAA CTG TGC AAC GGG TG-3’ and DRD4-R: 5’-GAG AAA CCG ACA AGG ATG GAG-3’ (NAPS IDT, Vancouver, BC, Canada). The PCR cycled in a MJ Mini Cycler (Bio-Rad Laboratories, Hercules, CA, USA) as follows: 95°C for 5 minutes, followed by 39 cycles of 95°C for 45s, 60°C for 45s, and 72°C for 2 minutes. The 25 µL reactions contained 20 mM Tris-HCl pH 8.4, 50 mM KCl, 1.8 mM MgCl2, 0.2 mM dNTP, 0.6 µM of each primer, 10% BSA, 1.0 U Taq polymerase (Invitrogen Corporation, Carlsbad, CA, USA), and 10-20 ng DNA template. PCR products were digested using FspI (recognition sequence: 5’ TGC!GCA 3’) restriction endonuclease (New England Biolabs, Beverly, MA, USA) at 37°C overnight, and genotypes were  visualized using 8% polyacrylamide gel electrophoresis stained with Sybr Safe gel stain (Life Technologies, Burlington, ON, Canada) (see Figure 4-1 for a sample gel photograph).  Figure 4-1. Sample gel electrophoresis photograph. Genotypes from left to right: -521 CC, -521 TT, -521 CT, 100-bp marker. Samples were run for approximately 45 minutes at 125 volts.  Pyrosequencing-based genotyping. Primers were designed using the Biotage AB PSQ Assay Design Software (Version 1.0.6, USA; DNA sequence shown in Appendix R): forward  99  primer: 5’-TAG GCG TCG GCG GTT GAG-3’; reverse primer: 5’-GAC TCG CCT CGA CCT CGT G-3’; and sequencing primer, 5’ TCG GGG GCA GGG GGA 3’ (the reverse primer was biotinylated and HPLC purified, NAPS IDT, Vancouver, BC, Canada). DNA was PCR amplified in 15 µL reactions containing 20 mM Tris-HCl pH 8.4, 50 mM KCl, 1.5 mM MgCl2, 0.2 mM dNTP, 0.4 µM of each primer, 1.0 U native Taq polymerase (Invitrogen), and 1.0 µL DNA (approximately 20 ng/µL). Cycling conditions were: 95°C for 5 minutes, followed by 45 cycles 95°C for 20s, 60°C for 20s, and 72°C for 20s, followed by a 5 minute final extension at 72°C. Single-stranded biotinylated PCR products were prepared as per manufacturers recommendations for sequencing using the Pyrosequencing Vacuum Prep Tool (Biotage AB, Uppsala, Sweden). The following sequence was analyzed by the PyroQ SNP software (Biotage) to determine sample genotypes: 5’GCGGGCGNGGAGGGYG 3’ (see Figure 4-2 for a sample PyroQ result; see Appendix R for additional result.  Figure 4-2. Sample PyroQ result. Far right portion of the image shows a peak at “T” and “C”, indicative of a -521 CT genotype.  100  4.3.4  Statistical analyses CSSQ-S and ImpSS data were screened for normality, missing data, and univariate  outliers. An analysis of co-variance (ANCOVA) was used for each dependent variable (ImpSS and CSSQ-S) to compare sensation-seeking scores between genotype groups (CC vs. CT vs. TT) in stage 1 (Pilot sample). As previous studies have shown a significant difference between sensation-seeking scores in males and females (males generally scoring higher (e.g., Zuckerman & Kuhlman, 2000)), sex was included as a covariate when there were significant differences in scores. Ability was included as a covariate for the CSSQ-S analysis since the sport-specific tool and ability were linearly correlated in Chapter 3. Dummy variables were used for sex (male = 0, female = 1) and ability (0 = beginner, 1 = novice, 2 = intermediate, 3 = advanced, 4 = expert); when ability was < 2 the CSSQ-S score was rejected. The CC genotype has been associated with approach behaviour (Bookman et al., 2002; Ronai et al., 2001); therefore, higher sensationseeking scores were expected in the -521 CC genotype group. The same statistical methods described in stage 1 were employed in stage 2 (the replication sample, n = 386), comparing the ImpSS and CSSQ-S scores across genotypes (CC vs. CT vs. TT) using a one-way ANCOVA. Fisher’s method for combining independent probabilities (Fisher, 1932) was used to combine p-values from the pilot and replication samples (similar to joint-analysis design suggested by Skol, Scott, Abecasis, & Boehnke, 2006). Details about Fisher’s method are shown in Appendix S. Joint-analysis is a form of internal replication that results in an increased power (compared to independent replication) to detect associations (Skol et al., 2006). Finally, the association was analyzed in the combined sample (pilot + replication, n = 503). In both stages, continuous variables (ImpSS and CSSQ-S scores) were analyzed by genotype, and therefore a control group is not needed. Similar single-cohort  101  research designs are common to behavioural genetic studies (Bookman et al., 2002; Okuyama et al., 2000; Ronai et al., 2001).  4.4 4.4.1  Results Stage 1 Genotype frequencies of the 117 participants from stage 1 (.33 CC, .41 CT, .26 TT) were  in Hardy-Weinberg Equilibrium (HWE), !2 = 3.57, p > .05. The allele frequencies in the participants were 54% -521C and 46% -521T. To confirm that the genotyping results would be consistent regardless of the genotyping technology used, a random sample of DNAs that had been genotyped using RFLP was genotyped again using pyrosequencing. There was 100% agreement between the results obtained by the two methods. The ImpSS and CSSQ-S data were normally distributed and satisfied univariate assumptions. Two subjects were excluded either due to not meeting ability for CSSQ-S, or high infrequency scores on ZKPQ. ANCOVAs revealed that the -521 C/T polymorphism had a significant effect on both sensation-seeking measures: ImpSS (F(2, 115) = 3.63, p = .03), and CSSQ-S adjusted for sex and ability (F(2,116) = 8.73, p < .001, Table 4-3). CSSQ-S scores were linearly correlated with ability in the Pilot sample, r(116) = .66, p < .001) and males scored significantly higher than females on the CSSQS (p < .001), but there were no significant differences on ImpSS scores between the sexes (p = .20, see Table 4-4). There were no significant correlations between age and sensation-seeking measures (p > .4).  102  Table 4-3 Stage 1: A summary of sensation-seeking scores by genotype Stage 1: A summary of sensation-seeking scores by genotype Genotype  ImpSS  CSSQ-S  N  Mean  SD  Range  n  Mean  SD  Range  CC  38  12.97  3.84  2-19  38  36.35  6.60  21-46  CT  49  10.65  4.43  2-19  48  33.10  7.64  18-45  TT  29  11.45  3.41  4-18  30  33.67  6.22  23-45  p=  .03  p < .001  Note. p values for univariate ANOVA are shown in bold. Sex and ability were included as covariates for CSSQ-S. Genotype group totals vary due to questionnaire exclusions (total n = 117). SD = standard deviation.  4.4.2  Stage 2 Genotype data. In the replication sample (stage 2), genotypes at the -521 C/T SNP were  established for 376 subjects; however, four subjects were excluded from subsequent association analyses due to missing data. Further subject exclusions specific to each questionnaire are described below. The genotype frequencies (.21 CC, .48 CT, .30 TT) showed no significant deviation from those predicted if the population was in Hardy-Weinberg equilibrium (!2 = 0.13, p > .05) and the frequencies did not differ between stage 1 and 2 (!2 = 5.71, p > .05). Genotypes did not differ on any other demographic variables (e.g., sex, age, education, marital status, p > .5). Questionnaire data. The questionnaire data were normally distributed and variances for both personality and ski data sets were homogeneous (Levene’s statistic, p > .05). Fifteen CSSQ-S scores were removed from stage 2 analysis for not meeting eligibility requirements (intermediate ski/snowboard ability), and three ZKPQ ImpSS scores were rejected due to high infrequency (social desirability scale) scores as per ZKPQ criteria (Zuckerman et al., 1993). The mean ImpSS score (M = 12.45, SD = 3.84, n = 487) for skiers and snowboarders (total sample, 103  males and females combined) was significantly greater than “norms” described elsewhere (M= 10.18, SD = 4.10, N = 2969 (Zuckerman et al., 1993); t(3454) = 11.42, p < .0001). Males scored significantly higher than females on both the ImpSS (p < .05) and the CSSQ-S scale (p < .001, Table 4-4). CSSQ-S scores were linearly correlated with ability in both the replication sample (r(362) = .57, p < .001) and the combined sample (r(478) = .59, p < .001). The relationships between sensation-seeking scores (ImpSS and CSSQ-S) and other demographic variables (e.g., marital status, age, education) were either not significant (p > .05) or only modestly correlated with small effect sizes (r < .2) and were not suitable for inclusion as covariates (Cohen, 1992).  Table 4-4 Differences in sensation-seeking scores between males and females Differences in sensation-seeking scores between males and females Pilot Sample  Replication Sample  Measure  Sex  n  Mean  SD  n  Mean  SD  ImpSS  Male  59  12.08  3.30  219  13.14  3.27  Female  57  11.12  4.76  156  12.09  4.25  a  t, p CSSQ  1.27 ,  .20  a  2.59 ,  .010 5.92  Male  59  37.92  5.42  219  39.58  Female  57  30.59  6.61  148  32.02  t, p  a  6.51 ,  <.001  7.11 a  11.05 ,  <.001  a  Note. Equality of variances between males and females was violated, therefore an alternate test that does not assume equal variances was employed. SD = standard deviation.  Gene associations. Stage 1 resulted in significant associations with both sensationseeking measures (ImpSS and CSSQ-S); therefore both variables were analyzed in stage 2. In the replication sample, analysis of variance revealed a significant relationship between genotype at the -521 C/T polymorphism and ski/snowboarding behaviours (CSSQ-S scores). The -521 C/T polymorphism was significantly associated with sensation seeking in skiing when tested using an additive model with sex and ability as covariates (n = 359, F(2, 354) = 3.85, p < .05, !p2  104  = .021, Table 4-5). The joint-p-value, obtained using Fisher’s method for combining probabilities between the pilot and replication samples was also significant, p < .0001. Finally, an ANCOVA on the combined sample revealed a significant association between -521 C/T and CSSQ-S score (n = 475, F(2, 470) = 7.93, p < .001, !p2 = .033, Table 4-5). Pair-wise comparisons (using least significant difference, LSD) between genotype groups revealed that individuals with the CC genotype scored higher than CT individuals (LSD = 2.37± 0.61, p < .001) and TT individuals (LSD = 2.09 ± 0.68, p = .002). No significant differences between genotype groups and ImpSS scores were observed in the replication sample (n = 372, F(2, 368) = 0.08, p = .92; adjusted for sex only). The relationship between genotype and CSSQ-S remained significant when a correction for testing two variables (Bonferroni, p = "/2 (Attia et al., 2009)) was applied.  Table 4-5 Stage 2: A summary of contextual sensation seeking (CSSQ-S) scores by genotype Stage 2: A summary of contextual sensation seeking (CSSQ-S) scores by genotype Replication Sample  Combined Sample  Genotype  n  Mean  SD  n  Mean  SD  CC  77  37.84  7.62  116  37.33  7.28  CT TT  173 107  36.13  a  36.64  a  p < .05  6.89 7.76  222 137  35.51  b  7.16  35.99  b  7.53  p < .001  Note. p-values for univariate analysis are shown in bold. Sex and ability were included as covariates. Means from respective samples that have no superscript in common are significantly different from each other (LSD p < .01). SD = standard deviation.  4.5  Discussion There was a significant association between the DRD4 -521 C/T polymorphism and  patterns of skiing/snowboarding behaviours as assessed using the CSSQ-S. There was an  105  association in a small sample of skiers and snowboarders, which was replicated in a larger independent sample recruited at a later date. The combined sample of skiers and snowboarders had higher ImpSS scores compared to other populations, which supports previous findings that high-sensation seekers are often involved in high-risk activities (reviewed in Goma-i-Freixanet et al., 2012); however, the ImpSS scores were not associated with the -521 C/T genotype in the replication sample. This study presented in this chapter is the first to investigate the genetics of sensation seeking in a high-risk sport population using a tool specific to the population, and to my knowledge it is also the first to examine the relationship between the -521C/T DRD4 polymorphism and sensation seeking (previous studies have investigated other approach traits such as novelty seeking and extraversion). One other sports study investigated genetic variants (DRD4 VNTR, DAT1 VNTR, and HTR2A 102T/C) and “Big Five” personality traits in high- and low- risk sport participants, but found no differences between groups (Cam et al., 2010). There was no association using the ZKPQ measure of impulsive sensation seeking in the current chapter, but the result of higher ski/snowboard-specific sensation seeking scores (CSSQ-S) in individuals with -521 CC genotypes is consistent with other studies on novelty seeking (Golimbet et al., 2005; Okuyama et al., 2000; Ronai et al., 2001). Studies that have investigated associations between the -521 C/T polymorphism and approach-related traits have found femalespecific associations (Bookman et al., 2002; Ronai et al., 2001), while others have observed the association in all-male cohorts (Okuyama et al., 2000). Sex was included as a covariate since sensation-seeking scores differ significantly between males and females, and there was a significant relationship between -521 C/T polymorphism and patterns of ski behaviours (p < .001) that survived correction for multiple tests.  106  Results from previous studies of DRD4 polymorphisms and novelty seeking have been inconsistent (M. R. Munafo et al., 2008), which may be due to broad phenotyping by general, trait measures. To address this, sensation seeking was measured using two scales: a global, traitscale and a context-specific sensation-seeking scale. Many trait questionnaires inquire about responses to hypothetical situations, but these may not be aligned with what the individuals actually do. Although ski behaviours were not observed in the field, the CSSQ-S was designed to inquire about actual, rather than hypothetical behaviours (Chapter 3). This focused phenotyping impacted the sample size (e.g., exclusion of less proficient skiers/snowboarders), but the cost of losing power due to smaller sample sizes can be offset by the gain of generating fewer false-positives or -negatives (Kreek et al., 2005). A note of caution, when interpreting data obtained from highly specific cohorts such as this skiing/snowboarding cohort: while there are benefits to sampling from a homogeneous population (e.g., reducing extraneous variability) there is a possibility of sampling bias and the results have limited generalizability. It should also be noted that no psychiatric screening was carried out, and that future studies would benefit from including additional screening. A role for DRD4 in approach-related behaviours has been supported in neurophysiology and genetic studies, but the effect sizes of complex phenotypes (e.g., personality traits) are usually small (Ebstein, 2006). While there was a significant association between genotype at 521 and patterns of skiing behaviour in the study described in this chapter, the locus accounted for only approximately 3% of the variance18 in contextual sensation seeking ski scores in males and females. Sensation seeking has moderate heritability, with over 50% of the trait-variance  !"  #Based on ANCOVA partial eta (!p2), a measure of effect size. 107  due to genetics (Stoel et al., 2006), and is likely a polygenic trait with the inter-individual variance influenced by genotypes at multiple loci (Golimbet et al., 2007). The -521 C/T polymorphism is in a known regulatory region of the DRD4 gene and has established phenotypic consequences (Okuyama et al., 2000); however, there may be other functional variants elsewhere in the gene (or in other genes involved in dopamine pathways) that could also affect sensation seeking. Other genes that have been previously investigated in association with approach-related traits include those encoding enzymes that metabolize, transport, or regulate dopamine (e.g., monoamine-oxidase B (MAO-B), catechol-O-methyl transferase (COMT), dopamine-!-hydroxylase (DBH), dopamine transporter (SLC6A3), and other dopamine receptor genes (e.g., dopamine receptors (DRD1, DRD2) (Ebstein & Israel, 2009; M. R. Munafo et al., 2003; Roe et al., 2009; Varga et al., 2012). As well as the serotonin receptor gene (HTR2A) and the gene encoding stathmin (STMN1), which regulates microtubule formation, both have been associated with harm avoidance and fear-related behaviours (Brocke et al., 2010; Ebstein & Israel, 2009), making all of these genes potential candidates in sensation seeking in sport. Data from analyses including variants from the above-mentioned genes are presented in the following chapters. In summary, genotyping revealed a significant association between alleles at a functional polymorphism in the DRD4 gene and sensation seeking in a sports context in male and female skiers and snowboarders. This observation raises the intriguing possibility that the predilection to risk-taking behaviour in sports is influenced by genetics.  108  Chapter 5: Association of a common D3 dopamine receptor gene variant is associated with sensation seeking in skiers and snowboarders 19  5.1  Summary The “sensation seeking” trait has been associated with risky behaviours including high-  risk sport participation. Genes involved in dopaminergic neurotransmission have been investigated in studies of approach traits; however, not in sporting contexts. Using a jointanalysis the relationships between monoamine-neurotransmitter gene variants and impulsive and sport-specific sensation seeking were investigated in skiers and snowboarders from Western Canada (n = 599). Twenty-six SNPs in eight genes involved in dopaminergic and serotonergic transmission (DRD1, DRD2, DRD3, DAT1, COMT, MAO-B, DBH, HTR2A) were genotyped. An initial screen identified a few variants that were associated with sensation seeking, one of which, a G/A transition (rs167771) in the D3-receptor gene (DRD3), remained significant in the combined sample and after correction for multiple testing (p = .004, !p2 = .02). DRD3 variants have been associated with approach traits; however, the results presented in this chapter are the first to suggest a role for rs167771 in sensation seeking.  5.2  Introduction Many genes have been explored in association with novelty and sensation seeking, and  there are likely a number of genes contributing to this polygenic trait (Ebstein, 2006). Increased reward-seeking behaviour (an animal phenotype similar to sensation seeking) in  19  The data in this chapter was published as: “Association of a common D3 dopamine receptor gene variant is associated with sensation seeking in skiers and snowboarders” in the Journal of Research in Personality, 2013. 109  hyperdopaminergic mice (Pecina et al., 2003), and correlations between serotonin levels and impulsive behaviour in humans (C.S. Carver & Miller, 2006) suggest that genes in the dopaminergic and/or serotonergic pathways could contribute to this trait. Variants within candidate genes that encode proteins involved in dopaminergic and serotonergic neurotransmission are extensively reviewed in Chapter 1. There have been only a handful of studies that have investigated genetic associations with the sensation-seeking trait (specifically). Most genetic studies have characterized approach phenotypes using the TCI novelty seeking subscale; however, Derringer et al. (2010) similarly employed a candidate gene association study design to test for associations with sensation seeking (measured using the SSS) by investigating 273 SNPs in eight candidate genes in a sample of alcoholics. They reported significant associations with 12 SNPs; however, a commentary on the study noted that the authors failed to correct for the number of SNPs tested and the findings were not significant after correction (Powell & Zietsch, 2011). To investigate the possibility that variants in genes that are involved in dopaminergic and serotonergic neurotransmission affect sensation seeking, a two-stage joint-analysis (Skol et al., 2006) candidate-gene association study was carried out in a cohort of proficient skiers and snowboarders. Polymorphic loci at which both alleles were moderately common (heterozygosity > .2) were chosen in genes that encode proteins involved in transport (dopamine transporter: DAT1 (SLC6A3)), function (dopamine and serotonin receptors: DRD1, DRD2, DRD3, HTR2A), and metabolic inactivation (catechol-O-methyltransferase, (COMT), monoamine oxidase B (MAO-B), and dopamine-!-hydroxylase (DBH)). Initially (stage 1), 26 single nucleotide polymorphisms (SNPs) in the aforementioned genes (see Table 5-1) were assessed for associations with sensation-seeking phenotypes in a sub-sample of participants randomly chosen  110  from the full sample. SNPs significantly (p < .05) associated with any of the sensation-seeking phenotypes in this stage were subsequently tested in the remainder of the sample (stage 2). The results from both stages were combined and corrected for multiple testing. Each athlete’s global sensation seeking was measured using the (ZKPQ; Zuckerman et al., 1993) impulsive sensation seeking (ImpSS) scale, which can be analyzed as a single subscale or as two constructs: impulsivity and sensation seeking (Zuckerman et al., 1993). Athlete’s sport-specific sensation seeking was measured using the Contextual Sensation Seeking Questionnaire for Skiing and Snowboarding (CSSQ-S), a tool that inquires about patterns of actual behaviours that relate to sensation seeking in sport (see Chapter 3 for details about CSSQS development). Many high-risk sport participants actively seek out risks and have higher than average sensation-seeking scores (Goma-i-Freixanet et al., 2012). Numerous studies have explored genetic associations with approach-related behaviours and traits in deviant risk-taking populations (e.g., substance users, alcoholics, gamblers) or in populations exhibiting externalizing disorders (Gorwood et al., 2001; Ting Li et al., 2011; Lobo et al., 2010; Nemoda et al., 2010); however, practitioners of high-risk sports may represent a population that seeks stimulation though less deviant, more health-benefiting outlets.  111  Table 5-1 List of single nucleotide polymorphisms chosen for analysis List of single nucleotide polymorphisms chosen for analysis Gene (abbreviated and full name) COMT Catechol-O-methyl transferase  Physiological function Metabolizes DA  SNP rs737865a, rs4633, rs4680, rs165599b  DAT1 or SLC6A3  Dopamine transporter  Reuptake of DA from the synapse  rs6347, rs27072, rs463379, rs2937639, rs2975226b  DBH  Dopamine B-hydroxylase  Converts DA to norepinephrine  rs1611115  DRD1  Dopamine receptor D1  DRD2  Dopamine receptor D2  Involved in dopaminergic neurotransmission Involved in dopaminergic neurotransmission  DRD3  Dopamine receptor D3  rs686, rs4532, rs251937, rs4867798 rs6277, rs1076560, rs1079597, rs1800497, rs1800498, rs2283265, rs2734831, rs4245147, rs7131056, rs17601612 rs6280, rs167771  MAO-B  Monoamine oxidase B  HTR2A  Serotonin receptor  Involved in dopaminergic neurotransmission Metabolizes DA  rs1799836  Involved in serotonergic rs6311 neurotransmission Note. Heterozygosity of all SNPs > .2 (ALFRED database). DA = dopamine, 5HT = serotonin. a  lower call rate (marker performance 88%).  b  Two SNPs failed optimization and were not genotyped.  5.3 5.3.1  Methods Participants Skiers and snowboarders were recruited at the 2010 Telus World Ski and Snowboard  Festival in Whistler, British Columbia, Canada (details in Chapter 2). Participants completed a set of brief questionnaires and provided a buccal (cheek) cell swab for DNA preparation. A majority of the initial sample (89% of n = 668) self-reported as “White/European descent”; therefore, to avoid the potential confounding effect of inter-population variation in allele frequencies all non-Caucasian participants were excluded from analyses. After exclusions, the genotyped sample included 599 skiers and snowboarders, ages 17 to 49 years (341 male, M = 112  26.91 years, SD = 6.81; 258 female, M = 27.41 years, SD = 5.93)20, and all were at least intermediate in skiing or snowboarding ability.  5.3.2  Measures ZKPQ Impulsive Sensation Seeking (ImpSS). Participants completed the 19-item  subscale scored true/false (Appendix D) (Zuckerman et al., 1993). In the current study, the Cronbach alphas for the combined sample (n = 599) were .80, .73, and .68 for ImpSS, Imp, and SS, respectively. Contextual Sensation Seeking Questionnaire for Skiing and Snowboarding (CSSQS). Participants completed the 10-item CSSQ-S, scored on a five-point Likert scale (Appendix F). In this study the internal consistency of the CSSQ-S was .87. Participants provided demographic information including age, sex, ethnicity, marital status, education, and occupation; and sport information including sport (skiing, snowboarding, telemarking), ability (same as previous chapters), and number of days skied per year. Participants’ CSSQ-S score was included only if they reported at least intermediate ability.  5.3.3  Genetic analysis Buccal cell DNA was isolated from cytobrushes using alcohol purification technique  described in Chapter 2 (Saftlas et al., 2004), and samples were diluted to 20 ng DNA/µL. A total of 26 SNPs from eight genes were successfully genotyped using the Sequenom iPLEX®  20  Subset of Telus Festival Sample (n = 668 described in Chapter 2), after exclusions for ethnicity and age. Exclusion for ability were only applied to CSSQ-S analyses. 113  technique (San Diego, California, USA) at the McGill University and Genome Québec Innovation Centre, Montréal, Quebec, Canada. The mean marker-call rate was 99 ± 0.021% (sample plate layout shown in Appendix N; marker list and project report from Genome Quebec shown in Appendix T).  5.3.4  Statistical analysis To increase the power to detect associations, a two-stage joint-analysis model was used  (similar to that described by Skol and colleagues (2006)). In stage 1, approximately 50% of the sample (n = 291, “discovery sample”) was selected at random using the PAWS-SPSS (version 18.0) random case selector and screened for the 26 polymorphisms, with significance was set at  ! = .05 (i.e., no correction for multiple testing). In stage 2, SNPs that met this inclusion threshold were analyzed in the remainder of the sample, herein referred to as the “replication sample” (n =308), and then in the combined sample (n = 599). A joint-p-value from the two independent samples was calculated (using Fisher's combined probability test; Fisher, 1932) and a correction for multiple testing was applied using the Bonferonni method (! = .05/x, where x is the number of independent tests performed (e.g., Lunetta, 2008)). Details about Fisher’s method are shown in Appendix S. An analysis of co-variance (ANCOVA) was used to compare sensation-seeking measures (ImpSS and CSSQ-S, quasi-dependent variables) between genotypes for each SNP (betweensubject factors). In both stages of analysis, additive models of inheritance (three genotype groups, i.e., three factor levels) were tested. Sex was considered a covariate as significant gender differences were expected based on previous ImpSS (e.g., Zuckerman & Kuhlman, 2000) and  114  CSSQ-S data (Chapter 3). Skiing or snowboarding ability was considered as a covariate only for CSSQ-S analyses since ability significantly correlated with CSSQ-S score in a Chapter 3.  5.4  Results Genotype frequencies for all of the SNPs tested in stage 1 were in Hardy-Weinberg  Equilibrium (HWE, p > .05). Scores derived from CSSQ-S and ImpSS were normally distributed. As indicated in Table 5-2, the selected discovery sample for stage 1 analyses (n = 291) did not differ from the replication sample (n = 308) in terms of education, sport, and sensation seeking (i.e., the phenotypes of the study) variables; however, there were differences in the ability and sex compositions of the samples (Table 5-2; there were approximately 10% more females and more intermediate level participants in the replication sample). In both stages, males scored higher than the females on impulsive sensation seeking (discovery: t(289) = 2.12, p < .05; replication: t(276) = 2.51, p < .05) and on the contextual sensation seeking in skiing and snowboarding assessment (discovery: t(282) = 9.30, p < .001; replication: t(290) = 9.81, p < .001); therefore, supporting the use of sex as a covariate. CSSQ-S score was significantly correlated with ability (discovery: r(279) = .58, p < .0001, replication: r(289) = .62, p < .0001) providing support for including ability as a covariate when analyzing CSSQ-S scores.  115  Table 5-2 Descriptive statistics and results comparing the samples at stages 1 and 2 Descriptive statistics and results comparing the samples at stages 1 and 2 Descriptive statistics Variables Grouped Sex Education Sport Ability  Discovery sample Stage 1 (nD = 291) 180 males, 111 females 64% post secondary or higher 60% skiers, 35% snowboarders, 5% other 19% intermediate, 30% advanced, 47% expert, 4% other  Replication sample Stage 2 (nR = 308) 162 males, 146 females 65% post secondary or higher 52% skiers, 40% snowboarders, 8% other 27% intermediate, 31% advanced, 35% expert, 7% other  Stage 1 vs. 2 Statistic  p  !2 =5.24 !2 =3.18 !2 =2.96  .02* .20 .23  !2 =8.81  .01*  Continuous Age M = 26.88 SD = 6.61 M = 27.35 SD = 6.61 t = 0.90 .37 † CSSQ-S M = 36.89 SD = 7.05 M = 35.83 SD = 7.65 t = -1.73 .09 ImpSS M = 12.73 SD = 3.85 M = 12.42 SD = 3.88 t = -0.58 .55 Imp M = 3.83 SD =2.29 M = 3.71 SD = 2.23 t = 0.64 .53 SS M = 8.93 SD = 2.12 M = 8.70 SD = 2.25 t = 1.25 .21 Note. Chi-square tests were used for grouped variables and t-tests for the continuous variables (age, CSSQ-S, and ImpSS). M = mean, SD = standard deviation. †  Total participants included in CSSQ-S means differ from sample total (nD = 284 and nR = 293) because  skiing/snowboarding ability was less than intermediate or missing. *p < .05.  