(Advances in Development and Psychopathology_ Brain Research Foundation Symposium Series 2) Patrick H. Tolan, Bennett L. Leventhal (eds.)-Gene-Environment Transactions in Developmental Psychopathology.pdf

Advances in Development and Psychopathology:
Brain Research Foundation Symposium Series
PatrickH.Tolan
BennettL.Leventhal
Editors
Gene-Environment
Transactions in
Developmental
Psychopathology
The Role in Intervention Research
Advances in Development and Psychopathology:
Brain Research Foundation Symposium Series
Series Editors: Patrick H.Tolan and Bennett L.Leventhal
More information about this series at http://www.springer.com/series/8544

Patrick H. Tolan Bennett L. Leventhal
Editors
Gene-Environment
Transactions in Developmental
Psychopathology
The Role in Intervention Research
Editors
Patrick H. Tolan Bennett L. Leventhal
Youth-Nex Center Department of Psychiatry
University of Virginia University of California at San Francisco
Charlottesville, VA, USA San Francisco, CA, USA
Advances in Development and Psychopathology: Brain Research Foundation

Symposium Series
ISBN 978-3-319-49225-4ISBN 978-3-319-49227-8(eBook)
DOI 10.1007/978-3-319-49227-8
Library of Congress Control Number: 2017932277
Springer International Publishing AG 2017

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of
the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation,
broadcasting, reproduction on microfilms or in any other physical way, and transmission or information
storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology
now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication
does not imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this book
are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the
editors give a warranty, express or implied, with respect to the material contained herein or for any errors
or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims
in published maps and institutional affiliations.
Printed on acid-free paper
This Springer imprint is published by Springer Nature

The registered company is Springer International Publishing AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Contents
1 What Can andShould BetheRole ofIntervention Studies

inGEX Research?.................................................................................... 1
Patrick H. Tolan and Bennett L. Leventhal
Part I Gene-Environment Transaction Framework

2 Genetics, Behavioral Intervention, andHuman Development............ 9
David Reiss
3 GeneEnvironment Interaction intheBehavioral Sciences:
Findings, Challenges, andProspects...................................................... 35
Matt McGue and Bridget E. Carey
4 Epigenetics andtheBiology ofGene Environment Interactions...... 59
Michael J. Meaney
5 Using Genetically Informed Designs toUnderstand
theEnvironment: TheImportance ofFamily-Based Approaches....... 95
Chang Liu and Jenae M. Neiderhiser
6 GeneEnvironment Correlation asaSource ofStability
andDiversity inDevelopment................................................................. 111
Christopher R. Beam and Eric Turkheimer
7 Sampling intheExamination ofGene-Environmental
Interactions Within aNeurodevelopmental Disorder Framework..... 131
Young Shin Kim
Part II Application to Developmental Psychopathology

8 Do Gene-by-Environment Interactions Offer Potential
Intervention Strategies inAnxiety Disorders?...................................... 147
Neal Ryan
v
vi Contents
9 Challenge and Potential for Research on Gene-Environment

Interactions in Autism Spectrum Disorder............................................ 157
Carly L.A. Wender and Jeremy Veenstra-VanderWeele
10 How Can GxE Research Help Prevent theDevelopment
ofChronic Physical Aggression?............................................................ 177
Richard E. Tremblay, Linda Booij, Nadine Provenal,
and Moshe Szyf
Part IIIIntervention Research: Implications for Gene

Environment Transaction Framework
11 Using Genetically Informed Prevention Trials toTest
GeneEnvironment Hypotheses........................................................... 211
Gene H. Brody
12 Developmental Mechanisms inGene-by-Intervention (GxI)
Effects intheFast Track Trial................................................................. 235
Kenneth A. Dodge
13 Applying Behavioral Genetics Research toInform
thePrevention ofDevelopmental Psychopathology:
Drawing fromthePrinciples ofPrevention Science............................. 251
Leslie D. Leve
14 Challenges forIntervention Research Within theGEX
Framework............................................................................................... 283
Patrick H. Tolan and Bennett L. Leventhal
Index.................................................................................................................. 295
Contributors
ChristopherR.Beam University of Southern California, Los Angeles, CA, USA

LindaBooij Concordia University, Montreal, QC, Canada
GeneH.Brody Center for Family Research, University of Georgia, Athens, GA,
USA
Bridget E. Carey Department of Psychology, University of Minnesota,
Minneapolis, MN, USA
KennethA.Dodge Duke University, Durham, NC, USA
YoungShinKim Langley Porter Psychiatric Institute, University of California at
San Francisco, San Francisco, CA, USA
LeslieD.Leve University of Oregon, Eugene, OR, USA
BennettL. Leventhal Department of Psychiatry, University of California at San
Francisco, San Francisco, CA, USA
Chang Liu Department of Psychology, The Pennsylvania State University,
University Park, PA, USA
MattMcGue Department of Psychology, University of Minnesota, Minneapolis,
MN, USA
MichaelJ.Meaney Department of Psychiatry, Ludmer Centre for Neuroinformatics
and Mental Health and Sackler Program for Epigenetics and Psychobiology at
McGill University, Douglas University Mental Health Institute, McGill University,
Montreal, QC, Canada
Jenae M. Neiderhiser Department of Psychology, The Pennsylvania State
University, University Park, PA, USA
NadineProvenal Max Planck Institute of Psychiatry, Munich, Germany
vii
viii Contributors
DavidReiss Yale School of Medicine, Child Study Center, New Haven, CT, USA
NealRyan University of Pittsburgh, Pittsburgh, PA, USA
MosheSzyf McGill University, Montreal, QC, Canada
PatrickH.Tolan Youth-Nex Center, University of Virginia, Charlottesville, VA,
USA
RichardE.Tremblay University College Dublin, Dublin, Ireland
GRIP, Universit de Montral, Montreal, QC, Canada
Eric Turkheimer Department of Psychology, University of Virginia,
Charlottesville, VA, USA
JeremyVeenstra-VanderWeele Department of Psychiatry and Sackler Institute
for Developmental Psychobiology, Columbia University Medical Center, New
York, NY, USA
Mortimer D.Sackler Associate Professor of Psychiatry, New York State Psychiatric
Institute, New York, NY, USA
Center for Autism and the Developing Brain, New York Presbyterian Hospital, New
York, NY, USA
Carly L.A. Wender Department of Psychiatry, Columbia University Medical
Center, New York, NY, USA
Chapter 1
What Can andShould BetheRole
ofIntervention Studies inGEX Research?
PatrickH.Tolan andBennettL.Leventhal
This volume offers the presentations and related commentary from a multidisci-
plinary conference entitled: Gene-Environment Transactions in Developmental
Psychopathology: Role in Intervention Research. The meeting was held in Chicago
on September 10 and 11, 2014 under the auspices of the Brain Research Foundation.
This working conference was the second in the series of efforts designed to examine
substantive scientific and methodological issues organized to utilize a developmen-
tal psychopathology framework to identify important scientific topics and to focus
on substantive and methodological issues in advancing knowledge and practice
related to that topic. The participants were asked to join in this effort to identify
novel opportunities for scientific and clinical advances by exploring translational
science and its ability to broaden our knowledge base, pursue novel methods, and
set an agenda for the next phase in the study of developmental psychopathology.
Developmental psychopathology is the broad context for this discourse and,
hence, this volume. For the purposes of this discussion, developmental psychopa-
thology is a construct that is characterized by transactions between individuals and
the environment over the life course which contribute to variations in biological and
psychological processes that adversely affect functioning and lead to
psychopathology.
As with the prior conference, leading scientists from diverse backgrounds and
with widely varying areas of expertise were assembled to offer position statements
P.H. Tolan (*)

Youth-Nex Center, University of Virginia, 405 Emmet Street South, Charlottesville, VA
22904-4281, USA
e-mail: pht6t@virginia.edu
B.L. Leventhal
Department of Psychiatry, University of California at San Francisco,
Box 0984-CAS, Room LP-152, 401 Parnassus Avenue, San Francisco, CA 94143-0984, USA
e-mail: Bennett.Leventhal@ucsf.edu
Springer International Publishing AG 2017 1

P.H. Tolan, B.L. Leventhal (eds.), Gene-Environment Transactions
in Developmental Psychopathology, Advances in Development
and Psychopathology: Brain Research Foundation Symposium Series 2,
DOI10.1007/978-3-319-49227-8_1
2 P.H. Tolan and B.L. Leventhal
and discuss questions critical to advancing the science and practice of developmental
psychopathology. At the working sessions of the conference, attendees addressed
approaches, new findings, continuing challenges, and promising new directions; they
were shared, critiqued, and debated. Moreover, the presence of experts from diverse
disciplines helped to provide unique perspectives on developing intersections as well
as divergences within the work. With this information, participants were able to
refine and formulate bridges across different areas of the work presented, while pro-
viding insight into promising new directions. Ultimately, it is our collective hope and
ambition that the conference and this volume offer important next steps for research
and theory development that will expedite understanding and potential actions to
effect new and different approaches to developmental psychopathology.
The specific goal for this seminar was to probe the role and utility of intervention
research and how it can expand knowledge about Gene-Environment Transactions
(GEX) and, conversely, how this framework might inform intervention studies.
Intervention or experimental manipulation research is fundamental to scientific
study across topics and disciplines as experiments permit valid causal inference. As
acknowledged for over a century, precisely constructed and carefully executed
experimental studies provide the most convincing test of theorized causal processes
(Radder, 2009). With appropriate controls, precise formulations, and absence of
bias in comparison conditions, experiments (unlike observation studies, epidemio-
logical tracking or passive longitudinal studies) permit the conclusion that the rela-
tion between the manipulated variable and the dependent variable is causal; the
manipulated variable is seen as controlling the occurrence, level, or influence of the
dependent variable. Moreover, differential reactions to interventions can suggest
differential susceptibility (genetic, environmental and, in most cases, combinations
of these contributors). This unique capability means that intervention studies are an
important component of any substantial advancement in scientific knowledge. How
and when this utility is best applied is often controversial; it rests on work that sen-
sitively and precisely considers how major precepts, assumptions, and framework
implications impact what is to become valid and useful research.
In several chapters of this volume, the question is less about the comparative value
or whether or not intervention research is essential for understanding gene-environment
transactions, rather it is about what level of knowledge accumulation and precision of
understanding is needed for a sound experimental study to follow after substantial
observational and pattern tracing studies. At present, the field is characterized by rap-
idly developing understanding of increasing complexity about how gene and environ-
ment confluences and interdependencies act in developmental psychopathology. In
addition, there is appropriate concern about interventions that are misdirected or
poorly informed as they may mislead investigators about the roles of genes and envi-
ronment and their relation in the development and maintenance of psychopathology.
For some, it can be argued that useful intervention studies may not yet be identifiable.
However, as is shown here, the time has come to discuss how these studies might be
constructed and what questions they can address. This volumes content suggests that
in order to be prepared for next steps, there is need for immediate attention to these
issues so that powerful scientific tools can be a ppropriately applied when warranted.
The time for this work is upon us, as examples in this volume illustrate.
1 What Can andShould BetheRole ofIntervention Studies inGEX Research? 3
A second major focus of these discussions includes the notion of interventions

within a GEX framework for developmental psychopathology; there is the increas-
ing interest in applying this framework to understanding effects of psychosocial
interventions. Such a framework considers genetic variations as moderators of
intervention effects and of mediation of effects by theorized genetic processes.
Increasingly, clinical interventions, across the spectrumfrom promoting healthy
development to prevention of disorder to treating manifest psychopathologyare
tracking genetic variation and intervention impact on specific genes and/or patterns
of genetic variance. The potential in this work not only suggests biological mecha-
nisms of effects of such interventions, but also offers opportunities to explicate how
genetic variation might relate to susceptibility to disorders as well as the potential
benefit from interventions. In addition, mediational models can test theorized pro-
cesses of gene expression, environmental dependence, and pathways from potential
risk to likely expression of psychopathology. For example, as suggested in the chap-
ters by Tremblay, Booij, Provenal and Szyf (2017), and Brody (2017), interven-
tions are being studied to test theorized processes of gene expression that may be
modified by intervention. Dodge (2017) and Leve (2017) provide informative dis-
cussions of how preventive interventions can provide information about the validity
of theories of gene-environment transactional developmental models. Intervention
studies that include a G-E transaction frame have great potential to address two
critical intervention research questions: (1) for whom is the intervention beneficial
and (2) how does this intervention have its effects?
Both questions are essentially about utilizing interventions to inform develop-
mental psychopathology and to test causal relationships by attempting to change
one variable that is thought to be under the control of another. As with other experi-
mental methods, one important prerequisite for either question is to adequately
develop and specify theoretical models as well as developing prior descriptive,
empirical studies that support specific valid experiments (see McGue & Cariey,
2017; Reiss, 2017). A second prerequisite is the identification of thoughtful, reli-
able, and replicable experimental methods and appropriate designs. Experimental
trials can provide some of the most reliable and robust bases for causal inferences
about developmental psychopathology if formulation is sensitive and sophisticated;
if they are fully informed by prior descriptive work, and if the studies are well
designed and executed. Whether occurring at the micro-level to test specific theo-
rized developmental processes or as clinical interventions thought to affect psycho-
pathology, these studies provide efficient and more authoritative tests of critical
theoretical contentions. This means that such findings may point directly to practi-
cal actions that can be taken to prevent or treat psychopathology and refine under-
standing of multiple correlations. When applied within preventive and treatment
interventions intervention trials can test likely benefits of a given intervention for
groups, based on environmental or gene variants, suggesting better methods for
targeting or personalizing interventions. Within the developmental psychopathol-
ogy framework, these processes are transactions between the organism at a given
point in development and reflect environmental influences on that development with
genetic contributions potentially occurring through multiple systems.
ene-Environmental Transaction Framework

