1. Department of Preventive & Social Medicine Dunedin School of Medicine University of Otago Dunedin, New Zealand Professor Richie Poulton Director, Dunedin Multidisciplinary Health and Development Research Unit; Co-Director, National Centre for Lifecourse Research Gene-environment interplay and human behaviour: Implications for theory, research and practice

2. Background Human beings, their problems and their diseases are complex. Human beings, their problems and their diseases are complex. Most research aims to identify genes that relate directly to specific disorders. Most research aims to identify genes that relate directly to specific disorders. Results have been disappointing and inconsistent. Results have been disappointing and inconsistent.

4. 1.Identify a plausible combination of disorder, environmental risk factor and candidate gene to generate a hypothesis. 2.Test the interaction hypothesis between the gene, the environmental risk factor, and the disorder. 3.Replicate the findings with alternative measures of the disorder. 4.Check to see other genes are interacting with the gene being tested. 5.Test if the gene-environment interaction predicts other disorders. Five Steps

5. Retention in the Dunedin Study Age Year Number Percent* Birth1972-73 Birth1972-73 31975-761037100% 31975-761037100% 51977-78 991 96% 51977-78 991 96% 71979-80 954 92% 71979-80 954 92% 91981-82 955 92% 91981-82 955 92% 111983-84 925 90% 111983-84 925 90% 131985-86 850 82% 131985-86 850 82% 151987-88 976 95% 151987-88 976 95% 181990-91 993 97% 181990-91 993 97% 211993-94 992 97% 211993-94 992 97% 261998-99 980 96% 261998-99 980 96% 322004-05 972 96% 322004-05 972 96% * Percentage seen of those who were eligible (i.e. alive) at each age

6. Location of Study Members at age 32

7. Current research activities include studies of: SES inequalities - selection v causation SES inequalities - selection v causation Employment Employment Personality continuities across the life-course Personality continuities across the life-course Antisocial behaviour and criminality Antisocial behaviour and criminality Long-term consequences of child abuse Long-term consequences of child abuse Intergenerational relations (Study members & their parents) Intergenerational relations (Study members & their parents) Maori health/cultural identity Maori health/cultural identity Mental health (including substance abuse) Self-harm/coping Intimate relationships and domestic violence Dental health Sexual & reproductive health Cardiovascular risk factors Cardiovascular reactivity Respiratory functioning

8. Current research activities (contd) Blood based studies Blood based studies – HPV (wart virus) – Herpes immunity – Cardiovascular disease risk factors – IgE, alpha 1 antitrypsin Genetic studies –Mental health –Asthma/allergy –Cardiovascular risk factors Methodological studies Methodological studies – Comparison of Dunedin sample with national data – Attrition analyses

9. The cycle of violence Childhood maltreatment is a universal risk factor for antisocial behaviour increasing later criminality by about 50%. Childhood maltreatment is a universal risk factor for antisocial behaviour increasing later criminality by about 50%. But the majority of maltreated males do not become delinquents or criminals? But the majority of maltreated males do not become delinquents or criminals? Caspi et al (2002), Science, 297; 851-854

10. Genetic factors may act to increase (or decrease) environmental risk exposure Passive gene-environment correlations Passive gene-environment correlations Evocative gene-environment correlations Evocative gene-environment correlations

11. Genetic factors may act to modify the influence of environmental risks on psychiatric outcomes Genotypes may exacerbate reactions to environmental adversities Genotypes may exacerbate reactions to environmental adversities Genotypes may have a protective function Genotypes may have a protective function

12. In this study… We characterised genetic susceptibility to childhood maltreatment in terms of individual differences in a functional polymorphism in the promoter of the monoamine oxidase (MAOA) gene.

13. The MAOA gene is located on the X chromosome (Xp11.23-11.4). The MAOA gene is located on the X chromosome (Xp11.23-11.4). It encodes the MAOA enzyme, which metabolizes neurotransmitters such as norepinephrine (NE), serotonin (5-HT) and dopamine (DA), rendering them inactive. It encodes the MAOA enzyme, which metabolizes neurotransmitters such as norepinephrine (NE), serotonin (5-HT) and dopamine (DA), rendering them inactive. Genetic deficiencies in MAOA activity have been linked with aggression in mouse and man. Genetic deficiencies in MAOA activity have been linked with aggression in mouse and man. Deficient (i.e. “knockout” or low) MAOA activity may dispose the organism toward neural hyper- reactivity to threat. Deficient (i.e. “knockout” or low) MAOA activity may dispose the organism toward neural hyper- reactivity to threat. Why the MAOA gene?