Genotype frequencies for three of the 26 polymorphisms were associated with a sensation-seeking phenotype in stage 1 ANCOVA (at p < .05; see Table 5-3) and therefore included in the stage 2 analysis of the replication sample. Data for all SNPs are shown in Appendix U. In stage 2, one SNP: rs167771 in DRD3 was marginally associated with the sensation-seeking (SS) construct from the ZKPQ (p = .055, Table 5-3). The joint p-value, obtained by combining the p-values for the association between sensation seeking and rs167771 in the discovery and replication analyses, was significant (p = .012) and remained significant after correcting for multiple testing (at ! = .017). The association between sensation seeking and rs167771 was also present in the combined sample (p = .004, "p2 = .02). Pair-wise comparisons 116  (using the Least Significant Difference, LSD) between genotype groups in the combined sample revealed that individuals with the rs167771 GG genotype scored lower than AG individuals (LSD = -1.05, p = .049) or AA individuals (LSD = -1.46, p = .005), and AG individuals scored lower than AA individuals (LSD = 0.39, p = .047), and this trend was present in the independent samples (Table 5-4). In other words, the number of A alleles carried was positively correlated with sensation seeking. The other two SNPs tested in stage 2 were not associated with either sensation-seeking measures (p > .2) and the joint p-values were above the corrected level of significance (Table 5-3). Genotype frequencies for the SNPs (including rs167771) that advanced to stage 2 remained consistent with HWE.  117  Table 5-3 ANCOVA results from stage 1 and 2 ANCOVA results from stage 1 and 2 Gene  1  Marker  DRD2  C>G  dbSNP  rs17601612  DV  N  ImpSS  290  Discovery sample MAF F .39  Imp 2  DRD2  3  DRD3  †  pD  n  3.37  .036  308  4.22  .016  Replication sample MAF F .42  pR  Combined sample F pC  Joint Analysis pJointa  0.57  .564  1.35  .260  .100  0.06  .938  2.41  .090  .078  Taq1A  rs1800497  ImpSS  290  .23  3.73  .025  308  .21  0.08  .923  2.22  .110  .110  A>G  rs167771  ImpSS  291  .17  308  .16  3.98  .047  1.43  .243  5.14  .006  .063  Imp  3.47  .024  0.13  .877  2.56  .078  .102  SS  3.55  .030  2.93  .055  5.54  .004  .012*  Note. Only p < .05 are shown. Sex was included as a covariate. MAF = minor allele frequency, DV = dependent variable. †  rs1800497 is located in ANKK1 (downstream from DRD2), but is often considered a “DRD2” polymorphism.  a  pJoint Independent p-values from the discovery and replication samples were combined using Fisher’s method (Fisher, 1932).  *significant at ! = .017.  118  Table 5-4 Descriptive statistics for ImpSS scale and subscales grouped by DRD3 rs167771 Descriptive statistics for ImpSS scale and subscales grouped by DRD3 rs167771 Sample  Genotype n  ImpSS  Imp  SS  Discovery (n = 291)  AA  203  13.07 (3.75)  4.01 (2.29)  9.05 (1.98)  AG  77  12.38 (3.85)  3.56 (2.25)  8.81 (2.23)  GG  11  9.73 (4.56)  2.36 (1.86)  7.36 (3.26)  AA  216  12.68 (3.81)  3.76 (2.25)  8.91 (2.17)  AG  85  11.86 (4.05)  3.60 (2.23)  8.26 (2.35)  GG  7  11.29 (3.90)  3.57 (1.99)  7.71 (2.50)  AA  419  12.87 (3.78)  3.89 (2.27)  8.98 (2.08)  AG  162  12.10 (3.95)  3.58 (2.23)  8.52 (2.31)  GG  18  10.33 (4.27)  2.83 (1.95)  7.50 (2.92)  Replication (n = 308)  Combined (n = 599)  Dependent variable  Note. Values represent mean (SD).  5.5  Discussion There was a significant association between one SNP, an intronic A to G transition  (rs167771) in the DRD3 gene and sensation seeking: homozygotes for the G allele scored lower on sensation seeking than individuals homozygous for the A allele, with heterozygotes showing an intermediate level. No associations were found between the SNPs in any of the other genes investigated (DRD1, DRD2, DAT1, COMT, MAO-B, DBH, 5HT2A), or between the most commonly studied DRD3 rs6280 (Ser9Gly) variant, which previously has been associated with sensation seeking and novelty seeking (Duaux et al., 1998; Staner et al., 1998). DRD3 is expressed in limbic regions of the brain (e.g., hippocampus, nucleus accumbens, ventral striatum) suggesting that the receptor is involved in emotion and cognition (Bouthenet et al., 1991). A role for the gene in novelty seeking may be hypothesized based on the observation that Drd3 knock-out mice display hyperactivity in a test for exploratory behaviour (Accili et al., 1996). The DRD3 gene has been investigated as a candidate gene for a number of phenotypes  119  including alcoholism (Gorwood et al., 2001), drug dependence (Duaux et al., 1998), and attention deficit hyperactivity disorder (ADHD) (Muglia, Jain, & Kennedy, 2002), all of which have been associated with impulsivity and sensation seeking (commentary from Kaplan, B. within Depue & Collins, 1999; Roberti, 2004; Verdejo-Garcia et al., 2008). Recently associations were reported between the G allele in DRD3 rs167771 and autism spectrum disorder (ASD) (de Krom et al., 2009) and extrapyramidal symptoms in psychiatric patients (Gasso et al., 2009). Another ASD study reported a conflicting finding: an association between the rs167771 G allele and a decreased risk for “insistence on sameness”, a domain within the Revised Autism Diagnostic interview (Staal et al., 2012). Although no direct comparisons between ASD studies and these findings can be made, the relationship between ASD and sensation-seeking phenotypes may be relevant to this study as the sensation-seeking trait is associated with a desire for novelty, while the insistence on sameness domain includes the items: “difficulties with minor changes in personal routine and environment” and “resistance to trivial changes in environment” (Szatmari et al., 2006). High sensation seekers crave stimulation, respond better to novelty in the environment, and have less preference for routine than low sensation seekers (Zuckerman & Kuhlman, 2000) (e.g., ZKPQ SS item, “I would like the kind of life where one is on the move and travelling a lot, with lots of change and excitement”). The current finding that lower scores on sensation seeking were associated with the G allele is consistent with that of de Krom et al. (2009) who reported that the same allele (G at rs167771) was associated with stereotyped (or repetitive) behaviour in ASD. Another important candidate gene in personality research is the DRD4, a highly polymorphic gene that has been associated with novelty seeking, extraversion, ADHD, schizophrenia, and drug use (Bookman et al., 2002; Ebstein et al., 1996; Kereszturi et al., 2007;  120  Okuyama, Ishiguro, Toru, & Arinami, 1999). Alleles at several loci in the DRD4 promoter were investigated during the current series of research projects and results are presented in Chapter 6.21 While the results presented in the current chapter for DRD3 and DAT1 are consistent with earlier published results (Jonsson et al., 2003; Jorm et al., 2000), they differ from those reported for other approach and/or externalizing behaviour studies; for example: sensation seeking with COMT Val158Met (U. E. Lang et al., 2007), DRD2 Taq1A with impulsivity and alcohol use (Esposito-Smythers, Spirito, Rizzo, McGeary, & Knopik, 2009), rs463379 (DAT1) with ADHD (Friedel et al., 2007), and rs686 (DRD1) with drug use (Liu et al., 2006). The discordances between the current results and those listed above may be due to a number of factors, including the polygenic inheritance of complex phenotypes, heterogeneous phenotyping methods, and differences in sample demographics (e.g., ethnicity and sex) (M. R. Munafo et al., 2003). Comparing studies is difficult due to varying allele frequencies between populations and inconsistency in whether the sexes are combined for analysis. There was a significant association between sensation seeking and rs167771 with sex as a covariate and the relationship between sex and sensation seeking was significant. In many cultures, sensation seeking has been associated with “masculine” characteristics and negatively associated with “feminine” characteristics (Daitzman & Zuckerman, 1980; Kish, 1971), and studies investigating associations between sensation seeking and dopaminergic genes (COMT) have found female-specific associations (Kang et al., 2010; U. E. Lang et al., 2007). The current findings and those mentioned above emphasize the importance of considering the influence of sex when analyzing a trait that differs consistently in magnitude between males and females.  21  Had these results been included in the analyses in the current chapter, the additional four SNPs would not have impacted the significant results, as none of them reach significance in stage 1. 121  Although the total sample (n = 599) was sufficient to achieve modest power (~ .80) based on small effect sizes (!2 = .01 to .02) typical of SNP associations, the sample was split in order to carry out an internal replication in an attempt to reduce chance findings. As a result, there was reduced power in the split samples, but taking the joint p-value into account improves the overall power of the study compared to discovery and replication samples considered independently (Skol et al., 2006). To compensate for the risk of false findings (both positive and negative (Christley, 2010)) corrections for multiple testing were applied in the second stage. In summary, 26 SNPS in eight dopaminergic and serotonergic genes were tested for associations with general and sport-specific sensation-seeking behaviour using a two-stage, jointanalysis design. After replication and correction for multiple testing, there was a significant association between sensation seeking and the DRD3 rs167771 polymorphism in skiers and snowboarders. Other DRD3 variants have been associated with externalizing phenotypes related to sensation seeking, but the results presented in this chapter are the first to suggest a role for the DRD3 rs167771 in sensation seeking. While the present findings are intriguing, they are preliminary and the association should be investigated in other sports, cultures, and populations.  122  Chapter 6: No association between promoter variants of the dopamine-4 receptor gene and sensation seeking in skiers and snowboarders.  6.1  Summary Twin studies have shown that impulsivity and sensation seeking are heritable traits, and  candidate genes encoding components involved in dopaminergic transmission have been targets for association studies. The gene that is most commonly implicated in approach-related phenotypes is the dopamine-4-receptor gene (DRD4). The promoter of the DRD4 contains a number of polymorphisms that have not been explored in association with sensation seeking. Five such common polymorphisms (-1106T/C, -906T/C, -809A/G, -291C/T, 120-bp duplication) were analyzed in a cohort of skiers and snowboarders: practitioners of commonly practiced but potentially hazardous sports. Impulsive sensation seeking and domain-specific (skiing) sensation seeking were compared between genotype groups in 599 skiers/snowboarders. No association was seen between genotype(s) and general or domain-specific sensation seeking.  6.2  Introduction The promoter region of the dopamine-4-receptor gene (DRD4) is highly polymorphic  (Okuyama et al., 2000) and multiple variants within the gene have been associated with novelty seeking, schizophrenia, and externalizing disorders (Golimbet et al., 2005; Lai et al., 2010; Mitsuyasu et al., 2001; M. R. Munafo et al., 2008; Rogers et al., 2004). A number of polymorphisms exist in a span of approximately 1000 bases upstream of DRD4, which despite their proximity, appear to independently segregate in Japanese and Caucasian populations (Mitsuyasu et al., 2007; Nakajima et al., 2007; Szantai et al., 2005). Among these are 19 SNPs  123  and four variable length polymorphisms (two repeat variants and two insertion/deletion) in the promoter region of the DRD4 (Mitsuyasu et al., 2007 and shown in Table 6-1); however 13 of them are rare with minor allele frequencies (MAF) ! .05, and would not be very informative for studying continuous personality traits (Balding, 2006). DRD4 is highly expressed in the frontal cortex and limbic regions of the brain and the receptor encoded by this gene is thought to be involved in attention, motivation, and emotional processing, all of which play a role in decisionmaking and risk-taking (Kreek et al., 2005). Decision-making is central to the personality trait “impulsivity”, which often involves acting without forethought (Evenden, 1999). The ZKPQ clusters impulsivity with a correlated trait, “sensation seeking”, which involves a desire for novelty and seeking stimulation through intense experiences and taking risks for the sake of these experiences (Zuckerman et al., 1993). Impulsivity and sensation seeking have been associated with disinhibited behaviours including gambling (Alessi & Petry, 2003), binge drinking (Carlson et al., 2010), and drug use (VerdejoGarcia et al., 2008), but also to non-deviant, prosocial outlets like travel and entrepreneurship (Lepp & Gibson, 2008; Nicolaou et al., 2008) (reviewed in Chapter 1). Sports are another common prosocial outlet for sensation seekers, and numerous studies have putatively linked high-risk sports with sensation seeking (reviewed in Goma-i-Freixanet et al., 2012); however, the relationship with impulsivity is less clear (D. J. Llewellyn, 2008). Sports like skiing and snowboarding are considered high-risk due to the relatively high chance of severe injury (Goma-i-Freixanet et al., 2012). To my knowledge other than the studies presented in this dissertation, there has only been one genetic association study on personality traits in high-risk sport populations (Cam et al., 2010); whereas, there have been numerous genetic studies (including several which investigate DRD4) on other risk-inclined populations (reviewed  124  in Dick et al., 2009; Goodman, 2008). The physiological mechanisms that underlie the motivation to participate in antisocial pastimes may be similar to those that attract people to high-risk sports (Zuckerman, 1983). To test the hypothesis that variants in DRD4 influenced general sensation seeking and domain-specific sensation seeking, a cross sectional, single cohort design was employed to test for association(s) between DRD4 genotypes in proficient skiers and snowboarders and their scores from: 1) the ZKPQ trait measures for impulsive sensation seeking (ImpSS) and 2) the Contextual Sensation Seeking Scale for Skiing and Snowboarding (CSSQ-S). Sensation seeking measures were compared between genotype groups at multiple loci in the 5’-upstream region of DRD4. Inclusion of polymorphisms was limited to informative, common SNPs (MAF > .05 in Caucasian populations) that could be analyzed using a multiplex genotyping technique (leaving a total of six SNPs, see Table 6-1), and an additional promising candidate, the 120-bp tandem duplication that required genotyping by standard PCR. Of the remaining six common promoter SNPs, the -521C/T (dbSNP rs180955) had already been previously studied in a portion22 of this sample (see results in Chapter 4) and this SNP along with the nearby -616 G/C (dbSNP rs747302) failed to amplify at the genotyping facility. Excluding failed attempts, the following variants were analyzed: -1106 T/C, -906 T/C, -809 G/A, -291 C/T (see Table 6-1 for dbSNP ID), a few of which have been studied in association with disinhibited behaviours and traits (Mitsuyasu et al., 2001; Nemoda et al., 2010; Okuyama et al., 2000). As well, a 120-bp tandem duplication (starting at position -1.24 KB) that reportedly reduces transcriptional efficiency  22  The -521C/T failed amplification at the Genome Quebec genotyping facility using a Sequenom approach and failed to amplify in a large portion of the Festival sample when genotyping was attempted locally by RFLP and pyrosequencing. The -521C/T results were treated as a separate study, because the aim was to replicate findings from my MSc., and the manuscript (Chapter 4) was written long before carrying out the analyses presented in the current chapter. 