G
forIntervention Research
The general framework for most genetic studies within developmental psychopa-
thology focuses on the intersection or transactions between genetic variations and
environmental exposures. The term initially used to describe these processes was
gene-environment interactions, with the simple model of greater genetic suscepti-
bility (variation) to a given disorder or endophenotypes (symptom of a disorder;
specified expression of genetic influence) and exposure to the promotive or precipi-
tating environment, resulting in differential psychopathology expression. Early
behavioral genetic studies showed that variations in symptoms of psychopathology
could be traced to differential genetic patterns in reaction to the same environmental
risk factor as well as to differences in the likelihood of inheriting the genetic makeup
that correlated with a greater probability of developing specific pathological symp-
toms and syndromes of developmental psychopathology (Meaney, 2001; Meaney,
2017; Moffitt, Caspi, & Rutter, 2005). Advances in technology allowed for scanning
and probing of the genome; modeling multi-gene interactions; and tracking gene
expression. As a result of this progress, contributions to risk gene-environment rela-
tions became multitudinal. Thus, arose the need for a more general term for gene-
environment transactions. Several chapters in this volume explain gene-environment
correlations and specify when a relationship does not simply act as a moderator,
hence becoming less clearly determinable, and perhaps less important (to make
such categorical distinctions of gene-environment relations). These specifics are
being subsumed under a broader framework of interesthow patterns of genetic
interactions and gene-environment transactions summatively produce a definable
outcome. Similarly, the framework expands to include multiple forms of interde-
pendency across development and help to explain the evolution of pathology. Thus,
as is illustrated in this volume, more complex GEX models with greater specifica-
tion of component contributors, including greater attention to gene-environment
correlations, expressive variations across development, attention to differences in
gene development and expression, and exploration of epigenetic processes are
emerging that can inform about psychopathology (Lester, Conradt, & Marsit, 2016).
These developments have launched a rapidly growing number of studies and an
equally impressive rate of development in the sophistication of studies and in the
variation in focus of gene-environmental transaction studies (Keating, 2016; Lester
etal., 2016); transforming the scientific discourse and modifying almost every theo-
retical framing of developmental psychopathology.
This volume is organized into three sections. The first focuses on key concepts
and issues in current understanding of a gene-environment transaction framework for
developmental psychopathology. It examines the implications for the role of inter-
vention research and how this framework might be applied for intervention studies.
This includes examinations of technical and methodological challenges. The second
section consists of applications of a gene-environment transaction framework to
three prevalent disorders: anxiety disorder (Ryan, 2017), physical aggression/con-
1 What Can andShould BetheRole ofIntervention Studies inGEX Research? 5
duct disorders (Tremblay, Booij, Provencal, & Szyf, 2017), and autism spectrum
disorders (Wender & Veenstra-Vander Weele, 2017). The final section provides
reports about application of gene-environment framework and hypotheses through
intervention studies. Collectively, these chapters provide a deeply informative and
broadly useful mapping of the current understanding of and potential for intervention
research for developmental psychopathology, suggest some new and potentially very
valuable approaches, and outline several pathways forward for the field.
Within the first hours of our discussions, one of our colleagues remarked about
how much they were learning and how that was not only personally satisfying but
also that this steep learning curve was reflective of the state of the field. This volume
reflects that there is much promise and much to learn about the role of intervention
research in GEX approaches to developmental psychopathology. We are very
pleased to have been part of this work and to have had the opportunity to learn
together and from our generous and capable colleagues. We are confident that each
chapter will provide a great learning opportunity and, collectively, they can move
forward our capability to understand and ultimately prevent major forms of devel-
opmental psychopathology.
References
Brody, G. (2017). Using genetically informed prevention trials to test gene environment hypoth-
eses. In P.H. Tolan, & B.L. Leventhal (Eds.), Gene-environment transactions in developmental
psychopathology: Role of intervention research. NewYork: Springer.
Dodge, K.A. (2017). Developmental mechanisms in gene by intervention (GI) effects. In P.H.
Tolan, & B.L. Leventhal (Eds.), Geneenvironment transactions in developmental psychopa-
thology: Role of intervention research. NewYork: Springer.
Keating, D.P. (2016). Transformative role of epigenetics in child development research: Commentary
on the special section. Child Development, 87(1), 135142. doi:10.1111/cdev.12488.
Lester, B.M., Conradt, E., Marsit, C. (2016). Child Development, 87(1), 2937. Hoboken, NJ:
Wiley. doi:10.1111/cdev.12423
Leve, L. (2017). Applying behavioral genetics research to inform the prevention of developmental
psychopathology: Drawing from the principles of prevention science. In P. H. Tolan & B. L.
Leventhal (Eds.), Advances in development and psychopathology. Brain Research Foundation
Symposium Series, Volume II: Gene-environment transactions in Developmental
Psychopathology: Role of intervention research. New York: Springer.
McGue, M., & Cariey, B. E. (2017). Geneenvironment interaction in the behavioral sciences:
Findings, challenges, and prospects. In P.H. Tolan, & B.L. Leventhal (Eds.), Geneenviron-
ment transactions in developmental psychopathology: Role of intervention research. NewYork:
Springer.
Meaney, M.J. (2001). Maternal care, gene expression, and the transmission of individual differ-
ences in stress reactivity across generations. Annual Review of Neuroscience, 24, 11611192.
doi:10.1146/annurev.neuro.24.1.1161.
Meaney, M.J. (2017). Epigenetics and the biology of gene environment interactions. In P.H.
Tolan, & B.L. Leventhal (Eds.), Geneenvironment transactions in developmental psychopa-
thology: Role of intervention research. NewYork: Springer.
Moffitt, T.E., Caspi, A., & Rutter, M. (2005). Strategy for investigating interactions between mea-
sured genes and measured environments. Archives of General Psychiatry, 62(5), 473481.
Radder, H. (2009). The philosophy of scientific experimentation: A review. Automated

Experimentation, 1, 2. doi:10.1186/1759-4499-1-2.
Reiss, D. (2017). Genetics, behavioral intervention and human development. In P. H. Tolan, &
B. L. Leventhal (Eds.), Geneenvironment transactions in developmental psychopathology:
Role of intervention research. NewYork: Springer.
Ryan, N. (2017). Do gene by environment interactions offer potential interventions strategies in
anxiety disorders? In P.H. Tolan, & B.L. Leventhal (Eds.), Geneenvironment transactions in
developmental psychopathology: Role of intervention research. NewYork: Springer.
Tremblay, R.E., Booij, L., Provencal, N., & Szyf, M. (2017). How can GE research help prevent
the development of chronic physical aggression? In P. H. Tolan, & B. L. Leventhal (Eds.),
Geneenvironment transactions in developmental psychopathology: Role of intervention
research. NewYork: Springer.
Wender, C.L. A., & Veenstra-Vander Weele, J.(2017). Challenge and potential for research on
geneenvironment interactions in Autism Spectrum Disorder. In P.H. Tolan, & B.L. Leventhal
(Eds.), Geneenvironment transactions in developmental psychopathology: Role of interven-
tion research. NewYork: Springer.
Part I
Gene-Environment Transaction
Framework
Chapter 2
Genetics, Behavioral Intervention,
andHuman Development
DavidReiss
This book explores the role of new findings in genetics to more fully understand
development across the life span. It gives equal attention to the role of genetic stud-
ies in identifying new targets for interventions and the sources of individual differ-
ences among individuals in their response to interventions. In the last decade, there
have been a number of publications that have reawakened interest in this interplay
between genetics and behaviorally relevant interventions but this interplay began
almost as soon as there were tools to study genetic influences on behavior.
The ancients had already mastered practical genetics through their artful use of
selective breeding. They learned that complex patterns of behavior were highly
heritable. For example, corgis were bred as cattle herders and could be relied on to
circle their herd and barking to keep them together; indeed, this trait readily gener-
alizes to herding humans who are also herded when hiking, walking, or even swim-
ming. Selective breeding is central to Darwins Variations in Plants and animals
under domestication (Darwin, 1920 (1868)) who notes the accomplishments of
Plato, Alexander the Great, and Virgil.
Scientific genetics began with Mendels famous pea experiments. Though the
main results were published in a well-known journal in 1866 (Mendel, 1948 (1866)),
they were little noticed until highlighted by the German botanist, Correns (1950), in
1900. Mendelian patterns of inheritance became widely explored and understood in
many areas of science.
Among its most notable applications to human development was the discovery of
phenylketonuria by Asbjorn Folling in 1932 in two siblings afflicted with severe
developmental delays and mental retardation (Centerwall & Centerwall, 2000).
D. Reiss (*)
Yale School of Medicine, Child Study Center,
P.O.Box 207900, New Haven, CT 06520, USA
e-mail: david.reiss@yale.edu