14. The gene-maltreatment link Animal studies show that maltreatment stress in early life alters NE 5-HT, and DA neurotransmitter systems leading to an increase in aggression in “adulthood”. Animal studies show that maltreatment stress in early life alters NE 5-HT, and DA neurotransmitter systems leading to an increase in aggression in “adulthood”. In humans, altered NE and 5-HT activity is linked to aggressive behaviour and childhood maltreatment has lasting neurochemical correlates. In humans, altered NE and 5-HT activity is linked to aggressive behaviour and childhood maltreatment has lasting neurochemical correlates. Step 1

15. DNA samples obtained via blood (93%) and saliva (7%) DNA samples obtained via blood (93%) and saliva (7%) Study members were grouped as “High” MAOA activity (carrying the 3.5 or 4 repeat variants) and “Low” MAOA activity (carrying the 2, 3 or 5 repeat variants). Study members were grouped as “High” MAOA activity (carrying the 3.5 or 4 repeat variants) and “Low” MAOA activity (carrying the 2, 3 or 5 repeat variants). Ethnic stratification? Ethnic stratification? Testing Gene x Environment Interaction in a Cohort Design

16. Childhood maltreatment ages 3-11 years Observations of mother’s behaviour (age 3) Observations of mother’s behaviour (age 3) Parental reports of harsh discipline (ages 7-9) Parental reports of harsh discipline (ages 7-9) Multiple caregiver changes (through age 11) Multiple caregiver changes (through age 11) Physical harm (retrospective report) Physical harm (retrospective report) Unwanted sexual contact (retrospective report) Unwanted sexual contact (retrospective report)

17. Childhood maltreatment (ages 3 -11 years) For each Study child, we derived a cumulative exposure index by summing the number of maltreatment experiences reported during the first decade of life. %

18. % of study members Exposure to childhood maltreatment as a function of MAOA activity Childhood maltreatment

19. Genetic status: High vs low MAOA activity Genetic status: High vs low MAOA activity Severity of childhood maltreatment, between ages 3-11 years Severity of childhood maltreatment, between ages 3-11 years Evidence for gene x environment interaction would be found if the MAOA gene modified the influence of maltreatment on children’s development of severe antisocial behaviour. Testing Gene x Environment Interaction in a Cohort Design

20. Diagnosis of conduct disorder in adolescence According to criteria of DSM-IV According to criteria of DSM-IV Based on diagnostic interviews with Study members (at ages 11, 13, 15, 18) and checklists completed by their parents and teachers Based on diagnostic interviews with Study members (at ages 11, 13, 15, 18) and checklists completed by their parents and teachers At each assessment age, symptom information was gathered with reference to the "past 12 months" At each assessment age, symptom information was gathered with reference to the "past 12 months" 28% of the Study males were diagnosed with CD at one or more assessment ages 28% of the Study males were diagnosed with CD at one or more assessment ages

21. n = 108 42131807920 Conduct Disorder (ages 10-18) by MAOA gene activity and childhood maltreatment Conduct disorder (%) Low MAOA activity High MAOA activity p<.001 p=.15 No maltreatment Some maltreatment High maltreatment Step 2

22. Personality disposition toward violence (self-report at age 26 years) Sometimes I daydream about injuring or hurting someone Sometimes I daydream about injuring or hurting someone My temper is quick and hot My temper is quick and hot When I get angry, I fly off the handle before I know it When I get angry, I fly off the handle before I know it When I get angry I am ready to hit someone When I get angry I am ready to hit someone I admit that I sometimes enjoy hurting someone physically I admit that I sometimes enjoy hurting someone physically I enjoy a good brawl I enjoy a good brawl Sometimes I hit people who have done something to deserve it Sometimes I hit people who have done something to deserve it