125  (D'Souza et al., 2004) was included in the study. There has been no functional support for the other deletion/repeats; therefore these were not genotyped. Other than the commonly studied 521C/T, -616G/C, and 120-bp duplication, only two studies investigating DRD4 promoter SNPs in association with approach traits have been published, both of which analyzed the -906T/C variant (Derringer et al., 2010; Heck et al., 2009).  6.3 6.3.1  Methods Participants Skiers and snowboarders visiting the 2010 Telus World Ski and Snowboard Festival in  Whistler, British Columbia, Canada completed two questionnaires and provided a buccal (cheek) cell swab for DNA preparation (methods described in Chapter 2). Unrelated participants were between 17 and 49 yrs of age (n = 59923, M = 27.12 years, SD = 6.45) and were pre-screened for ability, reporting at least intermediate ability, defined as capable of skiing/snowboarding a “blue square run” comfortably. The majority of the participants were either skiers (56%) or snowboarders (38%) while the remaining practiced both sports (5%) or were telemarkers (1%). To minimize confounding effects due to differing biogeographical background, only Caucasian participant (self-reported as of “European descent”) were included in the genetic analysis (participant demographics are shown in Table 6-2).  23  See also Chapter 5. 126  Table 6-1 A summary of DRD4 promoter polymorphisms A summary of DRD4 promoter polymorphisms Marker  dbSNP  Reference  MAF  Population  120-bp repeat  rs4646984  (Seaman, Fisher, Chang, & Kidd, 1999)  .20  European  -1217 G/del  rs12720364  (Okuyama et al., 2000)  .17  Japanese  Functional support Affects transcriptional efficiency (D'Souza et al., 2004) -  -1123 C/T  -  (Okuyama et al., 2000)  <.01  Japanese  -  -1106 T/C  rs936460  .26  European  -  -1102 G/A  -  NCBI dbSNP (Wang et al., 2004) (Mitsuyasu et al., 2007)  <.01  Japanese  -  -930 C/G/T  -  (Mitsuyasu et al., 2007)  <.03  Japanese  -  -906 T/C  rs3758653  .20  European  -  -809 G/A  rs936461  .45  European  -  -768 G/A  rs4987058  NCBI dbSNP (Wang et al., 2004) (Mitsuyasu, Ozawa, Takeda, & Fukumaki, 1999) (Mitsuyasu et al., 1999)  <.01  European  -  -713 C/T  rs11246224  (Mitsuyasu et al., 2007)  <.03  Japanese  -  -616 G/C  rs747302  (Mitsuyasu et al., 1999)  .44  European  -615 A/G  rs936462  NCBI dbSNP  .05  European  Possible AP-2 binding site gain (Barr et al., 2001) -  -603 del/T  rs747303  (Mitsuyasu et al., 1999)  .46  Japanese  -  -600 G/C  rs10902180  (Mitsuyasu et al., 1999)  <.05  Japanese  -  -598 G/T -597 (G)2-5 -521 C/T  rs3842250 rs1800955  (Mitsuyasu et al., 2007) (Mitsuyasu et al., 2007) (Mitsuyasu et al., 1999)  <.01 .40  Japanese Japanese European  Affects transcriptional efficiency (Okuyama et al., 2000) -376 C/T rs916455 (Mitsuyasu et al., 1999) .03 European -364 A/G rs916456 NCBI dbSNP <.01 Japanese -291 C/T rs916457 (Mitsuyasu et al., 1999) .07 European -234 C/A (Mitsuyasu et al., 2007) <.01 Japanese -128 G/T (Mitsuyasu et al., 1999) .03 Japanese -11 C/T (Cichon et al., 1995) <.01 European Note. MAF = minor allele frequency. (-) refers to missing information, e.g., no rs# available, or a lack of studies investigating potential functional effects.  127  6.3.2  Measures Impulsive Sensation Seeking (ImpSS). Participants completed the 19 item ImpSS,  scored true/false (Appendix D) (Zuckerman et al., 1993). Data derived from the ImpSS subscales in this study demonstrated acceptable internal reliability (ImpSS ! =.80). If there are significant differences between genotypes and ImpSS, the subscale may be divided into its components (SS and Imp) to investigate whether the association is driven by differences in impulsivity or sensation seeking. Contextual Sensation Seeking Scale for Skiing (CSSQ-S). Participants completed the 10-item CSSQ-S, anchored on a Likert scale by 1 (strongly disagree) and 5 (strongly agree) (described in Chapter 3 and Appendix F). Data derived from the instrument in the current sample demonstrated high internal consistency (! = .87).  6.3.3  Genotyping Buccal cell DNA was isolated from cytobrushes (Fisher Scientific, Ottawa, ON, Canada)  using an alcohol purification technique (Saftlas et al., 2004) (described in Chapter 2; Appendix L) and diluted to 20 ng DNA/µl. Four polymorphisms (DRD4 -1106 T/C, -906 T/C, -809 A/G, -291 C/T) were genotyped using the Sequenom iPLEX® technique (San Diego, California, USA) at McGill University and Génome Québec Innovation Centre, Montréal, Quebec, Canada. The 120-base pair duplication was amplified using primers and PCR reagents described by Seaman et al. (1999) and using the “touchdown” thermal profile described in McCracken et al. (2000). Primers are shown in Appendix O. The PCR cycled in a MJ Mini Cycler (Bio-Rad Laboratories, Hercules, CA, USA) as follows: 95°C for 15 minutes, followed by 10 cycles of 95°C for 30s, 66°C for 30s, and 72°C for 90s, and then 25 cycles of 95°C for 30s, 55°C for 30s, 128  and 72°C for 90s, finishing with 72°C for 10 minutes. The 25 µL reactions contained 20 mM Tris-HCl pH 8.4, 50 mM KCl, 1.5 mM MgCl2, 0.2 mM dNTP, 0.2 µM of each primer, 5% DMSO, 0.5 U Taq polymerase (Invitrogen Corporation, Carlsbad, CA, USA), and 10-20 ng DNA template. PCR products were sized by electrophoresis using an 8% PAGE gel stained using SYBR Safe DNA gel stain (Invitrogen, California), visualized using BIORAD Gel DocTM EZ System and then identified as containing “short” (S; 429 bp) versus “long” fragments (L; 549 bp). Sample gel photographs shown in Appendix V.  6.3.4  Statistical analyses An analysis of variance (ANOVA) was used to analyze main effects of individual  variants (120 bp duplication (LL + LS vs. SS), -1106 C/T (TT vs. CT vs. CC), -906 T/C (TT vs. TC vs. CC), -809 G/A (GG vs. GA vs. AA), -291 C/T (CC vs. CT vs. TT)) between polymorphisms and each of the dependent variables (Imp, SS, CSSQ-S). An additive model of inheritance was applied for all SNPs because no support for an alternative model has been proposed. The 120-bp tandem duplication was analyzed using both additive and grouped models based on previous support for grouping carriers of the L allele (Rogers, 2004). The threshold for significance was set at an alpha of .01 to account for the five variants tested (using a Bonferroni correction (Attia et al., 2009)), though no correction for testing multiple dependent variables was applied because they are not independent. The effects of potential covariates were investigated before carrying out the analyses. Sensation seeking putatively varies with age and between the sexes (Zuckerman et al., 1993); therefore, the relationship between age and sensation seeking was measured using Pearson’s  129  correlation, and the effects of sex across sensation-seeking variables were analyzed using an independent sample t-test.  6.4  Results The distributions of the five variants were consistent with Hardy-Weinberg Equilibrium  (p > .05). The mean marker call rate for SNPs analyzed using the Sequenom iPlex® was 98 ± 0.33%. Genotype frequencies are shown in Table 6-3. Scores derived from the ZKPQ ImpSS and CSSQ-S scales were normally distributed, with no univariate outliers. Sex was entered in the initial analysis as a covariate, because there were significant differences between the sexes in both ski and global sensation seeking (CSSQ-S: t(576) = 13.70, p < .001; ZKPQ ImpSS: t(494)24 = 3.20, p = .001). Age was related to both CSSQ-S and ImpSS (CSSQ-S r(575) = -.24, p < .001; ImpSS r(597) = -.09, p = .04), but the relationship between ImpSS and age was too weak to be included as a covariate. Finally, ability was significantly correlated with CSSQ-S, r(570) = .60, p < .001 and was included as a covariate for CSSQ-S analyses only.  24  Levene’s test of equal variances was violated; therefore a test that does not assume equal variances was used. 130  Table 6-2 Descriptive statistics for demographic and personality variables Descriptive statistics for demographic and personality variables Variable Sex Education  Descriptive statistics 341 males, 258 females 65% post-secondary or higher Ability 24% intermediate, 31% advanced, 41% expert, 4% other † CSSQ-S M = 36.34 SD = 7.36 ImpSS M = 12.59 SD = 3.87 Imp M = 3.78 SD =2.26 SS M = 8.82 SD = 2.18 Note. M = mean, SD = standard deviation. †Total participants included in CSSQ-S means differ from sample total (n = 578) because skiing/snowboarding ability was less than intermediate or missing.  The initial sample was 599 participants; however, not all were genotyped successfully (the 120-bp duplication was especially problematic) and the final sample sizes for the genetic analysis ranged from n = 445 to 578. Univariate analyses comparing ImpSS (with sex as a covariate) between genotypes at each loci revealed no significant associations, the results for the Imp and SS scales when analyzed separately were also not significant (see Appendix W). Similarly, there were no significant associations between domain-specific sensation seeking (with ability, age, and sex as covariates) and any of the polymorphisms tested (see Table 6-3).  131  Table 6-3 Descriptive statistics and ANOVA results for DRD4 promoter polymorphisms Descriptive statistics and ANOVA results for DRD4 promoter polymorphisms CSSQ-S Marker  dbSNP  Genotype  120-bp repeat  -  LL LS SS  rs936460  SD  n  M  SD  296  36.72  7.54  300  12.78  3.84  126  36.35  7.00  131  12.53  4.00  14  40.14  4.72  14  13.07  3.65  140  36.73  6.89  145  12.58  3.96  F A, F G  1.13,  0.37  0.06,  0.11  pA pG  .27,  .54  .94,  .75  TT  272  36.32  7.50  281  12.75  3.72  TC  244  36.07  7.45  256  12.61  4.06  CC  61  37.56  6.43  62  11.81  3.73  †  FG  -809 G/A  -291 C/T  rs3758653  rs936461  rs916457  1.47  0.71 .49  PG -906 T/C  ImpSS  M  LS + SS  -1106 T/C  n  .23  TT  393  36.30  7.46  409  12.58  3.94  TC  163  36.06  7.42  169  12.58  3.63  CC  21  39.38  4.41  21  12.81  4.63  F  0.85  0.01  P  .43  .99  GG  225  36.12  7.82  235  12.70  3.90  GA  274  36.40  7.13  285  12.46  3.79  AA  74  36.99  6.97  75  12.68  4.18  F  0.60  0.18  P  .55  .84  CC  507  36.10  7.45  528  12.52  3.91  CT  51  37.67  6.61  62  13.00  3.63  TT F P  5  39.91 2.88 .06  6.06  5  13.40 0.48 .62  2.51  Note. F-statistics and p-values for both additive and grouped models are shown for the 120-bp duplication, all other SNPs were analyzed using additive genetic models. Sex was included as a covariate for ImpSS, and sex, ability, and age were included as covariates for CSSQ-S analyses. FA = additive model, FG = grouped model, M = mean, SD = standard deviation. †Homogeneity of variances was violated; therefore, grouped model was tested.  132  6.5  Discussion The goal of this study was to investigate the individual effects between five DRD4  promoter polymorphisms and two measures of sensation seeking in skiers and snowboarders. There were no significant associations between either the CSSQ-S (the domain-specific sensation-seeking measure) or the ZKPQ ImpSS and any of the DRD4 promoter variants studied. The DRD4 has been implicated in numerous association studies on approach-related personality traits and externalizing disorders (Gizer et al., 2009; M. R. Munafo et al., 2008). The 120-bp tandem duplication is of interest because it contains binding sequences for transcription factors (Seaman et al., 1999) and the long (240-bp) allele has, in some studies, reduced transcriptional activity compared to the short allele (D'Souza et al., 2004). Lower transcription could lead to lower DRD4 expression, resulting in fewer D4 receptors and ultimately affecting levels of dopamine in the synapse (D'Souza et al., 2004). The long version of the tandem duplication has been associated with attention deficit hyperactivity disorder (ADHD) and schizophrenia, conditions that are sometimes characterized by impulsivity (Lai et al., 2010; McCracken et al., 2000). Associations between the long allele (possibly encoding fewer D4 receptors) are viewed as being congruent with one aetiology of ADHD, namely the “dopaminedeficit theory” (Swanson et al., 2007); however, there are inconsistencies in the data, as some studies have reported associations between the short allele and ADHD and/or high impulsivity scores (Kereszturi et al., 2007; Rogers et al., 2004). Additionally, two recent meta-analyses reported no associations with the 120-bp tandem duplication (Gizer et al., 2009; Sanchez-Mora et al., 2011) and ADHD. In the current study, there were no significant differences in selfreported impulsive sensation seeking (nor for impulsivity and sensation seeking measured separately) between the 120-bp duplication alleles.  133  There are few studies that include the four other promoter SNPs analyzed in the current study (-291 C/T, -809 G/A, -906 T/C, and -1106 T/C), and there are no reports in the literature demonstrating differential functionality for the alleles at these loci. The four SNPs have been investigated in association with schizophrenia in two Japanese populations (Mitsuyasu et al., 2007; Nakajima et al., 2007), in Caucasian family-based association tests for ADHD (Lasky-Su et al., 2007; Oades et al., 2008) (one of which only looked at two of the four SNPs, -906 C/T and -291 C/T), and in association with novelty and sensation seeking (both only looking at the -906 C/T) (Derringer et al., 2010; Heck et al., 2009), but none of the studies reported significant associations between DRD4 promoter alleles and any of the phenotypes. Similarly, there were no significant main effects for any of the promoter SNPs analyzed in this study. Nakajima et al. (2007) examined interactions between DRD4 promoter variants, and reported that a combination of four polymorphisms (-809 G/A, -291 C/T, -616 G/C, and 12-bp repeat) conferred a susceptibility to schizophrenia and they found a trend for overrepresentation of a haplotype that included -906 C and -1106 C in the cases. Oades and colleagues (2008) reported a trend for over-transmission of the -906 C allele in ADHD cases exhibiting behavioural impulsivity, but it was not significant after correction. Overall, there are no consistent results linking any of these polymorphisms to approach-related traits or externalizing disorders. The study described in the current chapter had sufficient power to test for multiple variants, but the sample size was too small to adequately observe the effects of the rare genotypes. Oddly, the CSSQ-S scores for each “rare” genotype group tended to be higher and it would be interesting to see whether these trends held in larger samples. The goal of the study presented in this chapter was to assay all of the polymorphisms in the promoter region with high heterozygosity, but two (-521 C/T rs1800955 and 616 C/G rs747302) failed optimization both at  134  the genotyping facility and in the Rupert laboratory. Additionally, two deletion polymorphisms, -603 del/T and -1217 G/del (described by Mitsuyasu et al., 2007) were not included in the analysis because the multiplex genotyping technique used in this study was limited to transition and transversion polymorphisms. As the alleles at -603 del/T and -1217 G/del are in moderateto-strong linkage disequilibrium with alleles at -1106 C/T (Mitsuyasu et al., 2007), which was assayed, an association would likely have been detected if the -603 del/T or -1217 G/del genotype contributed to the phenotype. While this sample is sufficient for assessing effects of highly heterogenic polymorphisms, it would be worthwhile to investigate the effects between all common polymorphisms in the 1.2 KB region of the DRD4 promoter; however, a much larger sample would be necessary to obtain sufficient power to observe rare genotypes. A homogeneous cohort of experienced athletes participated in this study, in whom both general and domain-specific sensation seeking could be measured. The sample was limited to include only experienced skiers and snowboarders (turning away beginners) to ensure that respondents had sufficient ability to carry out the sport behaviours described in the CSSQ-S. While there are benefits to sampling from a homogeneous population (e.g., reducing extraneous variability), the results have limited generalizability. In summary, five common polymorphisms in the DRD4 promoter were tested for associations with sensation seeking in a sample of skiers and snowboarders. There were no significant associations between genotypes at these loci and sensation-seeking measures. These findings are for the most part in line with previous studies, which reported no associations between approach-related traits and other DRD4 promoter variants (excluding the -521 C/T).  