DOI10.1007/978-3-319-49227-8_2
10 D. Reiss
Folling was born and raised in rural Norway but unusual among his rural peers in
those days, achieved entry into a technical college where he studied chemistry and
then completed medical training at the University of Oslo. Folling was the last sta-
tion in an agonizing search by the childrens parents, Borgny and Harry Egeland, for
the causes of their childrens severe behavioral problems. Folling has described how
he added ferric chloride to these childrens urineas a test for ketones ordinarily
found in diabetesand watching it turn green, a reaction he had never seen. Folling
used his background in chemistry to identify the abnormal urinary substance as
phenyl pyruvic acid. When he found eight additional cases after a survey of 400
institutionalized children, he published his results in 1935. Folling understood
Mendelian genetics well enough to deduce that the parents must be heterozygote
carriers and indeed published the first report on how to detect their status with a
phenylalanine challenge test (Sydnes & Flling, 1962).
The most pivotal work, relevant to this chapter, subsequent to Follings original
discovery, was the testing of a low phenylalanine diet by Horst Bickel a German
trained physician who was training as pediatrician in Birmingham University (UK)
Childrens hospital. In a single 2-year-old girl, he reduced and then augmented phe-
nylalanine in her diet and demonstrated the partial reversibility of the somnolent
retardation that was well established in this toddler (Bickel, 1953). The experiment
can be witnessed on You Tube (https://www.youtube.com/watch?v=-rs0iZW0Lb0).
Follings ferric chloride test was used for some time and could reveal the disease if
applied to a babys diaper; however, it was often not positive until well after the
typical newborn left the hospital and might be lost to screening. Therefore, the final
step in this storyfor our purposesis the perfection of a blood test for PKU that
would reveal the disease in newborns while they were still in the hospital for screen-
ing. Robert Guthrie a physician and microbiologist, working at Buffalo Childrens
Hospital, took advantage of the singular ability of phenyl pyruvic acid to foster the
growth of Bacillus subtilus in a restricted culture medium (Guthrie & Susi, 1963).
The availability of highly specific and sensitive test as well as an effective dietary
treatment, if started very early, led to legislation in all 50 states and the District of
Columbia requiring all hospital-born infants to be screened.
Both Gregor Mendel and Asbjorn Folling were, in some sense, fortunate. Their
agricultural and clinical experiments focused on a highly heritable phenotype where
one gene determined one distinctive phenotype. In Follings case, the gene was not
expressed in the exceptionally complex organ of the brain but chiefly in the liver.
Moreover, a straightforward environmental treatment, diet, could fully compen-
sateor nearly sofor the diseases prime deficiency.
Implicitly, the PKU success story has set high hopes that the genetic etiology of
a range of disorders is that are more complex that PKU will yield a set of biological
targets that could be discovered in practical programs of screening and that either
pharmacologic or behavioral interventions might either correct or compensate for
the basic deficiencies of the disorder. The genetic delineation of targets for interven-
tion will be one of the two themes of this chapter and we will return to it to survey
accomplishments thus far and anticipate the future.
2 Genetics, Behavioral Intervention, andHuman Development 11
The one-gene, one disorder paradigm has successfully uncovered other screen-
able and treatable inborn errors of metabolism but, in its simple form, is not appli-
cable to the complex syndrome of behavioral disorders or of behavioral health that
have been delineated over the decades (for the phenomenology and genetics of posi-
tive syndromes of well-being, see National Research Council, 2013; Rietveld etal.,
2013). However, another line of genetic research contributes to our understanding of
behavioral interventions. Ten years before Follings first Norwegian publication, a
graduate student of Lewis TermanCurtis Merriamwas the first to utilize the twin
method for estimating more global effects of genetic influence on human behavior;
his interest was in the full range of mental abilities rather than severe retardation.
However, Merriam lacked a secure method for distinguishing monozygotic from
dizygotic twins, a strategy that was developed by a German dermatologist Herman
Siemens (Rende, Plomin, & Vandenberg, 1990) who developed criteria for their dis-
tinction that presages those still used today (Siemens, 1927). Siemens was passionate
believer in heredity long before he first used twins to more precisely calibrate its
effects on both skin disease and behavior. In the deep dismay that swept post-World
War I Germany he inadvertently helped fan a growing enthusiasm for eugenics that
fueled a hope for the restoration of German racial superiority (Proctor, 1988). The
worldwide revulsion at this mixture of racism and genetics suppressed research on
the genetics of human behavior for a generation following World War II.1
The post-World War II revulsion against eugenics rendered all genetic research
on human intelligence and behavior suspect and added implicit weight to objections
to the twin method. Critics wedded with equal fervor to a fully environmentalist
position underscored the many ways in which MZ twins were treated more similarly
than DZ twins, hence upending the basic assumption of the method that the environ-
ments of MZ and DZ twins were equally correlated (the equal environments
assumption). However, a series of studies using twins reared apart served as a rigor-
ous test of this assumption and clarified the strengths of the twin method.2
Long before the controversy surrounding the twin method was resolved investi-
gators recognized that this method could help them learn more about behavioral
intervention. The first of these was Arnold Gesell who trained first as a psychologist
and then, while completing his medical degree at Yale, founded its Child Study
Center. Gesell recognized that he could use monozygotic twins to control the effects
of genetic influences not just on naturally occurring behavior as a phenotype but on
response to treatment as a distinctive phenotype. Thus, Gesell trained one member
of a toddler MZ pair to climb stairs and used the other as a control. Thus, he held
constant the effect of genetic influence on the childrens response to treatment. He
also held constant the effects of their correlated environments (e.g., their mothers
personality). In Gesells summary of this co-twin control method (Gesell, 1942), he
1
Siemens was appointed chair of dermatology at Leiden before the outbreak of World War II and,
objecting to Nazi occupation procedures, was jailed by the movement he inadvertently helped to
foster (Burgdorf, Bickers, & Hoenig, 2014).
2
For a full review of this controversy and its resolution, see Reiss (2016).
12 D. Reiss
clarified that his method could also be used to clarify the points in development in
which intervention might be most effective.
In sum, Folling used genetic inquiry to delineate a biochemically identified tar-
get for intervention. His specific discovery was to identify the urinary abnormalities
of his child subjects through chemistry alone. But his presumption that a single gene
caused the disorder and his knowledge that Mendelian genetics were crucial to his
identifying parents who were carriers. Gesell first formulated the idea that genetic
analysis could account for variations in response to treatment. Gesell chose to con-
trol these influences but later investigators chose to estimate more directly the
effects of genotypic variation on response to behavioral intervention.
The first study in this genre, following Gesell, was by Paul Fox and his col-
leagues using the widely publicized sample of twins reared apart that were recruited
worldwide to the University of Minnesota (see Bouchard, Lykken, McGue, Segal,
& Tellegen, 1990 for a summary of this study). As part of a lengthy assessment
period Fox observed his subjects learning to improve their performance on a pursuit
rotor task and correlated separately the performance of 64 MZ twins and 32 DZ
twins all reared apart (Fox, Hershberger, & Bouchard, 1996). The most critical com-
parisons were for slopes of performance across days; the slope reflects rate of
change in response to practice. For example, MZ twins had an intraclass correlation
of .56 for the magnitude of this slope and DZ twins a correlation of .24 suggesting
that heritability and sibling-specific environments accounted equally for difference
in rate of learning. What might have been the results had the twins been reared
together? Would there have been any evidence of a shared or between family effect
on rates of learning intervention and practice of the rearing environment the twins
had shared together? Follow-up studies have been surprisingly rare but two suggest
there would have been little difference between twins reared together and apart
(Missitzi etal., 2011, 2013). Experiments with children might show a more notable
effect of the shared rearing environment.
It is surprising how infrequently the twin design has been used to explore the
balance between environmental and genetic factors that contribute to success or
failure of behavioral interventions. Indeed, the twin method can be used not only to
clarify the balance between anonymous genetic and environmental influences but
also to track down, quite specifically, what those factors might be. For example, the
heritability of mens capacity to remain abstinentonce they commit to stopping
tobaccois highly heritable. Likewise, the intensity of nicotine withdrawal symp-
toms is also heritable though less so. Of special interest is that the genetic influences
for withdrawal and for difficulty in maintaining abstinence overlap (Xian, 2003)
providing not only an important clue for refining or improving cessation programs
but for searching for specific genes involved (Uhl etal., 2008).
To summarize, not long after developmental science acquired the tools for exploring
genetic influences, investigators explored two lines of inquiry exploiting those tools to
improve intervention. The research on PKU became a paradigm for the use of genetics
to better define targets for both preventive and therapeutic intervention. Heterozygote
parents, according to this vision, might be identified and provided genetic counseling,
and infants with both recessive genes could be economically identified shortly after
birth and provided with a specific environmental treatment whose mechanism of action
was clear compensatory rather curative, a distinction that remains important in both
pharmacological and behavioral therapeutics. The prescient work of Gesell and a gen-
eration later of Fox, Bouchard and their colleagues suggested a second line of work:
genetic analysis of variations in response to treatment. This work did not solidify into
a paradigm. The twin methodand its companion, the adoption methodare power-
ful and rarely used tools for understanding behavioral interventions. But, as we will
see, in unrelated developments, a growing interest in how specific genes moderate the
effects of environmental influences has reawakened an interest in genetic influences on
response to behavioral treatments. Likewise, the twin and adoption method has made a
steady, if unheralded, contribution to defining targets for behavioral intervention in
ways Gesell, Fox, and Bouchard could not have imagined. Each of these lines of inves-
tigation inevitably led to an expanded knowledge of developmental processes. Horst
Bickels 2-year-old PKU patient brightened up in response to reduce phenylalanine in
her diet: her face expressed interest in dangling keys that her eyes followed closely, she
climbed a chair that had been insurmountable on a normal diet and anyone watching
the film would feel her retreat into the doldrums of retardation once the phenylalanine
was added again to her diet. But she remained severely retarded throughout these trials;
indeed, these trials underscored a critical period, very early in development, for the
establishment and restoration of basic functions of the brain. Critical periods were also
of prime interest to Gesell and to those few investigators who followed his lead in using
the co-twin control design. Gesells lifelong scientific passion was mapping the devel-
opmental stages and landmarks of typical development. The MZ co-twin control design
enabled him to hold constant individual differences among children occasioned by
their genetic differences and differences in environmental factors common to siblings
in the same family. It was a strategy that helped him identify the developmental stages
that were common for all children he studied. It is a pity that investigators lost interest
in the use of twins by the advent of World War II.As we sketch two uses of genetics in
intervention researchidentify targets and exploring individual differences in
responsewe will suggest ways in which newer research can unravel some seemingly
intractable research problems in development.
Defining Targets forBehavioral Intervention
As noted, PKU research introduced a paradigm for the use of genetic information
for therapeutic intervention, specifically for environmental alteration of genetic
risk. However, PKU reflects the effects of a polymorphism of a single gene regulat-
ing the activity of a liver enzyme. Without major modification, this paradigm cannot
be applied to disorders influenced by many genes each of which may have pleiotro-
pic3 effects on brain mechanisms and each of which are densely interrelated with
other mechanisms and are also influenced by social and other environmental influ-
ences as well. Nonetheless, this paradigm has potential and we review here direc-
tions for research within its frame.
3
When allelic variation in a single gene affects two or more distinctly different phenotypes that are
unrelated to each other, the variation is set to be pleiotropic.
14 D. Reiss
Defining Groups ofIndividuals at Risk forDisease
The PKU paradigm suggests that if we can identify groups of individuals at risk for a
disorder before the disorder occurs we may prevent it even if we do not fully under-
stand its pathogenesis. For example, Uhl and his colleagues have extended his work
on assessing genes that favor successful cessation of smoking following a planned
intervention. Thesecombined in polygenic risk scorepostdict which adolescents
at age 14 will rapidly escalate drug use into adulthood and which adolescents will, by
and large, resist the addictive substances (Uhl, 2014). Uhl did not test whether his
polygenic risk score overlapped or exceeded in precision that of a simpler family his-
tory of smoking. However, Daniel Belsky and colleagues using a similar postdiction
strategy with a somewhat different polygenic risk score found family history and
genetic data independently contributed to the precision of postdiction (Belsky, 2013).
That is, they used a polygenic risk score to distinguish which children became smok-
ers and which did not across nearly 30years of observation. However, great caution is
necessary in identifying individuals at risk with polygenic risk scores. For example,
Gartner and colleagues using a polygenic risk in a simulation model found no advan-
tage over family history in screening for adolescent smokers (Gartner, Barendregt, &
Hall, 2009). Further, broad experience with these genetic strategies in prediction obe-
sity, cardiovascular disease, and diabetes has suggested they have little utility beyond
a family history (Evaluation of Genomic Applications in & Prevention Working,
2010; InterAct Consortium etal., 2013; Veerman, 2011).
Belsky and Uhl used polygenic risk scores to postdict patterns of behavior, poor
self-regulation, and addiction. However, this same strategy can be used to predict
sensitivity to environmental influences. For example, gene variants regulating vari-
ous CNS neurotransmitter systems postdict the responsivity of adolescents to par-
enting style: favorable parenting in these adolescents was associated with enhanced
self-regulation but those with unfavorable parenting developed serious problems in
self-regulation; those without these so-called plasticity variants showed little effect
of parenting either way (Belsky & Beaver, 2011). The role of these polygenic scores
as practical screening is untested but promising.
Suppose a single gene or a set of genes reliably predicted a health outcome what
is the prospect that a behavioral intervention might offset that risk analogous to the
salutary effects of low phenylalanine diet in PKU? In two related proof-of-concept
studies, Brody and his colleagues reported that a serotonin transporter gene polymor-
phism postdicted a higher risk in teens for a range of risk behavior (Brody, Beach,
Philibert, Chen, & Murry, 2009) and a set of additional genes regulating gabaergic
and dopaminergic brain function postdicted an increased risk for alcoholism (Brody,
Chen, & Beach, 2013). In each case a brief, family-oriented prevention curriculum
offsets these genetic risks. Additional information about these studies can be found
elsewhere in this volume. While needing replication, these findings are important
because they suggest that public health oriented interventions might, in some fash-
ion, offset risk for serious disorders that are identified through genetic assays.
Quantitative genetics provides two useful tools for identifying groups of children
at risk. The first is to identify the earliest appearing behavioral indicators of genetic
risk, in infants or toddlers. A powerful tool is the prospective adoption design where
children are placed for care at birth, their development is followed across time, and
detailed studies are made of both biological and rearing parents. Though prospective
adoption studies are very rare, properly designed they can answer an important ques-
tion: if one or both parents exhibit severe psychopathology what is the first manifes-
tation of this risk in infants and toddlers? The Early Growth and Development Study
is the only prospective adoption design to include birth parents, adopted parents, and
children placed for adoption within a short time after birth (Leve, Neiderhiser,
Scaramella, & Reiss, 2010) and to focus on social and emotional development in
children. It has found three characteristics of toddlers that are very early manifesta-
tion of the genetic risk for externalizing disorders (as indexed by the psychopathol-
ogy of their birth parents correcting for intrauterine factors): an inability to self sooth
in frustrating situations, a need for structured parenting, and a capacity to evoke
maternal negativity, especially in the context of marital difficulties (Fearon et al.,
2014; Leve etal., 2009, 2010). The practical utility of these discoveries is untested
but the findings suggest psychosocial interventions that may directly address these
early psychological and interpersonal difficulties of children who are at risk for a
broad variety of externalizing disorders including smoking and substance abuse.
A second contribution, drawn mainly from twin designs, is to identify natural envi-
ronmental variation that has a major impact on behavioral development independent
of a child genotype. These naturally occurring variations are good clues for innovative
interventions. For example, little noted in the genetic literature is redundant evidence
that sibling relationships differ in both warmth and conflict across families and these
between-family differences regularly anticipate the development of both aggression
and substance use in both siblings (Natsuaki, Ge, Reiss, & Neiderhiser, 2009;
Neiderhiser, Marceau, & Reiss, 2013; Reiss, Neiderhiser, Hetherington, & Plomin,
2000a; Slomkowski, Rende, Novak, Lloyd-Richardson, & Niaura, 2005). The value
of genetic information here is that it identifies these substantial influences as operating
entirely by environmental mechanism and invites sibling-focused preventive interven-
tions as part of an effective strategy for prevention.
Delineating Malleable Mechanisms ofGene Expression
The mechanisms by which genes are expressed in behavior are being defined on at
least three levels.
The most familiar level is the molecular one where technical progress now allows
a broad genome scans of patterns of methylation and expression of messenger RNA.
Following the path breaking work of Meaney and Szyf in rats (see a review Meaney
& Szyf, 2005 and material elsewhere in this volume), evidence is accumulating that
adverse experience in both childhood and adulthood can alter patterns of gene expres-
sion as examined in postmortem brain samples (McGowan etal., 2009) and in periph-
eral blood (Cole et al., 2007). Although stressful circumstances can induce very
distinct and contrasting gene expression profiles in circulating white cells and in brain
16 D. Reiss
(Provencal et al., 2012), peripheral blood has yielded fairly consistent patterns of
upregulated (genes regulating inflammatory responses) and downregulated (cortisol
monitoring systems and antiviral mechanisms) expression systems, all of which
heighten liability to a range of medical disorders (Cole, 2014; Miller etal., 2009).
Preliminary, proof-of-concept trials suggest that cognitive behavior (Antoni et al.,
2012) and mediation procedures (Bhasin, 2013; Creswell et al., 2012) can reverse
some of these stress-induced profiles of gene expression. However, even in prelimi-
nary studies, it remains unclear how important these reversals are to patient recovery.
A second level of study is influence of genetic allelic variation on brain function.
For example, two early publications in this field by Ahmad Hariri engendered con-
siderable excitement: he and his colleagues reported that research subjects with one
or two of the short versions of the serotonin transporter gene had stronger amyg-
dala responses to emotional stimuli on fMRI (Hariri etal., 2002, 2005). This finding
might explain why individuals with the short allele of this gene would be vulnerabil-
ity to stress-induced depression as illustrated by the oft-cited paper by Caspi etal.
(2003). However, many subsequent efforts to replicate this finding suggested that
the effect of the single gene was, at best, very small (Murphy etal., 2013). The use
of genetic variation to delineate the role of brain function in the pathogenesis of
mental disorders will almost certainly have to examine the effects of many genes
acting in concert and study variation in the function of neural networks and not
single brain regions (Birnbaum & Weinberger, 2013).
Despite uncertainties in this field of study, the use of genetics to identify brain
functions on the path to major disorders is an inviting target for behavioral therapeu-
tic and preventive interventions. The most specific behavioral interventions are
fMRI neurofeedback techniques where subjects learn to control the activity of spe-
cific regions or circuits by up- or downregulating MRI signals directly in their visual
field. Thus, fMRI neurofeedback techniques have been used to downregulate amyg-
dala response to adverse stimuli in normal subjects, thereby enhancing their emo-
tional self-regulation (Sarkheil et al., 2014) upregulating amygdala response to
pleasurable stimuli in depressed subjects suggesting the possibility of this treatment
for anhedonia (Young etal., 2014).
A third level of genetic expression occurs entirely through the medium of social
relationships. Heritable features of children evoke a broad range of responses from
parents, siblings (Klahr & Burt, 2014; Pike, McGuire, Hetherington, Reiss, & Plomin,
1996; Reiss et al., 2000a), and peers (Manke, McGuire, Reiss, Hetherington, &
Plomin, 1995; Rose, 2002), and heritable features of adults influence their level of
perceived social support and exposure to stressful events (Kendler & Baker, 2007;
Kendler & Karkowski-Shuman, 1997; Kendler etal., 1995). Findings such as these
led Reiss etal. (2000c) and Kendler (2001) to propose that these genetic effects on
environmental process constitute a major pathway of the expression of genetic influ-
ences on psychopathology. A number of longitudinal studies have supported this idea
using twin (Burt, McGue, Krueger, & Iacono, 2005; Larsson, Viding, Rijsdijk, &
Plomin, 2008), adoption (Elam etal., 2014; Harold etal., 2013), and molecular meth-
ods (Propper, Shanahan, Russo, & Mills-Koonce, 2012). Kendler estimated that 16%
of the genetic influence on depression was expressed through genetic influences on
exposure to stressful life events and on social support (Kendler, 2001); Neiderhiser
and her colleagues (2013) estimate that all of the genetic influence on initiation of
illegal substance use in young adults was expressed through genetic influence on their
families and peer selection when they were adolescents.
These outside the skin (Kendler, 2001) mechanisms of genetic expression are
conspicuous but never used targets for psychosocial interventions to blunt the expres-
sion of genetic influences. In a recent paper already cited, Fearon and his colleagues
(2014) showed that rearing parents marital satisfaction played a decisive role in the
impact of their adopted childrens heritable characteristics on their experience of par-
enting. Fearon characterized young childrens liability for externalizing disorders by
assessing birth parents psychopathology. Where marital satisfaction was low the rear-
ing mothers perception of her own parenting was negative, probably because the
child was perceived as vexatious. However, in the context of a favorable marriage,
mothers saw the child with the same genetic risk in positive terms and expressed posi-
tive parental feelings. The effect of marital dissatisfaction on the evoked maternal
feelings played a notable role in the evolution of conduct and related problems in the
developing toddlers. These findings suggested that early marital interventions might
abort the expression of genetic influence on child psychopathology by interrupting an
unheralded but important pathway of gene expression.
However, it is far from secret that early family interventions are effective in pro-
moting positive child development preventing the evolution of child psychopathol-
ogy (Cowan & Cowan, 2010). There is good evidence that focus such interventions
on the parents marriage has additional yield. Does genetic information add anything
to the mix already available? What would be possible now, to continue with the data
provided by Fearon and his collaborators, is to deliver marriage-centered preventive
intervention within the context of a genetically informed design. To what extent is
the success of a marriage-oriented intervention dependent on blocking the outside
the skin mechanisms of gene expression and to what extent is therapeutic success
attributable to social mechanisms independent of the childs genotype? For e xample,
does marriage-oriented intervention eliminate the correlation of birth parent psy-
chopathology with differences in behavior problems among children placed for
adoption? What behaviors in the child or alternations in maternal perceptions are
critical to this anti-genetic effect? What other outside the skin pathway might be
blocked by such an intervention and might some of the positive outcomes be mea-
sured by childs physical health?
Defining theTiming ofBehavioral Interventions
There is increasing interest in integrating both quantitative and molecular genetics