23. n =10842131807920 Aggressive personality (age 26) by MAOA gene activity and childhood maltreatment Disposition toward violence (z scores) Low MAOA activity High MAOA activity p=.015 p=.26 No maltreatment Some maltreatment High maltreatment

24. Antisocial personality disorder symptoms (informant report at age 26 years) Antisocial personality disorder symptoms (informant report at age 26 years) Has problems controlling anger Has problems controlling anger Blames others for own problems Blames others for own problems Does not show guilt after doing something bad Does not show guilt after doing something bad Impulsive, rushes into things without thinking Impulsive, rushes into things without thinking Good citizen (reversed) Good citizen (reversed) Does things against the law Does things against the law Gets into fights Gets into fights

25. Antisocial Personality Disorder symptoms (z scores) n =107 39 12 171 74 18 Low MAOA activity High MAOA activity No maltreatment Some maltreatment High maltreatment Antisocial Personality Disorder symptoms (informant reports) by MAOA gene activity and childhood maltreatment p=.001p=.19

26. Conviction of a violent crime by age 26 Court records searched for all Study members Court records searched for all Study members 25% of Study males received 987 criminal convictions 25% of Study males received 987 criminal convictions 11% of Study males received 172 convictions for violent crimes 11% of Study males received 172 convictions for violent crimes Common assault Assault with intent to injure Manslaughter Rape Indecent assault on female Aggravated cruelty to animal

27. Violent convictions to age 26 by MAOA gene activity and childhood maltreatment p=.24 p<.001 Convicted for violent offence (%) n =1084213 1807920 Low MAOA activity High MAOA activity 0 10 20 30 40 50 60 No maltreatment Some maltreatment High maltreatment Step 3

28. Multiple Indicator Model of Antisocial Behaviour Antisocial Behaviour χ 2 (2) = 2.56, p=.28, CFI=.99, RMSEA=.02 Diagnosis of conduct disorder Violence conviction Informant reports of ASPDSx Self-reports of disposition towards violence.74.69.70.64

29. The Interaction between MAOA and Childhood Maltreatment Composite Index of Antisocial behaviour (z scores) Childhood maltreatment Low MAOA activity High MAOA activity

30. Employment problems (e.g. being laid off) Money problems (e.g. heavy debt) Relationship problems (e.g. break-up) Health problems (e.g. disabling injury) Housing problems (e.g. homelessness) Depression Adult Stress Load Caspi et al (2003). Science, 301; 386-389

31. In this study… We characterised genetic susceptibility to stressful life events in terms of individual differences in a functional polymorphism of the serotonin promoter (5-HTT) gene.

32. Why focus on 5-HTT gene? Pharmacological evidence Pharmacological evidence Experimental evidence from animal studies Experimental evidence from animal studies Neuroimaging evidence Neuroimaging evidence Step 1

33. The short ‘s’ allele has lower basal transcriptional activity than the long ‘l’ allele. The short ‘s’ allele has lower basal transcriptional activity than the long ‘l’ allele. Study members were grouped into 3 groups: s/s (n=146, 17%), s/l (n=435, 52%), and l/l (n=264, 31%). Study members were grouped into 3 groups: s/s (n=146, 17%), s/l (n=435, 52%), and l/l (n=264, 31%). Stressful life events between ages 21-26 years. Stressful life events between ages 21-26 years. Evidence for gene x environment interaction would be found if the 5-HTT gene modified the influence of stressful life events on depression. Evidence for gene x environment interaction would be found if the 5-HTT gene modified the influence of stressful life events on depression. Testing gene x environment interaction in a cohort design

34. The association between stressful life events and self-reports of depression symptoms at age 26, as a function of 5-HTTLPR Step 2 No. of stressful life events Self reports of depression symptoms, age 26 s/s s/l l/l

35. The association between stressful life events and informant-reports of depression at age 26, as a function of 5-HTTLPR No. of stressful life events Informant reports of depression, age 26 s/s s/l l/l

36. The association between stressful life events and a major depression episode at age 26, as a function of 5-HTTLPR No. of stressful life events Probability of major depression episode, age 26 s/s s/l l/l

37. The association between stressful life events and suicide attempt/ideation at age 26, as a function of 5-HTTLPR Step 3 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 01234 or more s/s No. of stressful life events Probability of suicide ideation/attempt, age 26 s/l l/l