135  Chapter 7: Interaction between allelic variants in the dopamine-4-receptor gene is associated with patterns of downhill sport behaviour  7.1  Summary The objective of this chapter was to investigate main effects or interactions of two  variants (-521C/T and the 48-bp VNTR) in the DRD4 in association with global and domainspecific sensation seeking. Global (ZKPQ ImpSS) and domain-specific sensation seeking (CSSQ) was measured in downhill-sport participants (n = 155, after exclusions; independent from Chapter 4 sample) and scores were compared between genotypes at two variants in DRD4: a 48-bp repeated segment encoding a region in the third cytoplasmic loop and a polymorphism in the promoter region (-521 C/T). There was a significant interaction between the two variants and sport-specific sensation seeking using both an additive (p = .001) and a dominant model (p = .001) for -521 C/T, but there were no significant associations with the global trait measure (p > .7). These results support previous studies that found associations between patterns of skibehaviour and the -521 C/T variant, and between novelty seeking and alleles at both loci.  7.2  Introduction High-risk downhill sports like skiing and mountain biking are popular and, despite  evidence for elevated rates of severe injury (Dodwell et al., 2010), have seen increases in participation rates (Puchan, 2004). Studies have shown that high-risk sports attract individuals who exhibit “thrill-seeking” tendencies and the personality trait sensation seeking has commonly been associated with high-risk sport participation (Goma-i-Freixanet et al., 2012), as well as with more “deviant” activities such as gambling, drinking, and drug use (M.T. Bardo et al., 2007;  136  Carlson et al., 2010). While environmental factors likely influence participation in mountain sports and/or other high-risk activities, twin-studies have shown sensation seeking to be moderately heritable (Stoel et al., 2006), suggesting that genotype may underlie some of the motivation for participation in such activities. Sensation seeking falls under the broader class of approach traits (e.g., extraversion and novelty seeking), which reflect a sensitivity to reward and strong behavioural activation systems (C. S. Carver & White, 1994). Most studies of approach traits have investigated variants in dopaminergic genes due to the purported role of dopamine in behavioural activation and instrumental learning (Lauzon & Laviolette, 2010; Salamone et al., 2007). The dopamine-4receptor gene (DRD4) is of particular interest in studies of approach traits as, unlike the four other common dopamine receptor sub-types, DRD4 expression is highly concentrated in areas of the brain thought to be involved in emotional regulation, attention, and motivation (i.e., prefrontal cortex, amygdala, and other regions of the limbic system) (Lauzon & Laviolette, 2010). DRD4 is a highly polymorphic gene, and two commonly studied variants are a single nucleotide polymorphism (SNP) -521 C/T (a cytosine to thymine transition at base -521 in the upstream promoter region; dbSNP rs1800955) and a 48 base-pair variable number tandem repeat (VNTR) in exon III. The 48-bp VNTR alleles vary between 2 and 11 repeats (often grouped by the presence or absence of the “long” (6 or more repeats) and differences between DRD4 alleles in their ability to inhibit cyclic AMP following dopamine stimulation have been observed (Asghari et al., 1995). Although the exon III VNTR is more commonly studied, a meta-analysis reported a larger effect size for the association between -521 C/T and “approach traits” (M. R. Munafo et al., 2008). The SNP could have an impact on function, since the -521 C allele is associated with a 40% increase in DRD4 transcription in cultured cells (Okuyama et al., 2000).  137  Numerous studies have reported individual associations between either the -521 C or the “long” VNTR and sensation/novelty seeking, impulsivity, and externalizing disorders; however, few have considered intragenic effects (in which the phenotype is influenced by a combination of alleles at both loci). The purpose of this study was to extend on the findings from Chapter 4 and to investigate interactions or main effects of alleles at the -521 C/T and the 48-bp VNTR in association with global (ZKPQ ImpSS) and domain-specific (CSSQ) sensation seeking in an independent sample of athletes. The CC genotype has been previously associated with sensation seeking in skiing (Chapter 4) and the VNTR long allele is more commonly associated with approach (M. R. Munafo et al., 2008); therefore, individuals carrying the CC genotype were expected to score higher on sensation-seeking measures but that the effect may vary depending on the VNTR genotype.  7.3 7.3.1  Methods Participants Individuals between 17 and 59 years of age (n =191: 139 male, 52 female; mean age =  27.25 years (SD = 8.71)) from the high- and low-risk sport cohorts (described in Chapter 2) who reported participation in a downhill sport (e.g., skiing, snowboarding, or mountain biking) at an intermediate or better ability participated in the study. Participants were recruited from Vancouver, British Columbia and Bordeaux and Chamonix, France. To minimize confounding effects of differing biogeographical backgrounds on allele frequencies, only participants selfreporting being of “European descent” (minimum third generation European) were included in  138  the genetic analysis. After providing informed consent, participants completed questionnaires (49% in French, 51% in English) and provided a buccal (cheek) cell sample for DNA analysis.  7.3.2  Measures Demographic information. Athletes completed a brief demographic assessment that  included questions about age, ethnicity, ancestry, language proficiency, and education. In addition, they provided information on sport, frequency, and ability (scored 0 = beginner, 1 = novice, 2 = intermediate, 3 = advanced or expert). Participants scoring ability < 2 were excluded from the analysis. Ability was converted to a dichotomous variable (intermediate vs. advanced/expert), because an ordinal variable with three levels led to a non-fitting model. As a proxy for psychiatric screening, participants completed a self-report section on history of prescribed medication (including type, reason for, and duration of use) that included a list of anxiolytics, anti-depressants, and neuroleptics (using common English and French names, Appendix X). Participants reporting current or past use of a medication for treatment of a psychiatric illness were excluded from the analysis. CSSQ. Sensation seeking in the context of downhill sports was assessed using the Contextual Sensation Seeking Scale (CSSQ, Appendix Y), a generalized version of the skispecific CSSQ-S described in Chapter 3 that was adapted to be compatible for skiing and biking25. The CSSQ comprises 10 items, anchored on a Likert scale by 1 (strongly disagree) and 5 (strongly agree). Exemplar items include “I like to ski/ride fast” and “I like to push my  25  The CSSQ adaptation involved minor word additions, such as adding “drops” where formerly the CSSQ-S included on the word “jump”. E.g., CSSQ item 5: “I like to attempt jumps/drops even if I’m not sure of the quality of the landing area”. 139  boundaries when I play my sport.” Using back-translation from English to French, and back to English until the translation agreed, a French version of the CSSQ was created (Appendix Z). Scores derived from the CSSQ demonstrated high internal consistency (Cronbach alpha French = .85, English = .90). ZKPQ ImpSS. Global sensation seeking was assessed using English (Zuckerman et al., 1993) and French (Rossier, Verardi, Massoudi, & Aluja, 2008) versions of the Zuckerman Kuhlman Impulsive Sensation Seeking scale (ZKPQ ImpSS). The scores derived from the ImpSS scales demonstrated acceptable internal reliabilities (Cronbach alphas: French = .71, English = .87).  7.3.3  Genotyping Buccal cells were obtained from inside of the participant’s cheek with a cytobrush  (Fisher Scientific, Ottawa, Canada). DNA was isolated from the recovered cells using an alcohol purification technique (Saftlas et al., 2004) (described in Chapter 2) and genotyped for the -521 C/T26 polymorphism at the McGill University Genome Québec Innovation Centre, Montréal, Canada using the Sequenom (San Diego, U.S.A.) iPLEX® technique (marker-call rate = 97.4%); and the exon III 48-bp VNTR was genotyped in the Robinson laboratory at Children’s and Women’s Hospital in Vancouver as described below. The VNTR was genotyped using a protocol created by Smolen et al. (2002). Primers used were D4-VNTR-F: 5’-AGG ACC CTC ATG GCC TTG-3’ (with a 5’ HEX fluorescent label) and D4-VNTR-R: 5’-GCG ACT ACG TGG TCT ACT CG-3’ (NAPS IDT, Vancouver,  26  The -521 C/T SNP failed amplification at the genotyping facility when the Festival samples were sent (Chapter 5), but for unknown reasons successfully amplified for the high- and low-risk samples. 140  BC, Canada). DNA was amplified in a MJ Mini Cycler (Bio-Rad Laboratories, Hercules, CA, USA) using touchdown cycling as follows: 95°C for 10 minutes, followed by 2 cycles of 94°C for 30s, 65°C for 30s, and 72°C for 60s. After 2 cycles the annealing temperature would drop 2°C and the cycle would repeat twice, and this continued until the annealing temperature reached 57°C. This was followed by 30 cycles of 94°C for 30s, 55°C for 30s, and 72°C for 60s, finishing with 30 minutes held at 72°C. The 25 µL reactions contained 20 mM Tris-HCl pH 8.4, 50 mM KCl, 2.0 mM MgCl2, 200 µM of each dNTP (+ 7-deaza-2-deoxy GTP; Roche Applied Science, Indianapolis, IN, USA), 245 µM of each primer, 10% DMSO, 1.0 U Taq polymerase (Invitrogen Corporation, Carlsbad, CA, USA), and 10-20 ng DNA template. PCR products (2 µL) were mixed with 20 µL of Hi-Di formamide and 0.5 µL of Genescan 2500 Rox and then were visualized on an ABI Prism 3100 Genetic Analyzer (Life Technologies, Carlsbad, USA). Amplicons varied in length between 379-bp (2-repeats) and 667-bp (8-repeats). Approximately 40 additional genotypes were identified using gel electrophoresis. PCR products were electrophoresed on an 8% PAGE gel stained using SYBR Safe DNA gel stain (Invitrogen, California), visualized using BIORAD Gel DocTM EZ System and then identified as containing 2 to 8 repeats. Each gel was run with a reference sample already genotyped using the ABI Prism. Sample gel photographs shown in Appendix AA.  7.3.4  Statistical analysis Sensation seeking scores (CSSQ and ImpSS) between genotype groups were compared  using a mixed model with two fixed effects corresponding to the two variants (VNTR and -521 C/T) as categorical factors with ability, sex, and age as covariates. Random effects were estimated for the model intercept with participants nested within their recruitment location  141  (France or Canada) in order to account for non-independence of sampling. Both a 2 x 3 and a 2 x 2 design were tested, each employing two factor levels for the 48-bp VNTR and three and two factor levels for the -521 C/T SNP, respectively. The 48-bp VNTR factor levels include alleles grouped by length: the S (short) allele (2 to 5 repeats) is more common than the L (long) allele (6 to 11 repeats, commonly referred to as 7R allele and second most common to the 4R allele) and as different binding profiles between the SS and SL genotypes have been reported (Asghari et al., 1994) genotype groups were collapsed based on the presence of the long allele (SS vs. SL + LL). The -521 C/T alleles are both common (i.e., minor allele > 40%) in European populations (Rajeevan, Soundararajan, Kidd, Pakstis, & Kidd, 2012), therefore both an additive (three factor levels: CC vs. CT vs. TT) and a grouped model in which homozygotes for the minor allele were grouped with heterozygotes (two factor levels: CC + CT vs. TT) were tested based on a previous DRD4 studies (Bellgrove et al., 2005; H. J. Lee et al., 2003; Nemoda et al., 2010). As CSSQ scores differ between males and females and are significantly correlated with ability and age (e.g., Chapter 6); relationships between these potential covariates and the CSSQ and ImpSS were tested.  7.4  Results After excluding participants for non-European ancestry (8), use of medications related to  a psychiatric illness (9) or for failure to genotype at both loci (19), 155 participants remained in the analysis (47 recruited from Canada, 108 from France). Scores derived from the CSSQ were significantly correlated with age (r(155) = -.20, p < .05), and there were significant differences between the sexes (t(153) = 5.15, p < .001) and abilities (t(153) = 6.3, p < .001), but no differences in sex, ability, or age were present across factor (genotype) levels (p > .3; shown in  142  Table 7-1); therefore, these variables were included as covariates. No significant trends were observed between the ZKPQ ImpSS measures and demographic variables (p > .3). The distributions of genotypes at each locus did not differ by country (VNTR: !2(2)= 1.5, p = .5; -521 C/T: !2(2)= 3.5, p = .2) and satisfied Hardy-Weinberg Equilibrium when assessed independently (all p > .3) and in combination (p > .6), suggesting that the alleles were segregating independently (genotype frequencies are shown in Table 7-2). The longest allele observed for the VNTR was 8 repeats.  Table 7-1 Comparison of demographic variables between factor levels Comparison of demographic variables between factor levels Genotype  Sex, n Male  Female  Ability, n Int Adv + Exp  Age, M (SD)  VNTR SS 65 24 26 63 27.20 (7.95) SL 43 15 12 46 27.04 (8.59) LL 7 1 2 6 26.43 (9.18) SL + LL 50 16 14 52 Statistical test F(2, 118) = 0.03 !2(1) = 0.15 !2(1) = 1.27 p-value .70 .26 .97 -521 C/T CC 22 8 9 21 28.65 (8.56) CT 57 17 18 56 26.54 (8.46) TT 36 15 13 38 27.10 (7.50) Statistical test F(2, 152) = 0.55 !2(2) = 0.67 !2(2) = 0.36 p-value .72 .84 .58 Note. Chi square tests were used to compare counts between groups and analysis of variance was used to compare age between groups. Due to small cell sizes for LL genotype, 2 x 2 contingency tables were used comparing counts for SS and SL + LL for sex and ability. n = counts per genotype group, M = mean, SD = standard deviation, Int = intermediate, Adv = advanced, Exp = expert, VNTR = variable number of tandem repeats, S = short allele, L = long allele.  