in the study of the life course. This has been aided by increasingly sophisticated,
genetically informed conceptual analyses of the life course (see Shanahan & Hofer,
2011) as well as by extended longitudinal studies of twins, adoptees and longitudi-
nal studies that have included both genotyping and assay of gene methylation or
18 D. Reiss
expression. So far evidence has been brought forward to tackle three major develop-
mental questions highly relevant to the timing of interventions.
The first use is to identify major developmental discontinuity, particularly periods
when earlier influences on development rapidly fade away and new ones take over.
The simplest strategy is to note, in longitudinal twin or adoption studies, the balance
over time between genetic and environmental influences on a particular line of devel-
opment. For example, the influence of environments shared by siblings, but differing
among families, is the predominant influence on general intelligence from ages two to
four but genetic differences among children play, by far, the major role from ages
seven to ten (Davis, Haworth, & Plomin, 2009). Davis and colleagues suggest a sim-
ple explanation for this finding: variability among families in intelligence-promoting
aspects of the environment may decrease owing to the uniformity of school curricula
in the UK, the site of their study. However, their data suggests to the present author
that variability in environment actually increases leaving the most likely explanation
as an absolute and very sharp increase in genetic influence from early to middle
childhood.
However, as Davis and his colleagues comment, there are two very different
mechanisms, each of which could explain this sharp increase. The first is that chil-
dren become dramatically more effective in eliciting reactions from their parents,
teachers, and friends: heritable features of intelligence such as verbal ability may
elicit responses from othersmore attention in the classroom and more intellectual
stimulation at homethat serve to amplify the childs intellectual abilities. This
gene-driven positive amplification process receives some support from molecular
genetic studies (Propper etal., 2012) and twin studies (Reiss etal., 2000c) and rep-
resents a potentially important outside the skin mechanism of gene expression. Both
these studies suggest that heritable factors in the child that elicit negative parenting
impair cognitive abilities. However, the Propper study focused on processes in early
childhood and Reiss and colleagues on adolescence so neither provides a clear
explanation for the dramatic discontinuity between ages four and seven. Davis and
colleagues advance a parallel explanation: the increase in genetic influence on corti-
cal thickening and myelination during a period approximating the period of discon-
tinuity they identified. However, even without clear mechanistic explanations these
longitudinal twin data have great relevance for behavioral interventions to enhance
childrens intellectual capacities: effective interventions, if they are designed to par-
allel or compensate for naturally occurring variation among children, are likely to
be different for early and middle childhood.
A second contribution of genetics to the timing of interventions is to estimate
when environmental influences that have been unstable and fluctuating become sta-
ble influences on development across time. These are influences most likely to
account for sustained influence on differences among individuals in their adjustment.
Quantitative genetics has become a particularly powerful tool here with the advent of
large numbers of longitudinal twins studies and a smaller number of longitudinal
adoption studies. The analysis proceeds in two steps. First, the variance accounting
for stability in individual adjustments is parsed into genetic influences and into two
great classes of environmental variance: environments that differ among families and
environments that differ among siblings in the same family. An example of the for-
mer is social class and the latter, differential treatment by parents. Then, these same
techniques can be used to specify specific environmental factors, of the between- or
within-family variety, that exert a causal influence independent of genotype. The
logic and methods of this analysis have been fully explained and exemplified (Plomin,
DeFries, Knopik, & Neiderhiser, 2013; Reiss etal., 2000a).
A good example is precise specification of the role of marital process in adult
development. Briley and Tucker-Drob (2014) and Tucker-Drob and Briley (2014)
gave recently reviewed all longitudinal twin and adoption studies to determine when
sibling-specific environments become both sizable and stable influences on indi-
vidual development for both cognitive and personality function.
Sibling-specific environments are important but unstable in early development but
by early adulthood their stability is comparable to genetic influences. Why might
these be the case?
A most likely contributor is the emerging role of marital status and marital satisfac-
tion in adult life. Indeed, whether one is married or not has major implications for
medical and behavioral health. While selection effects may have some role in this
remarkable effect, epidemiological (Laub, Nagin, & Sampson, 1998) and genetic
analyses (Burt etal., 2010) suggest that not only is marital status causal but that it is
differences between siblings in marital status that is decisive for the effect of marriage
on mental health. The same holds true for marital satisfaction that has been specifi-
cally tested for its effects on both depression and positive mental health (Spotts etal.,
2004, 2005). Genetic studies of the spouses of twins provide additional insight. On a
very broad range of characteristics spouses of identical twins are no more like each
other than spouses of fraternal twins (Lykken & Tellegen, 1993; Zietsch, Verweij,
Heath, & Martin, 2011) and the correlations for both identical and fraternal twins are
quite low for most spousal measures suggesting not only that genes play no role in
mate selection but neither do environments shared by siblings. Adults pick mates in
part as reflections of their own prior life course as it is c ontrasted with their siblings,
a life course that takes on even further uniqueness as consequence of the distinctive
relationship spouses build (and some destroy) together.
Despite strong evidence for the use of marital therapy to treat problems of indi-
vidual adjustment (for example, see Whisman etal., 2006), evidence that is unrelated
to genetic inquiry, genetic data provide novel perspectives for further development of
marital interventions. It is striking that genetic data is the most persuasive we have
that across the span of ordinary human development the quality of marital relation-
ships plays decisive effect on early child development whileat the same time
providing unique contributions to the psychological and medical health of the parents
themselves.
A third contribution to the timing of intervention, provided by genetic inquiry, is
the definition of critical or sensitive periods in development. While often clear in
animal development (e.g., Hubel & Wiesel, 1970; Liu etal., 1997; Meaney, 2001),
these periods are much harder to define in human development, particularly for com-
plex behaviors. Animal models provide decisive data because researchers can control
the on and off times of unfavorable or favorable environments. Thus Hubel and
20 D. Reiss
Wiesel could suture one eye of a kitten and remove the suture at varying subsequent
times in order to determine the critical period for the development of binocular
vision. A similar quasi experiment is available in congenitally deaf humans who have
hearing restored at various ages by cochlear implants.4 Other human designs permit
estimates of on and off time boundaries for adverse or favorable environmental
impact but genetically informed adoption studies provide the most decisive human
data. Adoption at conception designs (contrasting mothers who have their own egg
implanted versus mothers having eggs implanted from genetically unrelated donor)
distinguishes between two possibilities. First, is an adverse influence such as smok-
ing linked to a behavior outcome such as antisocial behavior because of genes shared
by mother and fetus? In this case there would be a correlation between maternal
smoking and child antisocial behavior only where the mothers own egg was
implanted. An exposure effect of smoking during pregnancy would requite the same
effect size for fetuses developing from a donated egg. Initial reports using this method
suggest that smoking exposure leads to restricted fetal growth but not to antisocial
behavior, thus pointing to genetic influences on the observed correlations (Rice etal.,
2009). Adoption at birth designs distinguishes between prenatal and postnatal influ-
ences and even their interactions. For example, an adoption at birth study suggested
that inconsistent parenting postnatally could influence a childs cortisol levels when
the fetus was exposed to a combination of maternal stress and drug use prenatally
(Marceau etal., 2013). Finally adoption at varying intervals post birth, particularly
for children subjected to harsh condition prior to adoption, allows an estimate of tim-
ing beyond which certain developmental achievements cannot be established despite
favorable rearing in the adoptive home. For example, a distinctive form of disinhib-
ited attachment (a child indiscriminately becomes attached to many different peo-
ple) persists for years, despite variation in adoptive families, if the child is raised in
an institutional setting for more than 2 years (in comparison to those adopted at
16months or less). These data suggest that a critical period for forming more focused
attachment is the latter half of the second year of life; after 2 years this focused
attachment (secure or insecure) cannot be established.
Genetic Influences onResponse toTreatment
Arnold Gesells experiments have long been forgotten and Foxs publication has
received only scattered citations. Only a single investigator has, in relation to that
paper, used the twin method to explore the role of genetic influences on motor learn-
ing and on a task pairing muscle stimulation with transcranial magnetic stimulation
4
In human development, considerable attention has been focused on hearing and language develop-
ment because of the natural experiment of cochlear implantation that provides hearing to congeni-
tally deaf children. While cochlear implants tend to be less effective over the age of 8, there is no
sharp temporal breaking point and the age at which cochlear implants start to become less effective
depends heavily on the nature of the auditory or perceptual test (Harrison, Gordon, & Mount, 2005).
and with muscle stimulation, a probe of neural plasticity (Missitzi etal., 2011, 2013).
Thus, although Gesell and Fox were vanguards, they were entirely without influence
on current efforts to use genetic techniques to explore individual differences in
response to behavioral treatments. Even more pertinent to this historical discontinu-
ity is the failure to use either the twin or adoption methods in exploring individual
differences in response to behavioral interventions although clear proposals have
been published (Plomin & Haworth, 2010; Reiss etal., 2000a). We will return to the
essential role of these approaches.
The current impetus in this field comes from two sources. First, is the growing
interest in measured gene by environment interaction on measures of adjustment
across the life span. Investigators have focused on genes that, in interaction with the
environment, have a plausible role in pathogenesis of behavioral disorders. They
have hypothesized that these same genes may also play a role in differential response
to treatment since treatment can be considered an experimentally controlled envi-
ronmental variable. We have already reviewed an example in the work of Uhl and
his colleagues for genes that are associated both with response to treatment and with
enhanced risk for adolescents to become smokers. However, in this work findings
on genetic influences on individual difference in response to treatment came first.
A particular incentive for research has been the theory of differential suscepti-
bility. Originally articulated by Kendler and Eaves (1986), this theory posits that
the behavioral effect of some alleles is to enhance children and adults susceptibility
to environmental influence, for better or for worse. The first publication illustrating
this principal, of which the writer is aware, was that of Lyman Wynne and his col-
leagues showing that children of mothers with schizophrenia and who were placed
for adoption early in life had a greater chance than control children of developing
thought disorder in a unfavorable rearing family but less of chance for disorder
when raised in a favorable family setting (Wahlberg etal., 1997). Ten years later Jay
Belsky revived this idea, apparently unaware of the work of Kendler and Wynne,
and pointed to a number of measured gene x environment interaction studies that
might support it (Belsky, Bakermans-Kranenburg, & Van Ijzendoorn, 2007). A criti-
cal test of this theory would be to demonstrate that an allele that conferred risk for
a behavioral disorder in children (or adults) subjected to an adverse environment
would also serve to enhance the effect of a positive environment such as that pro-
vided in a randomized clinical trial.
However, beyond the theory of differential susceptibility, the large force stimu-
lating work in this area is the enormous enthusiasm in the biomedical community
for the prospects of personalized medicine. In this initiative we can recognize the
same two aims we summarized for the genetics of behavioral intervention: defining
new targets for treatment and better discrimination among patients between those
who will respond to treatment and those who will not. Notable success, for example,
has been achieved in breast cancer. Three advances in distinguishing among breast
biopsies are now part of standard practice: the detection of progesterone and estro-
gen receptors that can be identified by tissue staining techniques and the detection
of human epidermal growth factor (HER2), an assay that often uses genetic tech-
niques (Giordano et al., 2014). Genetic analysis has also distinguished among
22 D. Reiss
patients those that effectively metabolize tamoxifen, a drug indicated for estrogen
receptor positive tumors, into its active metabolite. Good metabolizers can be iden-
tified before treatment by genotyping for a particular cytochrome enzyme. Poor
metabolizers are generally not good candidates for tamoxifen treatment (Higgins
& Stearns, 2011).
Identifying candidates for tamoxifen treatment is a good example of the value of
pharmacogenetics, a technique that has been applied to many psychotropic drugs.
Indeed, of over 170 drugs now required by the FDA to include genomic data on
labeling instructions, as of January 2015 there are 24 psychotropic drugs listed,
almost all of them are antipsychotic or antidepressant medications (http://www.fda.
gov/drugs/scienceresearch/researchareas/pharmacogenetics/ucm083378.htm).
However, in each instance, the required genetic data refers to the cytochrome sys-
tem. This system, sited mainly in the liver, has been known for years but genotyping
is an efficient way of ascertaining its properties and their influence on drug dosing
and drugdrug interactions. No genomic labeling data required by the FDA reflects
current advances either in neuroscience more generally or the genetics of brain
function. However, there is a burgeoning literature that does take advantage of some
of this knowledge, particularly about receptor systems in the brain (see Reynolds,
McGowan, & Dalton, 2014 for a recent review). Reliable findings across studies are
emerging. For example, in a meta-analysis of 15 studies of individual differences in
response to SSRIs, Serretti, Kato, De Ronchi, and Kinoshita (2007) reported that
the long form of the serotonin transporter gene enhanced response to treatment.5
The application of genetic techniques to analyze differences in patient response
to standardized behavioral treatments (Reiss, 2010a, 2010b) has been the topic of
considerable theoretical attention (Haworth & Davis, 2014; Plomin & Haworth,
2010; Reiss etal., 2000b). In principal, this approach can help improve the match-
ing of particular treatments to particular individuals and even explore some of the
mechanisms by which behavioral interventions might be effective; it can illumine
the field of gene x environment interaction, and can place behavioral interventions
within a broader study of developmental psychobiology.
The motivating vision of personalized medicine is to utilize insight from mod-
ern biology to better target treatments and select individuals who will respond to
those treatments. We have reviewed in some detail the relative success of behavioral
genetics in clarifying individuals at risk for behavioral disorders, mechanism of
gene expression that may be feasible targets, and the timing in development when
intervention may be most effective. Elsewhere in this volume researchers will report
on ongoing efforts to use molecular genetic techniques to account for individual
differences in response to treatment; we provide a brief introduction here.
Coining the term therapygenetics Thalia Eley with her colleague Kathryn
Lester, reviewed 14 studies of measured gene x therapy interaction which, at that
juncture in the field, yielded few impressive findings (Lester & Eley, 2013). For
5
Since Belsky and others have argued that the short form of this gene confers susceptibility to both
unfavorable and favorable environments, these pharmacogenetic effects would appear to weaken
the theory of differential susceptibility.
example, six of nine studies of the moderating effect of the short allele of the sero-
tonin transporter gene showed no effect. Some of the studies retrospectively geno-
typed already treated sample, and analyses simply omitted those not genotyped,
thus infringing on the benefits of randomization. Some studies followed only treated
patients (see a brief follow report) (Eley etal., 2014). While providing clues about
the role of genotype, DNA methylation and even SNP-based heritability6, as influ-
ences on patient change this design, cannot distinguish between treatment effects
and varying speeds of spontaneous remission.
At this early stage of therapygenetics preliminary reports are sufficiently
promising to pursue more adequately powered studies. These may give clues to
some of the biological mechanisms by which behavioral therapies reveal their
effects. As noted elsewhere (Reiss, Leve, & Neiderhiser, 2013), genetic main effects
are essential in understanding G x E interactions. What neural or cellular mecha-
nisms do the genes regulate that account for the differential effects of allelic varia-
tion on differential response of patients to treatment? For example, the association
of allelic variation in the serotonin gene with amygdale function once offered such
a hope. Could the gene-influenced enhanced reactivity, highlighted in initial reports,
account for the greater responsiveness of patients to the positive effects of treat-
ment? Tamoxifen is a good example as the main effect on the cytochrome system of
the gene that moderates its effect fully explains how this genotype moderates treat-
ment response. We are some distance however from deploying genetics to develop
biologically mechanistic explanations for the effect of behavioral interventions. A
more immediate yield may be to determine whether the effect of a broad range of
treatments is all moderated by the same allelic variation in the same gene or sets of
genes. This would aid the search for common factors among effective therapies, a
major issue in contemporary psychotherapy research. Finally genetics might explore
other central issues in the psychotherapy. For example, characteristics of parents
have substantial effects on outcome of children in treatment (e.g., Hoza etal., 2000).
In a sufficiently powered sample exploring the parents genotype as a moderator of
child treatment would underscore the importance of this effect. Indeed, some of the
apparent effect of child genotype in preliminary studies supported by Eley and oth-
ers may reflect, in part, parental genotype effects.
Genetically based studies of response to treatment can also aid the study of gene
environment interaction more broadly. Standards for statically inferring genotype x
environment interaction are clear but often honored in the breach. Investigators are
usually aware of the need to control for the effect of the child genotype on the environ-
ment being studied since such an effect can perfectly mimic a genotype x environment
interaction. Similarly, where the environment studied is shaped by a biological parent
6
This analysis proceeds in the opposite form from that of a twin design. This procedure compares
all possible pairs of individuals in a sample and asks whether their degree of phenotypic similarity
is matched by a comparable overlap in genetic structure. Because SNPs do not contain all possible
genetic information the estimates of heritability by this method are usually less than that of the twin
method. The genes involved are not identified and significant SNP heritability may be obtained in
circumstances where not even a single gene reaches a threshold for an effect on the phenotype under
analysis.
24 D. Reiss
or sibling the genes shared between these and the offspring under study, so-called pas-
sive gene-environment correlation, can also mimic gene x environment interaction but
is much less frequently accounted for (see Kim-Cohen etal., 2006 for a delineation of
the problem and its solution). A nearly intractable problem is the possibility that an
unidentified gene evokes the environment under study. Thus an unidentified gene x
identified gene interaction can, again, mimic a gene x environment interaction. All of
these problems are obviated when the researcher controls environmental variation as
in a randomized controlled trial. However, most treatments that are effective against
serious psychological problems often induce many changes in environments experi-
enced by children and by adults. Thus, simpler experiments that, for example, allow
researcher to control stress levels (e.g., Way & Taylor, 2010, 2011) may be part of
such an effort. As noted, genetic investigation of responses to treatment also provides
a stringent test of the differential susceptibility hypothesis.
Finally, the promise of genetic studies of differences in response to treatment has
the long-term promise of linking intervention studies to a more general science of
developmental psychobiology. Recall our example from the work of George Uhl on
a set of genes associated with response to smoking cessation trials; these same
genes distinguish among adolescents who go on to smoke and those who dont
clarify that biological processes that, in ordinary development, contribute to desis-
tance from smoking also favor response to treatment.
A Few Ideas About theFuture
It seems very unlikely that the current chapter will have an appreciable influence on
the course of research at the intersection of behavioral interventions and genetic
variation among individuals. The forces propelling some lines of research in this
area are very powerful and those slowing research are equally so. Thus a more mod-
est hope for a concluding statement here is that it might serve to round out a chapter
meant as one of many guides to both readers and student researchers entering this
field and perhaps to give emphasis to points made earlier but now rephrased in terms
of both encouragement and caution.
As the present writer has outlined elsewhere (Reiss, 2016), the history of behav-
ioral genetics, the quantitative inferences drawn from twin and adoption studies, is
the polar opposite from molecular genetics. The former is heading for its 100th birth-
day (although its birthdate is subject to some dispute); its methods were slow to
develop, its assumptions have been clarified and painfully examined across decades,
and its results tend to be highly replicable. However, as this chapter has illustrated,
its potential for identifying targets for interventionalong with the timing of when
interventions might be most effectiveis substantial. In contrast, the exact birthday
of the widespread use of molecular genetic analyses can probably be dated to the
hour if not the minute: Kary Mulliss presentation of the PCR method for amplifying
specific DNA sequences in 1983. At that moment, biology was forever changed; the
method (with a few critical tweaks from less acknowledged colleagues) was applied
worldwide and in less than a decade Mullis had his Nobel Prize, one of the most bit-
terly contentious ever awarded. Aided by large investments by private companies,
current methodsdirect outgrowths of the work of Mullis and colleaguesare
becoming less and less expensive. Molecular genetics has given the behavioral sci-
ences more generally, and behavioral interventions more particularly, a gift with
more apparently alluring properties than any method in the entire history of research
on human behavior. The measured polymorphism seems like a godsend in a field
with complex causes, endless problems of distinguishing cause and effect, multiple
methods purporting to measure the same construct, cultural effects on measurements
and results, and problems with the validity of retrospective assessment. First, current
dogma regards a persons genotype as a definitive first cause: preceding but not
caused by environmental exposure including participating in a clinical trial of behav-
ioral intervention.7 Second, unlike any other variable in behavioral science, research-
ers can revisit old cohorts of research subjects, obtain their genotypes, and reason
that their results are identical to those they would have obtained at the outset of their
study. Third, with suitable safeguards, a polymorphism is a polymorphism whether it
is measured in Mongolia or Manhattan. Finally, identifying a specific polymorphism
offers the promise of integrating behavioral intervention with the biology of the
brain, the immune system, and biological stress response systems.
Given the understandable allure of genotyping subjects enrolled in behavioral
interventions, this approach hardly needs encouragement. Rather, it needs some
thoughtful caution and more rigorous design and replication. The bulk of studies, in
this genre, reported thus far are those where genotyping is completed after trials have
been conducted and completed. Invariably, some subjects cannot be located or do not
give consent. Results are reported anyway even though it is impossible to maintain
the rigors of an intention-to-treat design. While results are intriguing, as noted, con-
spicuous failures to replicate (e.g., Lester etal., 2015) have not been weighted in
recent, highly selected reviews (e.g., Bakermans-Kranenburg & Van IJzendoorn,
2015). Moreover, many published studies failwith some exceptions (e.g., Schlomer
etal., 2015)to attend to the basic toilette of molecular genetic studies, most con-
spicuously the confounding effects of population stratification by allele frequency.
Finally, readers and journal reviewers of these studies have no way of knowing how
many polymorphisms were assayed by investigators, in any given study, and whether
those that are published were selectively harvested cherries. The stakes are high and
ethical issues preemptory: misused genetic studies might become a basis to dissuade
patients from receiving therapies that seem to be contraindicated by their genotype.
With these problems in mind, it is entirely reasonable to propose that post hoc geno-
typing of subjects in completed trials is reaching the end of a useful exploratory phase.
7
While it remains the case that the genotype is fixed at conception it can no longer be regarded as
nave to earlier experience. A range of studies, mostly in animals, suggests that parental and even
grandparental experience leaves an inheritable residue of altered gene expression. See, for exam-
ple, a particularly vivid example of the transmission of fear conditioning to a very specific odor in
an F0 generation of mice that is transmitted to the F1 and F2 generations through hypomethylation
of a highly specific genetic locus in the DNA of sperm cells (Dias & Ressler, 2013).
26 D. Reiss
Future studies of genetic moderation of behavioral therapies need to be prospective and