38. How do we know this is a G x E, and not a G x “G” interaction? Genes can cause people to select themselves into stressful situations Exposure to life events is influenced by genetic factors Life Stress Genes Environment

39. G x E or G x “G”? No correlation between 5-HTT gene and exposure to life stress Postdiction analyses: If life events are a genetic marker for depression, we should detect a 5-HTT gene x “G” interaction when postdicting depression that occurred prior to the life events

40. Is the genetic moderation specific to adult stress? If 5-HTT gene moderates the depressogenic influence of stressful life events, it should moderate the influence of stressful experiences that occurred not just in adulthood, but also of experiences that occurred in earlier developmental periods.

41. The association between childhood maltreatment and adult depression, as a function of 5-HTTLPR Childhood maltreatment Probability of major depression episode, ages 18-26 s/s s/l l/l

42. The moderation of life stress on depression was specific to a polymorphism in the 5-HTT gene because this effect was observed regardless of the individual’s MAOA gene status. Step 4 – genotype specificity

43. Step 5 – outcome specificity The relationship was specific to depression and not other psychiatric outcomes in adulthood such as anxiety or antisocial behaviour.

44. Innovation? Vulnerability to adversities is conditional upon genetic susceptibility factors on which individuals vary, i.e. nature via nurture. Vulnerability to adversities is conditional upon genetic susceptibility factors on which individuals vary, i.e. nature via nurture. To find associations between genes and behaviours, measure the participants’ developmental stress history. To find associations between genes and behaviours, measure the participants’ developmental stress history. The riddle of genetic penetrance. The riddle of genetic penetrance. QTL effects may not be so small after all. QTL effects may not be so small after all.

45. Research implications Genes are assumed to create vulnerability to disease, but from an evolutionary perspective they are equally likely to protect against environmental insult. Genes are assumed to create vulnerability to disease, but from an evolutionary perspective they are equally likely to protect against environmental insult. Bivariate association (main effect) studies are not the way forward. Two-way, three-way, four- way… interactions are the future. Bivariate association (main effect) studies are not the way forward. Two-way, three-way, four- way… interactions are the future.

46. Intervention implications Pharmacological and genetic treatments remain a long way off. Old fashioned fixes are supported. Pharmacological and genetic treatments remain a long way off. Old fashioned fixes are supported. Population screening inappropriate. Population screening inappropriate. Impact upon (↑or↓) a person’s sense of responsibility for their own wellbeing and behaviour. Impact upon (↑or↓) a person’s sense of responsibility for their own wellbeing and behaviour.

47. Legal implications Courts – Is there a case for mitigation in sentencing, i.e. lighter versus harsher sentencing linked to genes. Courts – Is there a case for mitigation in sentencing, i.e. lighter versus harsher sentencing linked to genes. Discrimination by insurance companies or employers? Require legislation to prevent this. Discrimination by insurance companies or employers? Require legislation to prevent this.

48. Are designer babies inevitable? Are designer babies inevitable? Pre-implantation and pre-natal diagnosis for positive and/or negative attributes. Pre-implantation and pre-natal diagnosis for positive and/or negative attributes. Social implications

49. Moral/ethical implications Reducing versus increasing stigmatisation. Reducing versus increasing stigmatisation. Social inequalities are not justified by the Bell Curve. Social inequalities are not justified by the Bell Curve. “There is a special obligation to guard against allowing research aimed at increasing knowledge and reducing suffering from being highjacked by the desire to justify the status quo.” (Parens, 2004)

50. Acknowledgements This on-going research would not have been possible without the co-operation and commitment of the Study members, their families and friends over a long period of time. This on-going research would not have been possible without the co-operation and commitment of the Study members, their families and friends over a long period of time. Core funding for the Dunedin Multidisciplinary Health and Development Research Unit comes from the Health Research Council of New Zealand. Core funding for the Dunedin Multidisciplinary Health and Development Research Unit comes from the Health Research Council of New Zealand. For copies of research articles referred to in this presentation or other information on the Study, contact Michelle McCann:  03 479-8507  email: dmhdru@otago.ac.nz  03 479-8507  email: dmhdru@otago.ac.nzhttp://dunedinstudy.otago.ac.nz