143  Table 7-2 Tests of Hardy Weinberg Equilibrium (HWE) and genotype frequencies Tests of Hardy-Weinberg Equilibrium (HWE) and genotype frequencies for each subsample Genotype France (n = 108) Canada (n = 47) Combined (n = 155) VNTR SS 60 29 89 SL 41 17 58 LL 7 1 8 HWE !2(2) = 0, p = 1.0 !2(2) = 0.69, p = .41 !2(2) = 0.13, p = .71 521 C/T CC 17 13 30 CT 52 22 74 TT 39 12 51 HWE !2(2) = 0, p = .96 !2(2) = 0.19, p = .66 !2(2) = 0.12, p = .73 Note. VNTR = variable number of tandem repeats, S = short allele, L = long allele.  A mixed model analysis of CSSQ scores and genotypes at -521 C/T and exon III VNTR, when adjusted for age, sex, and ability, revealed a significant interaction between the genotype groups using either an additive model for -521 C/T (F(2, 153) = 6.98, p = .001) or a dominant model (F(2, 152) = 10.52 p = .001). Both results remain significant after correcting for testing two variants and an interaction for two dependent variables, tested using two models of inheritance (e.g., corrected alpha = "/12 = .004). Least significant difference comparisons of simple effects based on model-estimated marginal means revealed that in the grouped -521 C/T model, individuals carrying at least one long VNTR allele had higher CSSQ scores if they also carried a -521 C allele (CT or CC) (p = .004) compared to those homozygous for the -521 T allele (shown in Figure 7-1 using marginal mean scores). Under the additive model, CSSQ scores for individuals carrying the CC genotype did not vary across levels of the VNTR (SS or SL + LL), but -521 C/T heterozygotes scored higher on CSSQ if they carried a long VNTR allele (p = .035) and conversely, TT homozygotes scored significantly lower if they carried a long VNTR allele (p = .003), Figure 7-2. No significant main effects or interactions between the loci  144  and ImpSS were observed (additive model: VNTR: ImpSS F = 0.17, p = .68; -521 C/T: ImpSS F = 0.32, p = .72; interaction: ImpSS F = 0.09, p = .92).  (44)  (30)  (60)  (21)  Figure 7-1. CSSQ scores grouped by genotypes for the 48-bp VNTR (SS vs. SL & LL) and the -521 C/T (CC & CT vs. TT). Cell sizes for each group are shown (n). CSSQ = contextual questionnaire for sensation seeking, VNTR = variable number of tandem repeats, SNP = single nucleotide polymorphism, S = short allele, L = long allele.  145  (16)  (30)  (30)  (14) (44) (21)  Figure 7-2. CSSQ scores grouped by genotypes for the 48-bp VNTR (SS vs. LL & LS) and the -521 C/T (CC vs. CT vs. TT). Cell sizes for each group are shown (n). CSSQ, contextual questionnaire for sensation seeking; VNTR, variable number of tandem repeats; SNP, single nucleotide polymorphism; S, short allele; L, long allele.  CSSQ scores for -521 C/T genotype groups are shown below in Table 7-3 to show that the results are in the same directions as the findings from Chapter 4. There was a trend for individuals with the C allele scoring the higher on contextual sensation seeking. Scores for the 48-bp VNTR genotype groups are also shown, to see if the trend is in keeping with the literature (carriers of the L-allele scoring higher on approach); however, since a significant interaction between the factors was present, the main effects are not reported.  146  Table 7-3 CSSQ scores by genotype for the -521C/T and 48-bp VNTR polymorphisms CSSQ scores by genotype for the -521 C/T and 48-bp VNTR polymorphisms Variant -521 C/T  Genotype  n  CSSQ, M (SD)  CC CT TT  30 74 51  35.60 (6.30) 35.83 (8.50) 34.89 (7.28)  VNTR LL 8 37.74 (7.71) LS 58 35.72 (7.93) SS 89 35.13 (7.58) Note. CSSQ = contextual sensation seeking questionnaire, M = mean, SD = standard deviation, VNTR = variable number of tandem repeats, S = short allele, L = long allele.  7.5  Discussion A significant interaction between two DRD4 variants was associated with contextual  sensation seeking; athletes carrying both “risk” alleles (-521 C and long VNTR) reported higher scores on the CSSQ (sport-specific questionnaire). Contrary to previous findings (Okuyama et al., 2000; Roussos, Giakoumaki, & Bitsios, 2009); however, there were no main effects on sensation-seeking scores for either variant, although trends for high contextual sensation seeking scores in carriers of the risk alleles were observed. Inconsistent results have been reported for both DRD4 variants in personality studies on healthy populations (M. R. Munafo et al., 2008). Though alleles at multiple loci in the DRD4 have hypothesized functions (Asghari et al., 1995; Okuyama et al., 2000), few studies have investigated interactions between variants in the gene. Consistent with a study finding an interaction between the DRD4 VNTR and -521 C/T was associated with novelty seeking; data presented in this chapter reveal that the combination of a long VNTR allele and a -521 C allele was associated with higher scores on approach (H. J. Lee et al., 2003). An interaction between alleles at the VNTR and -521 C/T was also observed in association with infant’s attachment  147  disorganization (Lakatos et al., 2002), a phenotype that has been associated with childhood behavioural problems (Fearon, Bakermans-Kranenburg, van Ijzendoorn, Lapsley, & Roisman, 2010), and synergistic genetic interactions between other DRD4 variants and externalizing disorders have been observed (Nakajima et al., 2007). A study that examined a -521 C/T-VNTR “genotype combination”27 revealed that children carrying a long VNTR and a -521 T allele showed decreased event-related potential indicative of brain activity associated with reflexive attention shifting and orienting reflexes (Birkas et al., 2006), both of which are behavioural correlates of sensation seeking (Zuckerman, 2007a). Though not directly comparable, carriers of both a long VNTR and -521 T allele reported the lowest sensation-seeking scores in the current chapter. Independently, both the -521 C and the VNTR long alleles have been associated with novelty seeking (Ebstein et al., 1996; Okuyama et al., 2000; Ronai et al., 2001), extraversion (Bookman et al., 2002; Golimbet et al., 2007), impulsivity (Congdon, Lesch, & Canli, 2008), financial risk taking (Dreber et al., 2009; Dreber et al., 2011), and sexual risk taking (Garcia et al., 2010); and while there are numerous null findings (Ekelund et al., 2001; Jonsson et al., 2002; M. R. Munafo et al., 2008), few studies report associations with the opposite alleles (e.g., the -521 T and VNTR short associated with risk-taking or approach). More recently, the DRD4 7R was associated with risk taking (Roussos et al., 2009) and slower response times (Szekely et al., 2011) in laboratory tasks, and while the latter study investigated other SNPs (including the -521 C/T), they did not explore interactions between variants. Interestingly, the current study reveals high CSSQ scores in carriers with one of each “risk” allele.  27  Birkas et al (2006) refer to the 7R-T genotype combination as a “haplotype”; however, the term haplotype implies the presence of linkage disequilibrium and studies on other European (Canadian) populations report no linkage between the -521 C/T and exon III VNTR loci (Barr et al., 2001). 148  The current study revealed an allelic interaction in association with domain-specific sensation seeking (CSSQ), but not with the ZKPQ ImpSS. Similarly, in the independent study described in Chapter 4, the -521 CC genotype was associated with high CSSQ scores in skiing and snowboarding, but there was no association with the broader trait measure. The sample used in the current study was smaller than the full sample used in Chapter 4 (n = 503); but was comparable to the Chapter 4 pilot sample (n = 117). The lack of significant findings in the current study could be due to low statistical power, or perhaps -521 C homozygotes in the Pilot sample in Chapter 4 were enriched with the VNTR long allele, but this is purely speculative. The loci are not in linkage disequilibrium (Barr et al., 2001) and the alleles should therefore segregate independently and combinations of alleles will vary from sample to sample. Perhaps many null findings associated with DRD4 variants are because researchers failed to consider genotypes at both loci, and the combination of genotypes at the -521 C/T and exon III VNTR may be important. As the popularity of downhill and adventure sports continue to increase, recreation sites create interesting natural settings for observing sensation-seeking behaviours in the field which may be used to measure correlates between self-report measures (i.e., the CSSQ) and physiological (e.g., cortisol, dopaminergic transmission, etc.) or genetic variables. An analysis of two variants in the DRD4 revealed an interaction between the exon III VNTR and the -521 C/T was associated with patterns of sensation-seeking behaviours in downhill sports. Both variants have previously been associated with approach traits, and though there were no main effects for either variant, the results are in the same direction as previous findings. A few limitations should be noted: all of the participants reported being of European descent and; therefore, the results may not generalize to other populations, and larger samples are preferable  149  for genetic association studies. The DRD4 is a highly polymorphic gene that contains a number of putatively functional variants. Future studies should consider both main and interaction effects of DRD4 genotypes on the expression of phenotypes. Such studies will require large samples in order to obtain a sufficient number of individuals who are homozygous for both minor alleles.  150  Chapter 8: Exploratory analysis of personality and genetic variables in highand low-risk sport participants 8.1  Summary Athletes participating in high-risk sports consistently report higher scores on sensation-  seeking measures than low-risk athletes or non-athletic controls, and may be an interesting group in which to study genetic variants commonly associated with sensation seeking. Personality traits (impulsivity and sensation seeking) and genetic variants (29 polymorphisms in 14 candidate genes) were compared between proficient athletes participating in high-risk sports (n = 141) and low-risk sports (n = 132). Sport cohorts differed on a number of demographic variables; therefore, whenever possible, matched subsamples were used to compare personality traits and were used for follow-up genetic analyses. Athletes participating in high-risk sports score higher than low-risk sport athletes on sensation seeking (p < .05), but not impulsivity. There were marginal associations between sport group and variants in two genes: stathmin (p = .004) and brain-derived neurotrophic factor (p = .03), trends that were also present when subsets of the sport cohorts were compared (p < .05), but the associations did not survive correction for multiple testing. This chapter also provides descriptive data on disinhibited behaviours (e.g., smoking, alcohol, substance use) in the two cohorts, and compares personality traits between athletes engaged only in prosocial sensation seeking (through sport) and other athletes engaged in both deviant (drug use) and prosocial outlets for sensation seeking.  8.2  Introduction Participation rates in high-risk sports have been increasing over the last 30 years (Celsi et  al., 1993). Certain sports, like skiing long ago attained mainstream popularity, but what were 151  once “fringe” sports, like snowboarding, rock climbing, kayaking, surfing, and mountain biking are becoming more common. The Outdoor Foundation (USA) (2010) studied participation rates for a multitude of outdoor activities in over 400,000 people and provided an estimate for growth within sports based on the percentage of first-time participants. In 2009, of the Americans reporting rock climbing, the percentage of first-time climbers was 43%. Similarly high growth was seen for a few other high-risk activities including whitewater kayaking (26.5%) and adventure racing (24%). It seems there has been a shift in leisure sport participation, and perhaps people are realizing that these high-risk sports satisfy a need for novelty that many people share. High-risk sport participants often report a need for stimulation, as measured by sensationseeking scales, and they consistently score higher than non-risky athletes or controls (Goma-iFreixanet et al., 2012). Numerous studies have compared personality traits between participants from high- and low-risk sports, but only one study has compared genetic variations between these two groups. Cam and colleagues (2010) grouped athletes recruited from a Turkish university by participation in high- or low-risk sports and compared the frequencies of three variants commonly studied in association with approach-related traits: the DRD4 VNTR, HTR2A 102 C/T, and the 3’ UTR VNTR in DAT1. They did not observe any differences in allele frequencies between the sports groups, but when they examined continuous personality traits in the combined sample, there were trends between HTR2A genotype and a number of “Big Five” dimensions. The study has several weaknesses: they do not provide details about the ability or frequency of participation in high-risk sports, nor do they provide any details about the demographic characteristics of the two sport groups (e.g., age, sex, ethnicity, etc.). Furthermore, it is unclear whether members of the “risk” group were regular, proficient participants in highrisk leisure activities or whether or not other differences existed between the two groups that  152  may have contributed to the null findings in the between-groups comparison. Athletes who participate in high-risk sports are rarely studied, whereas there are numerous studies that compare genotype frequencies between high- and low-risk groups defined by substance use disorder, alcoholism, externalizing disorders, and also extreme scorers on novelty- and sensationseeking subscales (Ekelund et al., 2001; M. R. Munafo et al., 2008; Nernoda et al., 2011). There are number of genes that are candidates for association with participation in highrisk sports. Athletes engaged in high-risk sport not only report high levels of sensation seeking, but they report low levels of avoidance and neuroticism (Castanier et al., 2010b; I. H. Franken et al., 2006; Goma-I-Freixanet, 1991; Schaal et al., 2011; Tok, 2011) and may experience less fear associated with common triggers (e.g., heights) (Zuckerman, 2007c). Genes involved in dopaminergic neurotransmission are of particular interest due to dopamine’s purported role in reward seeking and approach motivation (Depue & Collins, 1999; Pecina et al., 2003; Zuckerman, 2005a). Genes involved in serotonergic neurotransmission are also potential candidates because serotonergic genes have been associated with harm avoidance, and weak “avoidance” systems in combination with strong “approach” systems are thought to underlie the sensation-seeking trait (Zuckerman, 2007a). In addition, serotonin genes have been implicated in risk taking behaviour in animals (Fairbanks et al., 2001; Long et al., 2009). Dopamine and serotonin systems are thought to interact (Depue & Collins, 1999) supporting the inclusion of genes from both systems in an investigation of risk-taking through sport. A number of researchers suggest that high-risk sport participation is an expression of sensation seeking that is similar to taking drugs, drinking, or engaging in risky sex (M. T. Bardo, Donohew, & Harrington, 1996; Celsi et al., 1993; I. H. Franken et al., 2006), and therefore genetic variants associated with these behavioural expressions might also be associated with high-risk sport.  