adequately powered. Genotyping must be done in response to the best available evi-
dence of the role of the genetic variants selected for study in biological processes highly
likely to influence response to treatment. Moreover, the genetic variants selected for
study must be entered before the trial, along with other aspects of the design, in the
lock box of clinicaltrials.gov or similarly accessible public databases.
What then of the more senior but more ignored older sibling of molecular and
genetics studies, the quantitative genetic inferences drawn from twin and adoption
studies? Its imaginative originatorsthe graduate student Curtis Merriman, the der-
matologist Herman Siemens, and the psychologist Barbara Stoddard Burks8were
not ever nor will they be household heroes, even in the homes of geneticists. However,
as noted above, the potential contributions of these methods to targeting interventions
and to their timing are very promising. Detailed speculation about why these remark-
able resources remain unused is beyond the scope of this brief chapter. Surely, one
remedial approach is enhanced accessibility. A mournful fact is that inferences and
statistical models employed in the evaluation of twin data, and to a lesser extent for
data from adoption studies, are difficult to grasp. The work of the present author and
his colleagues are as good an example as any of the use of complex inferential models
that defy ready understanding, even by mathematically competent readers (see, for
example, Narusyte etal., 2011). The utility of twin and adoption methods as tools for
strong inference about environmental effects, including those induced by therapeutic
or preventive interventions, is rarely taught in research training programs. Further,
authors in this field need to improve their presentation and explication of inferences
and statistical models in their reports of results. Senior researchers, using these meth-
ods, have argued for the utility of conducting behavior trials with twin (Plomin &
Haworth, 2010) and adoption (Reiss etal., 2000c) designs but scattered attempts to
develop such programs have been of little interest to grant review committees.
Summary
Genetic inquiry has stimulated a strong interest in linking studies of behavioral

intervention to an understanding of human psychological development across the
life span. This synthesis now mirrors broader developments in biomedicine. Can
genetics aid in identifying more accurately targets for intervention and better predict
who will respond to those interventions?
Thus far, quantitative geneticsusing twin and adoption methodshas made the
most solid contributions in part because these methods can specify both genetic and
environmental factors within these two broad domains. Molecular genetics has
achieved some early success in identifying young people at risk for developing seri-
ous disorders later in life. It remains unclear whether that success improves on an
accurate history of behavioral difficulties in first- and second-degree relatives. Rapid
See Reiss (2016) for a brief historical summaries of genetics and behavior.
8
developments in gene expression assays currently attract monumental interest and

resources but enthusiasm for these techniques must be tempered by recognizing that
gene expression is tissue specific. The use of genotyping and gene-expression profil-
ing to distinguish responders from nonresponders has the same promise and obsta-
cles as these techniques face in broader biomedicine. We need a better understanding
of both psychopathology and of mechanisms of action of interventions to make most
effective use of these techniques. Or to put the matter another way these techniques
provide useful data only insofar as they aid in that understanding. Research on
tamoxifen, and similar drugs, where both mechanism of disease and drug action and
metabolism are better understood, is a limited but important standard for behavioral
research. However, unlike tamoxifen, the influence there is overwhelming evidence
that the efficacy of behavioral interventions is shaped in good measure by social fac-
tors such as patient-therapist relationships, the marital quality of adults in treatment
and parental attitudes of children in treatment. Data suggest this is likely to be true
for pharmacotherapy as well. Progress in this field is unimaginable without equal
attention to genetic and social determinants of treatment efficacy and how these two
interact. Here twin and adoption methods become essential.
The role of twin and adoption studies in understanding the difference between
responders and nonresponders to standardized behavioral interventions is tragically
underutilized and this under use serious retards progress in the field. These are pow-
erful techniques for clarifying both environmental and genetic contributions to dif-
ferential response to interventions and to identifying what these factors might be.
Because of the immense cost of sample accrual, they are impractical for individuals
already suffering major disorders. For example, to study environmental and genetic
effects on the treatment of childhood anxiety we would require a sample of twins all
of whom suffered child anxiety disorders. However, these techniques could be
deployed brilliantly in the study of preventive intervention in high-risk twin samples.
Such samples have been accrued and yield enormous insight into the interplay of
genetic factors and the environment. By their nature, prospective adoption studies
contain high-risk children because, in the USA at least, birth parents who place their
children for adoption at birth have high prevalence of behavioral and substance abuse
disorders. Indeed, the integration of quantitative and molecular genetic investigation
is nowhere more important than in linking behavioral intervention to an understand-
ing of human development.
References
Antoni, M.H., Lutgendorf, S.K., Blomberg, B., Carver, C.S., Lechner, S., Diaz, A., Cole, S.W.
(2012). Cognitive-behavioral stress management reverses anxiety-related leukocyte transcrip-
tional dynamics. Biological Psychiatry, 71(4), 366372.
Bakermans-Kranenburg, M.J., & Van IJzendoorn, M.H. (2015). The hidden efficacy of interven-
tions: Geneenvironment experiments from a differential susceptibility perspective. Annual
Review of Psychology, 66, 381409.
28 D. Reiss
Belsky, D.W. (2013). Polygenic risk and the developmental progression to heavy, persistent smok-
ing and nicotine dependence evidence from a 4-decade longitudinal study. JAMA Psychiatry
(Chicago, Ill.), 70(5), 534.
Belsky, J., Bakermans-Kranenburg, M.J., & Van Ijzendoorn, M.H. (2007). For better and for worse:
Differential susceptibility to environmental influences. Current Directions in Psychological
Science, 16(6), 300304 Retrieved from http://www.scopus.com/inward/record.url?eid=2-s2.0-
36348983842&partnerID=40&md5=0c74b54b36059d9afe41bc86f50562e0.
Belsky, J., & Beaver, K.M. (2011). Cumulative-genetic plasticity, parenting and adolescent self-
regulation. Journal of Child Psychology and Psychiatry, 52(5), 619626.
Bhasin, M.K. (2013). Relaxation response induces temporal transcriptome changes in energy metab-
olism, insulin secretion and inflammatory pathways. PLoS One, 8(5), e62817. doi:10.1371/jour-
nal.pone.0062817.
Bickel, H. (1953). Influence of phenylalanine intake on phenylketonuria. The Lancet (British edi-
tion), 265(6790), 812813.
Birnbaum, R., & Weinberger, D.R. (2013). Functional neuroimaging and schizophrenia: A view
towards effective connectivity modeling and polygenic risk. Dialogues in Clinical Neuroscience,
15(3), 279289.
Bouchard Jr., T.J., Lykken, D.T., McGue, M., Segal, N.L., & Tellegen, A. (1990). Sources of human
psychological differences: The Minnesota Study of twins reared apart. Science, 250(4978), 223228.
Briley, D.A., & Tucker-Drob, E.M. (2014). Genetic and environmental continuity in personality
development: A meta-analysis. Psychological Bulletin, 140(5), 13031331.
Brody, G.H., Beach, S.R., Philibert, R.A., Chen, Y.F., & Murry, V.M. (2009). Prevention effects
moderate the association of 5-HTTLPR and youth risk behavior initiation: Geneenvironment
hypotheses tested via a randomized prevention design. Child Development, 80(3), 645661.
Brody, G. H., Chen, Y. F., & Beach, S. R. (2013). Differential susceptibility to prevention:
GABAergic, dopaminergic, and multilocus effects. Journal of Child Psychology and Psychiatry,
54(8), 863871.
Burgdorf, W.H., Bickers, D.R., & Hoenig, L.J. (2014). Strange Bedfellows. JAMA Dermatology,
150(4), 365365.
Burt, S., McGue, M., Krueger, R.F., & Iacono, W.G. (2005). How are parent-child conflict and
childhood externalizing symptoms related over time? Results from a genetically informative
cross-lagged study. Development and Psychopathology, 17(1), 145165.
Burt, S.A., Donnellan, M.B., Humbad, M.N., Hicks, B.M., McGue, M., & Iacono, W.G. (2010).
Does marriage inhibit antisocial behavior?: An examination of selection vs causation via a
longitudinal twin design. Archives of General Psychiatry, 67(12), 13091315.
Caspi, A., Sugden, K., Moffitt, T.E., Taylor, A., Craig, I.W., Harrington, H., Poulton, R. (2003).
Influence of life stress on depression: Moderation by a polymorphism in the 5-HTT gene.
Science, 301(5631), 386389.
Centerwall, S.A., & Centerwall, W.R. (2000). The discovery of phenylketonuria: The story of a
young couple, two retarded children, and a scientist. Pediatrics, 105(1 Pt 1), 89103.
Cole, S.W. (2014). Human social genomics. PLoS Genetics, 10(8), e1004601.
Cole, S.W., Hawkley, L.C., Arevalo, J.M., Sung, C., Rose, R.M., & Cacioppo, J.T. (2007). Social
regulation of gene expression in human leukocytes. Genome Biology, 8(9), R189.181R189.113.
Correns, C. (1950). G. Mendels law concerning the behavior of progeny of varietal hybrids.
Genetics, 35(5), 3341.
National Research Council (2013). Subjective well-being. Washington, DC: The National
Academies Press.
Cowan, P.A., & Cowan, C.P. (2010). How working with couples fosters childrens development:
From prevention science to public policy. In Strengthening family relationships for optimal
child development: Lessons from research and intervention. Washington, DC: American
Psychological Association.
Creswell, J., Irwin, M.R., Burklund, L.J., Lieberman, M.D., Arevalo, J.M., Ma, J., Cole, S.W.
(2012). Mindfulness-based stress reduction training reduces loneliness and pro-inflammatory
gene expression in older adults: A small randomized controlled trial. Brain, Behavior, and
Immunity, 26(7), 10951101.
Darwin, C. (1920). The variation of animals and plants under domestication. NewYork: D.Appelton.
Davis, O.S., Haworth, C.M., & Plomin, R. (2009). Dramatic increase in heritability of cognitive
development from early to middle childhood: An 8-year longitudinal study of 8,700 pairs of
twins. Psychological Science, 20(10), 13011308.
Dias, B.G., & Ressler, K.J. (2013). Parental olfactory experience influences behavior and neu-
ral structure in subsequent generations. Nature Neuroscience, 17, 8996. doi:10.1038/
nn.3594.
Elam, K.K., Harold, G.T., Neiderhiser, J.M., Reiss, D., Shaw, D.S., Natsuaki, M.N., Leve,
L.D. (2014). Adoptive parent hostility and childrens peer behavior problems: Examining the
role of genetically informed child attributes on adoptive parent behavior. Developmental
Psychology, 50(5), 15431552.
Eley, T.C., Lester, K.J., Roberts, S., Coleman, R.I., Keers, R., Breen, G., Hudson, J.L. (2014).
S.25.02 Genetic predictors of psychotherapy for mood and anxiety disorders. European
Neuropsychopharmacology, 24, S141S142.
Evaluation of Genomic Applications in, P., & Prevention Working, G (2010). Recommendations from
the EGAPP Working Group: Genomic profiling to assess cardiovascular risk to improve cardiovas-
cular health. Genetics in Medicine, 12(12), 839843.
Fearon, R.M., Reiss, D., Leve, L.D., Shaw, D.S., Scaramella, L.V., Ganiban, J.M., & Neiderhiser,
J. M. (2014). Child-evoked maternal negativity from 9 to 27 months: Evidence of gene-
environment correlation and its moderation by marital distress. Development and
Psychopathology, 24, 115.
Fox, P.W., Hershberger, S.L., & Bouchard Jr., T.J. (1996). Genetic and environmental contribu-
tions to the acquisition of a motor skill. Nature, 384(6607), 356358.
Gartner, C.E., Barendregt, J.J., & Hall, W.D. (2009). Multiple genetic tests for susceptibility to
smoking do not outperform simple family history. Addiction, 104(1), 118126.
Gesell, A. (1942). The method of co-twin control. Science, 95(2470), 446448. doi:10.1126/
science.95.2470.446.
Giordano, S.H., Temin, S., Kirshner, J.J., Chandarlapaty, S., Crews, J.R., Davidson, N.E.,
American Society of Clinical, O (2014). Systemic therapy for patients with advanced human
epidermal growth factor receptor 2-positive breast cancer: American Society of Clinical
Oncology clinical practice guideline. Journal of Clinical Oncology, 32(19), 20782099.
doi:10.1200/JCO.2013.54.0948.
Guthrie, R., & Susi, A. (1963). A simple phenylalanine method for detecting phenylketonuria in
large populations of newborn infants. Pediatrics, 32(3), 338343.
Hariri, A. R., Drabant, E. M., Munoz, K. E., Kolachana, B. S., Mattay, V. S., Egan, M. F., &
Weinberger, D.R. (2005). A susceptibility gene for affective disorders and the response of the
human amygdala. Archives of General Psychiatry, 62(2), 146152.
Hariri, A.R., Mattay, V.S., Tessitore, A., Kolachana, B., Fera, F., Goldman, D., Weinberger,
D.R. (2002). Serotonin transporter genetic variation and the response of the human amygdala.
Science, 297(5580), 400403.
Harold, G.T., Leve, L.D., Barrett, D., Elam, K., Neiderhiser, J.M., Natsuaki, M.N., Thapar,
A. (2013). Biological and rearing mother influences on child ADHD symptoms: revisiting the
developmental interface between nature and nurture. Journal of Child Psychology and
Psychiatry, 54(10), 10381046. doi:10.1111/jcpp.12100.
Harrison, R. V., Gordon, K. A., & Mount, R. J. (2005). Is there a critical period for cochlear
implantation in congenitally deaf children? Analyses of hearing and speech perception perfor-
mance after implantation. Developmental Psychobiology, 46(3), 252261.
Haworth, C. M., & Davis, O. S. (2014). From observational to dynamic genetics. Frontiers in
Genetics, 5, 6.
Higgins, M.J., & Stearns, V. (2011). Pharmacogenetics of endocrine therapy for breast cancer.
Annual Review of Medicine, 62, 281293. doi:10.1146/annurev-med-070909-182545.
30 D. Reiss
Hoza, B., Owens, J. S., Pelham Jr., W. E., Swanson, J. M., Conners, C. K., Hinshaw, S. P.,
Kraemer, H.C. (2000). Parent cognitions as predictors of child treatment response in attention-
deficit/hyperactivity disorder. Journal of Abnormal Child Psychology, 28(6), 569583.
Hubel, D.H., & Wiesel, T.N. (1970). The period of susceptibility to the physiological effects of
unilateral eye closure in kittens. The Journal of Physiology, 206(2), 419436.
InterAct Consortium, Scott, R.A., Langenberg, C., Sharp, S.J., Franks, P.W., Rolandsson, O.,
Wareham, N. J. (2013). The link between family history and risk of type 2 diabetes is not
explained by anthropometric, lifestyle or genetic risk factors: The EPIC-InterAct study.
Diabetologia, 56(1), 6069. doi:10.1007/s00125-012-2715-x.
Kendler, K. S. (2001). Twin studies of psychiatric illness: an update. Archives of General
Psychiatry, 58(11), 10051014.
Kendler, K.S., & Baker, J.H. (2007). Genetic influences on measures of the environment: A sys-
tematic review. Psychological Medicine, 37(5), 615626 Retrieved from http://www.scopus.
com/scopus/inward/record.url?eid=2-s2.0-34047161579&partnerID=40&rel=R8.0.0.
Kendler, K.S., & Eaves, L.J. (1986). Models for the joint effect of geneotype and environment on
liability to psychiatric illness. American Journal of Psychiatry, 143(3), 279289.
Kendler, K.S., & Karkowski-Shuman, L. (1997). Stressful life events and genetic liability to major
depression: Genetic control of exposure to the environment? Psychological Medicine, 27(3),
539547.
Kendler, K.S., Kessler, R.C., Walters, E.E., MacLean, C., Neale, M.C., Heath, A.C., & Eaves,
L.J. (1995). Stressful life events, genetic liability, and onset of an episode of major depression
in women. American Journal of Psychiatry, 152(6), 833842.
Kim-Cohen, J., Caspi, A., Taylor, A., Williams, B., Newcombe, R., Craig, I., & Moffitt, T. (2006).
MAOA, maltreatment, and gene-environment interaction predicting childrens mental health:
New evidence and a meta-analysis. Molecular Psychiatry, 11(10), 903913.
Klahr, A.M., & Burt, S.A. (2014). Elucidating the etiology of individual differences in parenting:
A meta-analysis of behavioral genetic research. Psychological Bulletin, 140(2), 544586.
doi:10.1037/a0034205.
Larsson, H., Viding, E., Rijsdijk, F.V., & Plomin, R. (2008). Relationships between parental nega-
tivity and childhood antisocial behavior over time: A bidirectional effects model in a longitudi-
nal genetically informative design. Journal of Abnormal Child Psychology, 36(5), 633645.
Laub, J.H., Nagin, D.S., & Sampson, R.J. (1998). Trajectories of change in criminal offending:
Good marriages and the desistance process. American Sociological Review, 63(2), 225238.
Lester, K.J., & Eley, T.C. (2013). Therapygenetics: Using genetic markers to predict response to
psychological treatment for mood and anxiety disorders. Biology of Mood & Anxiety Disorders,
3(1), 4. doi:10.1186/2045-5380-3-4.
Lester, K.J., Roberts, S., Keers, R., Coleman, J.R., Breen, G., Wong, C.C., Eley, T.C. (2015).
Non-replication of the association between 5HTTLPR and response to psychological therapy
for child anxiety disorders. The British Journal of Psychiatry, 208(2), 182188. doi:10.1192/
bjp.bp.114.154997.
Leve, L.D., Harold, G.T., Ge, X., Neiderhiser, J.M., Shaw, D., Scaramella, L.V., & Reiss, D.
(2009). Structured parenting of toddlers at high versus low genetic risk: Two pathways to child
problems. Journal of the American Academy of Child and Adolescent Psychiatry, 48(11),
11021109.
Leve, L.D., Kerr, D.C., Shaw, D., Ge, X., Neiderhiser, J.M., Scaramella, L.V., . . . Reiss, D. (2010).
Infant pathways to externalizing behavior: Evidence of GenotypeEnvironment interaction.
Child Development, 81(1), 340356. doi:10.1111/j.1467-8624.2009.01398.x; CDEV1398 [pii]
Leve, L.D., Neiderhiser, J.M., Scaramella, L.V., & Reiss, D. (2010). The early growth and devel-
opment study: Using the prospective adoption design to examine genotype-environment inter-
play. Behavior Genetics, 40(3), 306314. doi:10.1007/s10519-010-9353-1
Liu, D., Diorio, J., Tannenbaum, B., Caldji, C., Francis, D., Freedman, A., Meaney, M. J.
(1997). Maternal care, hippocampal glucocorticoid receptors, and hypothalamic-pituitary-
adrenal responses to stress. Science, 277(5332), 16591662.
Lykken, D.T., & Tellegen, A. (1993). Is human mating adventitious or the result of lawful choice?
A twin study of mate selection. Journal of Personality and Social Psychology, 65(1), 5668.
Manke, B., McGuire, S., Reiss, D., Hetherington, E., & Plomin, R. (1995). Genetic contributions
to adolescents extrafamilial social interactions: Teachers, best friends, and peers. Social
Development, 4(3), 238256.
Marceau, K., Ram, N., Neiderhiser, J.M., Laurent, H.K., Shaw, D.S., Fisher, P., Leve, L.D.
(2013). Disentangling the effects of genetic, prenatal, and parenting influences on childrens
cortisol variability. Stress, 16, 607615. doi:10.3109/10253890.2013.825766.
McGowan, P.O., Sasaki, A., DAlessio, A.C., Dymov, S., Labonte, B., Szyf, M., Meaney, M.J.
(2009). Epigenetic regulation of the glucocorticoid receptor in human brain associates with
childhood abuse. Nature Neuroscience, 12(3), 342348. doi:10.1038/nn.2270 .nn.2270 [pii]
ences in stress reactivity across generations. [Review] [203 refs]. Annual Review of
Neuroscience, 24, 11611192.
Meaney, M.J., & Szyf, M. (2005). Maternal care as a model for experience-dependent chromatin
plasticity? Trends in Neurosciences, 28(9), 456463.
Mendel, G. (1948). Experiments in plant hybridization. Cambride, MA: Harvard University Press.
Miller, G.E., Chen, E., Fok, A.K., Walker, H., Lim, A., Nicholls, E.F., Kobor, M.S. (2009).
Low early-life social class leaves a biological residue manifested by decreased glucocorticoid
and increased proinflammatory signaling. Proceedings of the National Academy of Sciences of
the United States of America, 106(34), 1471614721. doi:10.1073/pnas.0902971106.
Missitzi, J., Gentner, R., Geladas, N., Politis, P., Karandreas, N., Classen, J., & Klissouras, V.
(2011). Plasticity in human motor cortex is in part genetically determined. The Journal of
Physiology, 589(Pt 2), 297306. doi:10.1113/jphysiol.2010.200600.
Missitzi, J., Gentner, R., Misitzi, A., Geladas, N., Politis, P., Klissouras, V., & Classen, J.(2013).
Heritability of motor control and motor learning. Physiological Reports, 1(7).
Murphy, S.E., Norbury, R., Godlewska, B.R., Cowen, P.J., Mannie, Z.M., Harmer, C.J., & Munafo,
M.R. (2013). The effect of the serotonin transporter polymorphism (5-HTTLPR) on amygdala
function: A meta-analysis. Molecular Psychiatry, 18(4), 512520. doi:10.1038/mp.2012.19.
Narusyte, J., Neiderhiser, J. M., Andershed, A. K., DOnofrio, B. M., Reiss, D., Spotts, E.,
Lichtenstein, P. (2011). Parental criticism and externalizing behavior problems in adolescents:
The role of environment and genotype-environment correlation. Journal of Abnormal
Psychology, 120(2), 365376. doi:10.1037/a0021815 .2011-01600-001 [pii]
Natsuaki, M. N., Ge, X., Reiss, D., & Neiderhiser, J. M. (2009). Aggressive behavior between
siblings and the development of externalizing problems: Evidence from a genetically sensitive
study. Developmental Psychology, 45(4), 10091018.
Neiderhiser, J.M., Marceau, K., & Reiss, D. (2013). Four factors for the initiation of substance use
by young adulthood: A 10-year follow-up twin and sibling study of marital conflict, monitor-
ing, siblings, and peers. Development and Psychopathology, 25(1), 133149. doi:10.1017/
s0954579412000958.
Pike, A., McGuire, S., Hetherington, E., Reiss, D., & Plomin, R. (1996). Family environment and
adolescent depressive symptoms and antisocial behavior: A multivariate genetic analysis.
Developmental Psychology, 32(4), 590603.
Plomin, R., DeFries, J., Knopik, V.S., & Neiderhiser, J.(2013). Behavioral genetics (6th ed., ).
NewYork: Worth.
Plomin, R., & Haworth, C. M. (2010). Genetics and intervention research. Perspectives on
Psychological Science, 5(5), 557563.
Proctor, R. (1988). Racial hygeine. In Medicine under the Nazis. Cambridge, MA: Harvard
University Press.
Propper, C.B., Shanahan, M.J., Russo, R., & Mills-Koonce, W.R. (2012). Evocative gene-parenting
correlations and academic performance at first grade: An exploratory study. Development and
Psychopathology, 24(4), 12651282. doi:10.1017/S0954579412000697 .S0954579412000697 [pii]
Provencal, N., Suderman, M.J., Guillemin, C., Massart, R., Ruggiero, A., Wang, D., Szyf, M.
(2012). The signature of maternal rearing in the methylome in rhesus macaque prefrontal cor-
32 D. Reiss
tex and T cells. The Journal of Neuroscience, 32(44), 1562615642. doi:10.1523/