153  To investigate the possibility that variants in genes that are involved in dopaminergic and serotonergic neurotransmission might be associated with high-risk sport participation, variants in candidate genes were compared between cohorts of proficient high- and low-risk sport participants. Biallelic polymorphic loci at which both alleles were moderately common (heterozygosity > .2) were chosen in genes that encode proteins involved in transport (dopamine and serotonin transporters: DAT1, SLC6A4), function (dopamine and serotonin receptors: DRD1, DRD2, DRD3, HTR1A, HTR2A), metabolic inactivation (catechol-O-methyltransferase, (COMT), monoamine oxidase A (MAO-A), dopamine-!-hydroxylase (DBH)) or precursors (tyrosine hydroxylase (TH)). Some of the same variants were tested for associations with sensation seeking in a cohort of skiers (see Chapter 5), but the choice of SNPs for the study described in the current chapter was based on the following criteria: 1) variants for which significant associations with global or contextual sensation seeking were found in Chapter 5 (i.e., rs167771), 2) any variants that have consistently been associated with approach-related traits and behaviours in the literature, 3) loci at which alleles have shown functional differences in the literature (i.e. affect structure or function of molecules encoded by the genes), and 4) additional candidate genes (that are not a part of dopamine and serotonin pathways) chosen based on genetic associations with fear-related phenotypes (stathmin, STMN1; Brocke et al., 2010; Shumyatsky et al., 2005) and sensation seeking (brain-derived neurotrophic factors, BDNF; Kang et al., 2010). Details about individual candidate genes are found in Chapter 1, and reasons for choosing SNPs are outlined in Table 8-1. The data presented in this chapter only include SNPs that were genotyped at Genome Québec (a genotyping facility based at McGill University). Two additional polymorphisms were analyzed in the high- and low-risk samples; however, analyses of one variant (the DRD4 exon III VNTR) are included in a separate chapter (Chapter 7) because  154  analyses were carried out only in high- and low-risk athletes that had completed the CSSQ (downhill sport questionnaire). The DRD4 VNTR and -521 C/T (rs1800955) were analyzed separately using a continuous trait design due to observed relationships between patterns of ski behaviour and -521 C/T in Chapter 4. Another copy number variant, the DRD4 120-bp tandem duplication (described and analyzed in skiers in Chapter 6), was successfully genotyped in only a portion of the high- and low-risk sport cohorts and analyses are shown in Appendix BB. In addition to comparing genetic variants, sensation seeking, impulsivity, and disinhibited behaviours were compared between sport groups. There is strong support for a link between substance use and sensation seeking (M.T. Bardo et al., 2007); therefore, it was important to consider the chance that participants in high-risk sports, who presumably score high on sensation-seeking measures (Goma-i-Freixanet et al., 2012), may also be more likely to experiment with drugs. A number of the candidate genes chosen for this study have previously been implicated in substance use disorders; therefore, to avoid confounding results it was important to exclude problematic substance users. The presence of ADHD symptoms may be another potentially confounding variable. ADHD has been suggested to be an extreme manifestation of sensation seeking (Brocke et al., 1999), and ADHD, like sensation seeking, is putatively related to dopaminergic neurotransmission (Swanson et al., 2007; Turic et al., 2010). A symptom checklist was included to screen for possible ADHD cases since a number of the genes proposed as candidates for sensation seeking in the current chapter have been implicated in ADHD (Gizer et al., 2009).  155  Table 8-1 List of SNPS chosen for analysis List of SNPS chosen for analysis Gene  Full Name  Protein Function  Marker  Ch. 5  1  BDNF  Brain-derived neurotrophic factor  rs6265  2  COMT  Catechol-O-methyl transferase  Nerve growth factor involved in neural plasticity Metabolizes DA  3  †  Functional support  Reason to include SNP  -  missense SNP  GWAS with approach traits  rs4633  Yes, ns  (Hirata et al., 2008)  Functional  COMT  rs4680  Yes, ns  Functional  4 5 6  COMT COMT DAT1  rs4818a rs6269 rs2652511  (Chen et al., 2004; Lotta et al., 1995)  -  -  Association with approach Association with approach Association with approach  7 8  DAT1 DAT1  rs27072a rs2975226a  Yes, ns Failed  (Pinsonneault et al., 2011) -  Functional Association with approach  9 10  DAT1 DBH  rs6347 rs1611122  Yes, ns -  (Pinsonneault et al., 2011) (Cubells et al., 2011)  11 12 13  DBH DBH DRD1  rs161115 rs6271 rs265981  Yes, ns -  (Zabetian et al., 2001) (Zabetian et al., 2001) -  Functional Contributed to linkage signal in externalizing disorders Functional Functional Association with approach  14  DRD1  rs4532  Yes, ns  (Ota et al., 2012)  15  DRD1  rs686  Yes, ns  16  DRD2  rs1076560  Yes, ns  (W. H. Huang & Li, 2009; W. H. Huang et al., 2008) (Y. Zhang et al., 2007)  17  DRD2  rs1800497  Yes, ns  18 19  DRD2 DRD2  rs2283265 rs6277  Yes, ns Yes, ns  Dopamine transporter  Dopamine-!hydroxylase  Dopamine receptor D1  Dopamine receptor D2  Reuptake of DA from the synapse  Converts DA to norepinephrine  Involved in dopaminergic neurotransmission Metabolizes DA, NE, 5HT  Involved in dopaminergic neurotransmission  (Noble et al., 1991) (Jonsson et al., 1999) (Y. Zhang et al., 2007) (Duan et al., 2003)  Association with approach, pharmaco-genetic marker for antipsychotic drugs Functional Functional Functional Functional Functional  156  †  Gene  Full Name  Protein Function  Marker  Ch. 5  20  DRD3  Dopamine receptor D3  Involved in dopaminergic neurotransmission  rs167771  21  DRD3  rs6280  Yes, p < .01 Yes, ns  22  DRD4  rs11246226  23 24 25 26 27  DRD4 DRD4 DRD4 DRD4 DRD4  28  HTR1A  Serotonin receptor 1A  29  HTR2A  Serotonin receptor 2A  30  HTR2A  31  HTR2A  32  MAOA  Monoamine oxidase A  33  SLC6A4  Serotonin transporter  34  STMN1  Stathmin  Dopamine receptor D4  Functional support  Reason to include SNP  -  Significant association in Chapter 5 Functional  -  (Lundstrom & Turpin, 1996) -  rs3758653 rs762502a rs916457 rs936460 rs936461  Yes, ns failed Yes, ns Yes, ns Yes  -  DRD4 is top candidate for approach DRD4 is top candidate for approach DRD4 is top candidate for approach DRD4 is top candidate for approach DRD4 is top candidate for approach  rs6295  -  (Lemonde et al., 2003)  Functional  rs6311  Yes, ns  (Smith et al., 2012)  Functional  rs6312  -  (Myers et al., 2007)  Functional  rs6314  -  Functional  Metabolizes DA, NE, 5HT  rs6323  -  (Hazelwood & SandersBush, 2004) (Jansson et al., 2005)  Reuptake of 5HT from the synapse Regulates microtubule formation (involved in neural plasticity)  rs25532  -  (Wendland et al., 2008)  Functional  rs182455  -  -  Involved in dopaminergic neurotransmission  Involved in serotonergic neurotransmission Involved in serotonergic neurotransmission  DRD4 is top candidate for approach  Functional  Association with startle and cortisol response 35 STMN1 rs213641 Association with startle and cortisol response 36 TH Tyrosin hydroxylase Catalyzes step in dopamine rs10770141 (Rao et al., 2008) Functional, association with NS synthesis (Sadaihiro et al., 2010) Note. Heterozygosity of all SNPs > .2 (ALFRED: www.alfred.med.yale.edu). SNPs that have been consistently associated with approach-related traits or that have purported functional differences between alleles were included in the current study. Additional details about functional properties associated with alleles at each loci are shown in Table 2-1. DA = dopamine, 5HT = serotonin, NE = norepinephrine, ns = not significant, GWAS = genome-wide association study. † a  Some of the SNPs were investigated in association with sensation seeking in skiers in Chapter 5.  Four SNPs failed optimization and were not genotyped.  157  8.3 8.3.1  Methods Participants A total of 146 high-risk athletes (mean age = 29.1 years, SD = 9.1; 81% male) and 141  low-risk athletes (mean age = 25.8 years, SD = 9.8; 55% male) participated in the study. After exclusions for missing data and sport group overlap (detailed below), the final sample included 141 high-risk athletes (age 29.2 years, SD = 9.2; 81% male), and 132 low-risk athletes (age 25.8 years, SD = 9.8; 52% male). All athletes were unrelated to each other. Further exclusions are detailed below. Demographic variables are shown in Table 8-2, and types of sports are shown in Table 8-3.  8.3.2  Procedures The majority of the recruitment took place in France, specifically in Bordeaux,  Chamonix, and Pau. Athletes were recruited using a variety of media, including Twitter, Facebook, online forums (e.g., www.basejump.org, www.chamonixdailydump.org), and through team managers (e.g., Red Bull Team), and posters displayed at the sites around Chamonix (e.g., Town Hall, Guides Bureau, Mountain Guide School, etc.). Participants were approached at locations frequented by high-risk sport enthusiasts (e.g., base of gondola to the Aiguille du Midi (a famous mountaineering and glacier traverse route in Chamonix), bus stops, the Chamonix Guides Bureau, adventure tour kiosks). While this one-to-one method was successful, it was difficult to obtain sufficient numbers for a genetic association study; therefore, additional recruitment occurred at festivals including the North Face Ski Challenge, Chamonix, France and the Openride Festival, Chamonix, France. Another “mass” recruitment occurred at an airfield in Pau, France where a number of BASE jumpers, skydivers, and wing-suit fliers participated in the  158  study. Low-risk athletes were recruited at the University of Bordeaux and through sports clubs in the city (a few were recruited in Chamonix). Due to the insufficient number of low-risk athletes recruited in France, recruitment continued after my return to Canada. A number of lowrisk athletes were recruited through UBC sports teams (track & field, golf, cross country, and triathlon). Once participants provided informed consent they completed a series of questionnaires (detailed below section 8.1.3). Both French and English versions of the instructions and the questionnaires were prepared so that athletes could complete the questionnaire in the language in which they were most proficient. The questionnaires took approximately 20 minutes to complete and the same instructions for completing the questionnaire package were used in France and in Canada. The participants could choose one of three media in which to complete the questionnaire (online, ipad, paper, detailed below). Surveys were available on www.surveymonkey.com through a secure account. If the participant did not have time to complete the survey at the time of recruitment, after providing informed consent and a DNA sample (if interested in the DNA component) he/she had the option to provide an email address. The survey link was promptly sent to the participants with instructions and an ID code in order to complete the survey online from home. A second paperless option involved a survey software program iFormBuilder mobile platform (Herndon, Virginia, USA) on an iPad (Apple, California, USA) for the purpose of providing an environmentally friendly option to data collection. The questionnaires and consent forms used in both the online and ipad formats were identical in content to the third option: the paper version. Both digital surveys had “skip logic”, meaning that if the participant answered “no” to a question such as: “do you participate in any of the following high-risk sports?” the survey would  159  automatically skip to the following question. Participants always had the option to return to the previous question, in order to make the paper and paperless options as similar as possible.  8.3.3  Measures The questionnaire component for the study described in the current chapter was  significantly more detailed than that in the projects on skiers and snowboarders described in Chapters 3 to 6. Additional screenings measures were included (e.g., psychiatric screening, ADHD screen, problematic substance use screen) to ensure that there was not an erroneous association with an underlying condition, because the study design involved comparing genetic and psychological variables between two sport groups (rather than comparing continuous variables between genotype groups).  8.3.3.1  Demographic variables Participants provided information about age, sex, marital status, dependents, occupation,  home country, education, ethnicity, grandparents’ ethnicity, first language, fluency and number of years speaking language used for survey (see Appendix CC). Participants reporting nonEuropean ancestry (3rd generation) were excluded from genetic analyses, but were not excluded from personality analyses.  8.3.3.2  Psychotropic medication As a proxy for psychiatric screening (in lieu of a structure interview), participants were  asked to report use of medications. A list of common names for anxiolytics, antidepressants, and neuroleptics was included in the questionnaire, and participants answered questions about history  160  and duration of use (Appendix X). Any participants reporting history or current use of psychotropic medication were excluded from all analyses.  8.3.3.3  Substance use inventories Smoking. Participants provided information about tobacco smoking status, years  smoking, and cigarettes per day. Smoking history was followed by a modified (four-item) version of the Fagerström Nicotine Dependence (Appendix CC) (Heatherton, Kozlowski, Frecker, & Fagerstrom, 1991) or the French version (Test de Fagerstrom; OMS, 1998). Alcohol use. Participant’s alcohol use was measured using a modified version of the CAGE questionnaire (named after acronym: “Cut down on drinking, Annoyances with criticisms about drinking, Guilt about drinking, and using alcohol as an Eye opener”) (Ewing, 1984), a brief 4-item questionnaire used to assess at-risk drinking behaviours. In addition, participants provided information about weekly frequency of alcohol consumption. A French version of the CAGE (called the DETA in French) is available (Reynaud et al., 1997) and had been used by the psychology laboratory in Bordeaux (Santé et Qualité de vie, EA4319). The French version does not use expressions, such as “eye opener” or “cut down” and since it was expected that some participants would be non-native English or French speakers, the CAGE was modified slightly to include clear language (without expressions). For example, item 1: “have you ever felt that you need to decrease your consumption of alcoholic beverages”, instead of “…cut down on your drinking” (Appendix CC). Illicit substances. From a list of illicit substances, participants selected any substances that he/she had tried (a modified list from the Adolescent Drug Involvement Scale (Moberg &  161  Hahn, 1991); see Appendix CC). The list was followed with questions about frequency of use and social pattern of consumption. If the participant had only tried a substance, but no longer reported use, he/she then skipped the rest of the questions pertaining to use of that particular substance. Participants reporting current use completed a modified version of the Cannabis Abuse Screening Test (CAST) in order to characterize their drug-use patterns (available in French and English; Legleye, Karila L, Beck F, & M:, 2008). The CAST is a 6-item questionnaire with dichotomous answers (yes/no) (Appendix CC). The word “cannabis” was replaced with the more general “drugs”, and will be referred to as the “modified-CAST”. Participants reporting problematic substance use (as defined by a modified-CAST score > 2) were excluded from all personality and genetic analyses, with the exception of a comparison of drug use between high- and low-risk sport groups.  