JNEUROSCI.1470-12.2012 .32/44/15626 [pii]
Reiss, D. (2010a). Genetic thinking in the study of social relationships. Perspectives on
Psychological Science, 5(5), 502515. doi:10.1177/1745691610383516.
Reiss, D. (2010b). Introduction to the special issue. Perspectives on Psychological Science, 5(5),
499501. doi:10.1177/1745691610383514.
Reiss, D. (2016). Genetics and family systems: Articulation and disarticulation. In D.Cicchetti
(Ed.), Developmental psychopathology. NewYork: Wiley.
Reiss, D., Leve, L.D., & Neiderhiser, J.M. (2013). How genes and the social environment moder-
ate each other. American Journal of Public Health, 103(Suppl 1), S111S121. doi:10.2105/
ajph.2013.301408.
Reiss, D., Neiderhiser, J.M., Hetherington, E.M., & Plomin, R. (2000a). The relationship code:
Deciphering genetic and social patterns in adolescent development. Cambridge, MA: Harvard
University Press.
Reiss, D., Neiderhiser, J.M., Hetherington, E.M., & Plomin, R. (2000b). The relationship code:
Deciphering genetic and social influences on adolescent development. Cambridge, MA:
Harvard University Press.
Reiss, D., Neiderhiser, J.M., Hetherington, E.M., & Plomin, R. (2000c). The relationship code:
Deciphering genetic and social influences on adolescent development. [References].
Rende, R.D., Plomin, R., & Vandenberg, S.G. (1990). Who discovered the twin method? Behavior
Genetics, 20(2), 277285.
Reynolds, G.P., McGowan, O.O., & Dalton, C.F. (2014). Pharmacogenomics in psychiatry: The
relevance of receptor and transporter polymorphisms. British Journal of Clinical Pharmacology,
77(4), 654672. doi:10.1111/bcp.12312.
Rice, F., Harold, G.T., Boivin, J., Hay, D.F., van den Bree, M., & Thapar, A. (2009). Disentangling
prenatal and inherited influences in humans with an experimental design. PNAS Proceedings of
the National Academy of Sciences of the United States of America, 106(7), 24642467.
Rietveld, C.A., Cesarini, D., Benjamin, D.J., Koellinger, P.D., De Neve, J.-E., Tiemeier, H.,
Krueger, R. F. (2013). Molecular genetics and subjective well-being. Proceedings of the
National Academy of Sciences, 110(24), 96929697.
Rose, R.J. (2002). How do adolescents select their friends? A behavior-genetic perspective. In
L.Pulkkinen, & A.Caspi (Eds.), Paths to successful development: Personality in the life course
(pp.106125). NewYork, NY: Cambridge University Press.
Sarkheil, P., Zilverstand, A., Kilian-Hutten, N., Schneider, F., Goebel, R., & Mathiak, K. (2014).
fMRI feedback enhances emotion regulation as evidenced by a reduced amygdala response.
Behavioural Brain Research, 281c, 326332. doi:10.1016/j.bbr.2014.11.027.
Schlomer, G. L., Cleveland, H. H., Vandenbergh, D. J., Feinberg, M. E., Neiderhiser, J. M.,
Greenberg, M. T., Redmond, C. (2015). Developmental differences in early adolescent
aggression: A geneenvironment x intervention analysis. Journal of Youth and Adolescence,
44(3), 581597. doi:10.1007/s10964-014-0198-4.
Serretti, A., Kato, M., De Ronchi, D., & Kinoshita, T. (2007). Meta-analysis of serotonin trans-
porter gene promoter polymorphism (5-HTTLPR) association with selective serotonin
reuptake inhibitor efficacy in depressed patients. Molecular Psychiatry, 12(3), 247257.
doi:10.1038/sj.mp.4001926.
Shanahan, M. J., & Hofer, S. M. (2011). Molecular genetics, aging, and well-being: Sensitive
period, accumulation, and pathway models. In Handbook of aging and the social sciences
(pp.135147).
Siemens, H.W. (1927). The diagnosis of identity in twins. The Journal of Heredity, 18(5), 201209.
Slomkowski, C., Rende, R., Novak, S., Lloyd-Richardson, E., & Niaura, R. (2005). Sibling effects
on smoking in adolescence: Evidence for social influence from a genetically informative
design. Addiction, 100(4), 430438.
Spotts, E.L., Neiderhiser, J.M., Ganiban, J., Reiss, D., Lichtenstein, P., Hansson, K., Pedersen,
N.L. (2004). Accounting for depressive symptoms in women: A twin study of associations
with interpersonal relationships. Journal of Affective Disorders, 82(1), 101111.
Spotts, E. L., Pederson, N. L., Neiderhiser, J. M., Reiss, D., Lichtenstein, P., Hansson, K., &
Cederblad, M. (2005). Genetic effects on womens positive mental health: Do marital relation-
ships and social support matter? Journal of Family Psychology, 19(3), 339349.
Sydnes, S., & Flling, A. (1962). On detection of heterozygotes for phenylpyruvic oligophre-
nia. Scandinavian Journal of Clinical & Laboratory Investigation, 14(1), 4446.
Tucker-Drob, E.M., & Briley, D.A. (2014). Continuity of genetic and environmental influences
on cognition across the life span: A meta-analysis of longitudinal twin and adoption studies.
Psychological Bulletin, 140(4), 949979.
Uhl, G.R. (2014). Smoking quit success genotype score predicts quit success and distinct patterns
of developmental involvement with common addictive substances. Molecular Psychiatry,
19(1), 5054. doi:10.1038/mp.2012.155.
Uhl, G.R., Liu, Q.-R., Drgon, T., Johnson, C., Walther, D., Rose, J.E., Lerman, C. (2008).
Molecular genetics of successful smoking cessation: Convergent genome-wide association
study results. Archives of General Psychiatry, 65(6), 683693.
Veerman, J.L. (2011). On the futility of screening for genes that make you fat. PLoS Medicine,
8(11), e1001114.
Wahlberg, K. E., Wynne, L. C., Oja, H., Keskitalo, P., Pyklinen, L., Lahti, I., Tienari, P.
(1997). Gene-environment interaction in vulnerability to schizophrenia: Findings from the
Finnish Family Study of Schizophrenia. American Journal of Psychiatry, 154(3), 355362.
Way, B.M., & Taylor, S.E. (2010). The serotonin transporter promoter polymorphism is associ-
ated with cortisol response to psychosocial stress. Biological Psychiatry, 67(5), 487492.
Way, B.M., & Taylor, S.E. (2011). A polymorphism in the serotonin transporter gene moderates
cardiovascular reactivity to psychosocial stress. Psychosomatic Medicine, 73(4), 310317.
PSY.0b013e31821195ed [pii]
Whisman, M.A., Beach, S., Wamboldt, M., Kaslow, N., Heyman, R., First, M., Reiss, D. (2006).
Role of couples relationships in understanding and treating mental disorders. In Relational pro-
cesses and DSM-V: Neuroscience, assessment, prevention, and treatment (pp. 225238).
Arlington, VA: American Psychiatric Association.
Xian, H. (2003). The heritability of failed smoking cessation and nicotine withdrawal in twins
who smoked and attempted to quit. Nicotine & Tobacco Research, 5(2), 245254.
Young, K.D., Zotev, V., Phillips, R., Misaki, M., Yuan, H., Drevets, W.C., & Bodurka, J.(2014).
Real-time FMRI neurofeedback training of amygdala activity in patients with major depressive
disorder. PLoS One, 9(2), e88785.
Zietsch, B. P., Verweij, K. J., Heath, A. C., & Martin, N. G. (2011). Variation in human mate
choice: simultaneously investigating heritability, parental influence, sexual imprinting, and
assortative mating. The American Naturalist, 177(5), 605.
Chapter 3
GeneEnvironment Interaction
intheBehavioral Sciences: Findings,
Challenges, andProspects
MattMcGue andBridgetE.Carey
Introduction
What is GeneEnvironment Interaction?
More than 60 years ago, the eminent psychologist Anastasi (1958) famously
declared an end to the acrimonious NatureNurture debate. She argued that the
debate was no longer productive because research had clearly established that both
genetic and environmental factors make fundamental contributions to individual
differences in behavior. While Anastasi may have been a bit premature in declaring
an end to the intellectual hostilities, researchers on both sides of the debate have
generally embraced Anastasis admonition that the key question is not so much
whether but rather how genes and the environment combine to influence behavior.
The phenomenon of geneenvironment interaction (GE) is seen by many to be key
to meeting Anastasis challenge of determining how (Rutter & Silberg, 2002).
While there may be general agreement as to the significance of GE, there has
been considerable difference of opinion as to what GE constitutes exactly (and
also, as we will see later, the extent to which behavioral geneticists have been able
to reliably document its existence). For some, GE simply means that behavior is
influenced by both genetic and environmental factors. Since complex behavioral
phenotypes are never genetically determined (Turkheimer, 2000), however, acknow
ledging this truism can do little to advance true understanding. A more broadly
endorsed and useful conceptualization of GE is that it involves some synergistic
relationship between genetic and environmental influences; that the p henotypic
M. McGue (*) B.E. Carey