8.3.3.4  Sport inventories Participants selected sports in which they currently participate from a list of high- and  low-risk sports. In order to exclude individuals who had only tried the sport once28, they were asked to list sports in which they participate “regularly” (more than twice per year29), followed by details about his/her most proficient sport (including ability and frequency). French translations for common names of sports were verified with local sport experts in Chamonix. The high-risk sport participants did not complete a low-risk sport inventory because there were no exclusions of high-risk sport participants for participating in low-risk sports as well (e.g., mountaineers might train during their off season by running). On the other hand, low-risk sport  28  Bungee jumping, paragliding, and skydiving are sports that many people try only once in a lifetime, but one attempt at a sport would not be sufficient for inclusion in the high-risk sport group. 29 The term “regularly” is used liberally here, because some sports may only be practiced twice a year (e.g., if a person considers a mountaineering trip a single bout of practice). 162  participants completed both inventories because an athlete approached because of their participation in a low-risk sport (i.e., golf team) may also practice BASE jumping, which would exclude them from the low risk group. All athletes were pre-screened for their participation in high- or low-risk sports in order to ensure they completed the correct questionnaire. Sports in the high-risk inventory were chosen based on web searches of “high-risk” and “extreme” sports, sports featured at mountain film festivals, and personal communication with athletes. There are no comprehensive lists of high-risk sports in the literature. The list included the choice “other” to avoid missing new or obscure sports (Appendix DD). In addition to the sport inventory, athletes reporting at least intermediate ability in mountain biking, skiing, or snowboarding completed a generalized version of the 10-item CSSQ (described in Chapter 3, see also Appendix Y), a context-specific sensation seeking measure. A French version of the CSSQ was created using back-translation from English to French and back to English until the translation agreed. Scores derived from the CSSQ demonstrated high internal consistency (Cronbach alpha French = .85, English = .90). Athletes reporting less than intermediate ability in the downhill sport of their choice were excluded from CSSQ analyses.  8.3.3.5  Adult Self-Report Symptom Checklist 1.1 (ASRS-V1.1) Athletes completed a brief 6-item scale to measure symptoms of attention deficit  hyperactivity disorder (ADHD; Appendix EE). The scale is available in English (Kessler et al., 2005) and French (Caci, Bayle, & Bouchez, 2008). The ASRS-V1.1 has been shown to be sensitive (Hines, King, & Curry, 2012), and has moderate internal consistencies (Kessler et al., 2005). An ASRS-V1.1 score > 3 suggests the participant has symptoms consistent with adult ADHD, but a formal diagnosis would require an in-depth clinical interview (Kessler et al., 2005).  163  Participants were not excluded from psychological analyses on the basis of high ADHD scores; however, individuals with ASRS-V1.1 > 3 were excluded from the genetic analysis due to the extensive overlap between candidate genes associated with sensation seeking and those that have been implicated in ADHD. Scores derived from the ASRS-V1.1 in the current study demonstrated modest internal consistency (Cronbach alpha French = .67, English = .67).  8.3.3.6  ZKPQ ImpSS Participants completed the English or French 19-item version of the ImpSS, scored using  true/false format (Appendix D) (French version validated by Rossier et al., 2008; Zuckerman et al., 1993). Scores derived from the ImpSS demonstrated acceptable internal consistency (Cronbach alpha French = .75, English = .86), and internal consistencies for the subfactors were modest (Cronbach alpha Imp (8 items): French = .65, English = .80; SS (11 items): French = .73, English = .79).  8.3.4  Genetic analysis DNA collection and preparation. Participants provided two cheek swabs (instructions  shown in Appendix K, a French translation of the instructions was available). Buccal cell DNA was isolated using the same alcohol-based purification technique described in Appendix L. The concentrations of all samples were measured using a Nanodrop ND-100 Spectrophotometer (Thermo Fisher Scientific Inc, Waltham, MA) and were diluted to approximately 20 ng/µl as per requirements outlined by the genotyping facility (Genome Quebec). Sample plate layout is shown in Appendix N.  164  Genotyping. The McGill University Genome Québec Innovation Centre (Montréal, Quebec, Canada) carried out the genotyping using the Sequenom iPLEX® technique (San Diego, California, USA). A total of 32 SNPs30 in 14 genes were successfully amplified (Table 8-1). Amplification failed for four SNPs. The mean marker call rate was 99.16% (SD = 0.012%). For additional information about markers and for the Genome Quebec project report see Appendix FF.  8.3.5  Analyses In order to carry out case-control analyses, the control group must be as similar as  possible to the case group for all variables except the grouping variable (reviewed by Attia et al., 2009). Attempts were made to 1) recruit equal representation of males and females in both groups, 2) to balance the age of the groups, and 3) to recruit the participants from similar geographic regions; however, the high-risk sport participants encountered during recruitment were largely male and over 30 years of age. Additionally, a majority of high-risk athletes were recruited in Chamonix, France, a place where high-risk recreation enthusiasts from around the world congregate,31 and the majority of people who practice low-risk sports regularly in Chamonix do so as training for their high-risk sports. As such, low-risk athletes were not recruited in Chamonix, and instead attempts were made to recruit low-risk athletes while in Bordeaux, France, but the success rate was low, so most were recruited at a later date in Vancouver, Canada. As recruitment at two sites could have contributed to potentially  30  Genome Quebec genotyped 33 SNPs, but the -521C/T (rs1800955) was analyzed separately in a subset of the participants to test for associations with the CSSQ using a single cohort design (described in Chapter 7). 31 A large number of the athletes recruited in Chamonix were not from France, they came from all over the world: e.g., USA, Canada, Australia, United Kingdom, Norway, etc. 165  confounding sample heterogeneity, the extent of the demographic differences between high- and low-risk groups were assessed and adjustments to the analyses are described below. Personality and disinhibited behaviours. Demographic variables were compared between sports groups using Student’s t-tests for continuous variables, and Chi square tests for categorical variables. Demographic characteristics are summarized for high- and low-risk sport groups in Table 8-2. Personality measures were compared between subsets of the high- and lowrisk groups grouped by sex and location of recruitment (e.g., males recruited in France participating in high-risk were compared to males recruited in France participating in low-risk sports). Specifically, ANOVAs were used to compare the mean scores from the following measures between subsets of the sports groups: impulsive sensation seeking (and each subfactor: impulsivity, sensation seeking), and CSSQ (for those who reported participation in downhill sports). Consumption patterns for drugs, alcohol, and smoking were also compared between sport groups using t-tests of scores from the modified-CAST, modified CAGE/DETA, and modified Fagerström, respectively. Finally, impulsivity and sensation seeking in high-risk male athletes reporting problematic drug use (modified-CAST > 2, from France only) were compared to other high-risk male athletes reporting no problematic use. It was expected that the individuals excluded from all other analyses based on their drug use would score higher on impulsivity than the athletes from the high-risk sport group reporting no problematic drug use. Genetic analyses. Genotype frequencies for all SNPs were tested to see if each satisfied Hardy-Weinberg Equilibrium. To provide support for comparing participants recruited in Europe with those recruited in Canada, additional analyses comparing genotype frequencies between recruitment locations were carried out in the low-risk sport participants. In order to investigate whether alleles in candidate genes involved in approach-related phenotypes were 166  overrepresented in athletes participating in high-risk sports, allele frequencies at each of the 32 SNPs were compared between high- and low-risk sport participants using Chi square analyses. When there were differences between allele frequencies (at an un-corrected alpha level of .05) in the un-matched sport samples, the analysis was re-tested in a subset matched for recruitment location. A Bonferroni correction for testing multiple SNPs was applied (Attia et al., 2009).  8.4 8.4.1  Results Participant exclusions There are a number of exclusion criteria, which vary depending on the analysis. The  following athletes were excluded from all analyses: nine low-risk sport participants who also regularly participated in high-risk sports, two low-risk and five high-risk sport participants with missing data, and 14 low-risk and two high-risk sport participants reporting a history of medication-use related to psychiatric illness. Athletes reporting problematic32 substance use based on a modified-CAST score > 2 were excluded from personality and genetic analyses that involved comparisons between sports groups (n = 1 from low-risk group, n = 25 from high-risk group); however, separate analyses were performed to investigate characteristics of the high-risk sport participants who reported problematic substance use (n = 25). After exclusions, there were 114 high-risk athletes and 117 low-risk remaining for the between-sport group personality analyses. Participant characteristics for these cohorts are shown in Table 8-2. For the genetic analysis, and additional three (leaving n = 111) high-risk and 17 (leaving n = 100) low-risk sport participants were excluded for reporting non-Caucasian ancestry among grandparents. Due to the overlap between candidate genes implicated in sensation  32  A CAST score > 2 suggests that the individual is exhibiting disordered or problematic use (Legleye et al., 2008). 167  seeking and those implicated in ADHD, participants with ASRS-V1.1 scores > 3 were excluded from the genetic analyses to avoid potentially confounding effects. There were 30 high-risk athletes and 28 low-risk athletes reporting ASRS > 3, leaving 81 and 71 in the high- and low-risk groups, respectively. Table 8-2 shows participant characteristics pre- and post-exclusions for both personality and genetics analyses. Participation rates for sports chosen by high- and lowrisk athletes are shown in Table 8-3.  Table 8-2 Participant’s characteristics pre- and post-exclusions Participant’s characteristics pre- and post-exclusions Pre-exclusions Demographic variable Sex M/F (%) Age, M (SD) Marital status (%) Single Common-law Married Other Dependents (%) Education (%) High school Post-secondary Graduate Language for questionnaire (%) English French Years spoken (%) <10 years 10-15 years 16-20 years >20 years First language (%) English French Other  Post exclusions (Personality) High Low (114) (117) 79/21 53/47 30.3(9.7) 25.1(9.2)  Post exclusions (Genetics) High Low (81) (71) 78/22 48/52 31.4(10.3) 26.7(10.8)  Betweensport group  High (141) 81/19 29.2(9.2)  Low (132) 52/48 25.8(9.8)  55 24 16 5 16  71 15 10 4 8  54 25 18 3 19  72 16 9 2 8  52 22 22 4 25  67 17 11 3 11  ns  21 60 19  30 53 17  21 59 20  32 52 17  22 54 24  31 49 20  ns  30 70  78 22  31 69  76 24  30 70  72 28  <.01  7 3 11 79  0 2 28 70  7 2 10 82  0 3 30 68  6 1 11 82  0 1 30 69  <.01  19 65 16  73 21 6  18 65 17  72 22 6  16 68 16  71 27 1  <.01  <.01 <.01  168  Pre-exclusions Demographic variable  High (141)  Post exclusions (Personality) High Low (114) (117)  Low (132)  Post exclusions (Genetics) High Low (81) (71)  Betweensport group  Country of recruitment (%)  Canada France European ancestry (%) Northern Southern Eastern Western Other (SE, SW, EW)  8 92  77 23  9 91  75 25  7 92  70 30  <.01  19 7 8 63 3  20 2 15 61 2  21 6 10 60 4  20 1 13 65 2  23 7 12 53 5  21 1 11 65 2  ns  Ethnicity (%) European 96 83 96 86 100 100 ns First Nations 0 2 0 1 Asian (Japan) 0 3 0 1 Asian (China) 0 4 0 3 South America 2 0 2 0 Other 2 8 2 9 Ability of most proficient sport Beginner 1 1 1a 0 1 0 <.01 Novice 1 10 2a 7 2 9 Intermediate 8 34 7 37 8 34 Advanced 19 36 18 36 17 40 Expert 71 19 70 20 72 17 Days per year at sport, M 136.8 134.9 142 136(103) 129(108) 128 (102) ns (SD) (113) (100) (118) Note. Sample sizes for each group are shown in the header row (n). p-values for between-sport group comparisons were obtained using t-tests for continuous variables and Chi square tests for categorical variables; ns = not significant. a  the three participants reporting “beginner” and “novice” ability were all skydivers who jumped an average of 60  times per year.  169  Table 8-3 Participation counts per high- and low-risk sports Participation counts per high- and low-risk sports High-risk sports n Most proficienta (n) Low-risk sports n Most proficienta (n) Adventure Racing 12 0 Athletics 47 11 BASE jumping 13 5 Badminton 10 2 BMX 11 0 Biathlon 1 0 Bungee jumping 8 0 Cross country skiing 24 4 Car racing 6 0 Curling 5 2 Cliff jumping 14 0 Cycling 64 5 Climbing (rock) 73 9 Dance/ballet 13 4 Dirt biking 15 0 Dragon boating 5 0 Freeride skiingb 79 44 Equestrian 2 1 Freeride snowboardingb 33 12 Figure skating 2 0 Freestyle skiing 4 4 Fishing 13 0 Ice Climbing 33 0 Golf 30 2 Kite surfing 13 3 Gym workouts 53 7 Luge 4 0 Gymnastics 6 1 Mountaineering (alpinism) 38 5 Iron Man 17 0 Mountain biking 57 2 Rollerblade 5 0 Paragliding 34 5 Rowing 19 5 Parkours (free running) 7 2 Running 83 30 Rappelling 47 0 Speed skating 13 0 Sailing 26 4 Swimming 64 16 Skateboarding 18 1 Tennis 26 5 Ski cross 9 0 Triathlon 34 16 Skydiving 46 20 Weight lifting 32 1 Snowboard cross 10 0 Yoga 48 4 Speed riding 17 3 Other 16 Speed skiing 8 0 Steep skiing ("pente raide") 17 4 Street luge 4 0 Surfing 41 8 Whitewater kayaking 13 0 Windsurfing 1 2 Wing-suit flying 11 4 Other 2 Note. Many athletes reported participation in multi