Department of Psychology, University of Minnesota,
75 East River Rd, Minneapolis, MN 55455, USA
e-mail: mcgue001@umn.edu; carey194@umn.edu

DOI10.1007/978-3-319-49227-8_3
36 M. McGue and B.E. Carey
whole is more than the simple sum of its genetic and environmental parts. Yet even
in this case, there are differences of opinion as to what GE is.
Tabery and his colleagues (Griffiths & Tabery, 2008; Tabery, 2008, 2014) have
traced the historical origins of two alternative conceptualizations of GE.The clas
sical, or biometric, definition of GE traces its roots to the seminal work of Fisher
(1918), who defined GE statistically within the analysis of variance framework he
developed. Specifically, biometricians seek to apportion variance (i.e., a statistic
first introduced by Fisher in his 1918 paper as a measure of individual differences in
a phenotype) among underlying genetic and environmental components. From a
biometric perspective, GE occurs when the phenotypic variance is not accounted
for by an additive function of a main effect of the genotype and a main effect of the
environment. A biometric GE is a specific example of a statistical interaction, that
is, where the effect of one variable is moderated by the other (De Gonzalez & Cox,
2007). A biometric GE reflects that the magnitude of the genetic influence depends
upon the level of the environment, or equivalently that the environmental influence
is conditioned on genotype.
An alternative conceptualization of GE is rooted more in developmental b iology
than statistics and is thus sometimes called developmental GE.This conceptual
ization can be traced to the contributions of Lancelot Hogben, Conrad Waddington,
and Gilbert Gottlieb (Tabery, 2008, 2014). The British geneticist Hogben was inter
ested in explaining individual development, which he argued was a result of the
interplay between a particular genotype and a particular environment that plays out
across development. From a developmental perspective, GE is ubiquitous, as an
individual organism is always a consequence of a unique combination of genetic
and environmental factors (Gottlieb, 1995). The goal for Hogben was to understand
how the phenotype was the result of this joint action.
While the developmental notion of GE may come closest to addressing
Anastasis call for research focused on how, the biometric model has been the
predominant approach taken to the study of GE within psychology and psychiatry
(indeed within human genetics more broadly) and so will be our focus here. A sta
tistical formulation of GE cannot fully answer the question How? Nonetheless,
it is a clear advance beyond the standard biometric approaches to estimating herita
bility that have dominated much behavioral genetic research and is arguably a key
first step to ultimately understanding the unique nature of the interplay of genotype
with experience in the course of individual development.
Shanahan and Hofer (2005) have provided a useful typology of biometric GE
(from here on, designated simply as GE). Here we highlight the two forms of GE
they identify that have received the most consistent attention in the behavioral
genetic literature. The diathesisstress model (Fig. 3.1a) has provided a conceptual
foundation for much of psychopathology research. In this model, genetic back
ground (i.e., diathesis) is conceptualized as establishing a level of vulnerability that
manifests as psychopathology only when triggered by exposure to a provocative
environment (i.e., stress). In the diathesisstress formulation, the effect of genotype
is minimized in a low-risk environment and maximized in a high-risk environment.
Alternatively, under the social control model of GE (Fig. 3.1b), the impact of
3 GeneEnvironment Interaction 37
Fig. 3.1 Alternative forms of genotypeenvironment interaction. In the diathesisstress model

(a), underlying genetic risk only manifests as pathology if triggered by environmental stress. In the
social control model (b), the impact of genetic risk on the frequency of psychopathology is mini
mized in environments characterized by high levels of monitoring or social control
genetic risk is minimized in environments in which the individuals capacity to

make choices is constrained by social norms or through close monitoring by author
ities and maximized in environments where individuals are largely free to pursue
their own interests.
hy is GenotypeEnvironment Interaction Considered

W
Important?
Several commentators have noted that there has been an exponential increase in
GE publications over the past decade (Dick, 2011; Manuck & McCaffery, 2014).
In the behavioral sciences, this growth in interest has sometimes been ascribed to
two seminal demonstrations of GE published by Caspi and colleagues in the early
2000s (Caspi etal., 2002, 2003). Figure 3.2 plots number of publications including
key terms involving GE by year of publication as determined using Thomson-
Reuters Web of Science. The number of publications is reported separately for
journals devoted to the behavioral and social sciences versus all other journals. The
plot confirms an exponential increase in GE publications over the past decade. The
plot does not, however, suggest that the Caspi papers, although clearly important
and heavily cited, had a singular effect on this growth. There was no noticeable
upsurge in GE publications following the two Caspi papers and the plot for publi
cations in social and behavioral science journals parallels that for all other journals.
(Note that behavioral research can be published in journals that are not devoted to
behavioral science, as was the case for the two Caspi et al. papers published in
Science, so that the separation between the two is not altogether perfect.)
Technological advances provide an alternative explanation for the growth in
GE research. Over the past 15years the efficiency of genotyping has increased
markedly as costs have dropped. As a consequence, researchers can now explore
models of GE they could only speculate about before. Yet while technological
advances might account for the marked increase in GE publications, other factors
Fig. 3.2 Number of publications with geneenvironment interaction key words as determined
from Thomson-Reuters Web of Science. Number of publications plotted separately for journals
in the behavioral and social sciences and all other journals
are needed to explain why researchers accord the topic great significance.
Researchers have identified many reasons to be interested in GE (Boffetta etal.,
2012; Hunter, 2005); we highlight several of the more important among these here.
First, GE may help explain a long-standing and perplexing observation within
psychologythere are marked differences in how individuals respond and react to
the same environmental provocation (Moffitt, Caspi, & Rutter, 2006). For example,
childhood maltreatment can result in an internalizing disorder for some, an external
izing disorder for others, and no discernable pathology for many (Anda etal., 2006).
Differential genetic vulnerability to internalizing and externalizing psychopathol
ogy may help explain this heterogeneous response to maltreatment.
Second, statistical GE may help identify the biological pathways underlying
disease and characterize how those pathways are modulated by exogenous factors
(Hunter, 2005), although to date the number of cases where the finding of a statisti
cal interaction has provided real biological insight is admittedly limited (Berrington
de Gonzales & Cox, 2007). Finally, GE is key to translating findings from genom
ics research into personalized approaches to medical care (Chan & Ginsburg, 2011;
Offit, 2011). Tailoring prevention efforts or clinical interventions to an individuals
genotype requires that individuals with different genotypes respond differently to
different treatments. There are of course many examples of the need for individual
ized treatment with Mendelian (i.e., single-gene) disorders, the geneticists favorite
example being phenylketonuria. Currently, there are, however, few examples of GE
interaction leading to differential clinical treatment with complex phenotypes.
Traumatic brain injury (TBI) and the ApoE-4 risk factor for late-onset dementia
provides one example for how GE might inform individualized treatment for a
complex phenotype. Carriers of the 4 allele appear to be especially vulnerable to the

clinical consequences of repeated head trauma. Greater TBI vulnerability among 4
carriers versus noncarriers has, for example, been reported in boxers (Jordan etal.,
1997), football players (Kutner, Erlanger, Tsai, Jordan, & Relkin, 2000), as well as
among those who suffer non-sports-related head injuries (Teasdale, Nicoll, Murray,
& Fiddes, 1997). Although additional evidence may be needed to support implemen
tation of strong policies (Baugh, Robbins, Stern, & McKee, 2014), research on ApoE
and TBI has motivated discussions concerning whether there should be screening for
4 status prior to participation in sports that involve a heightened risk of concussion.
Methods forDetecting GenotypeEnvironment Interactions
Latent-Variable Approach
GE occurs whenever the effect of the genotype is conditional on environmental

exposure. A common index of genetic effect is the heritability, defined as the pro
portion of phenotypic variance attributable to genetic factors (Visscher, Hill, &
Wray, 2008). Heritability is often estimated by a twin study, so that one approach to
detecting GE involves stratifying twin pairs by exposure on some environmental
factor and then determining whether heritability varies across the strata (Purcell,
2002). In this approach, since the genotypic effect is assessed indirectly, by compar
ing monozygotic to dizygotic twin similarity, rather than directly, by genotyping the
relevant causal genetic variant, inference is about a latent, aggregate genetic effect
rather than a specific genotype.
One of the most highly cited GE findings in the behavioral genetic literature
helps to illustrate the latent-variable approach. Turkheimer and colleagues (Turk
heimer, Haley, Waldron, DOnofrio, & Gottesman, 2003) stratified a sample of
twins according to their rearing socioeconomic status (SES) and then estimated the
heritability of IQ across the SES strata. They found that the heritability of IQ was
minimized in an impoverished rearing environment and maximized in an advan
taged one. That is, it appeared that individuals failed to fully reach their genetic
potential for intellectual achievement if they were reared in socioeconomically dis
advantaged homes.
In terms of psychopathology, the two forms of GE we highlighted from the
Shanahan and Hofer (2005) paper receive considerable support from research tak
ing the latent-variable GE approach. For example, in the area of substance use and
abuse, a consistent body of research has shown that genetic influences are attenuated
by elements of social control. The heritability of smoking is lower among adoles
cents who are closely rather than poorly monitored by their parents (Dick et al.,
2007) and for smokers of all ages within states that prohibit smoking in public
places, have high cigarette taxes, or restrict cigarette sales and advertisement versus
states that have policies that are more permissive of smoking (Boardman, 2009;
Boardman, Blalock, & Pampel, 2010). Religion, an aspect of social control, has also
been shown to moderate genetic influences on substance use, with genetic influ
ences on adolescent alcohol use being lowest among individuals who are highly
religious or being raised in a religious household (Button, Hewitt, Rhee, Corley, &
Stallings, 2010; Koopmans, Slutske, van Baal, & Boomsma, 1999). Additionally,
heritability of alcohol use is lower in rural environments than in urban environ
ments, presumably because rural communities place greater constraints on adoles
cent access to alcohol than urban communities (Legrand, Keyes, McGue, Iacono, &
Krueger, 2008; Rose, Dick, Viken, & Kaprio, 2001).
The diathesisstress model has also received considerable support from latent-
variable GE research. Peer affiliation, for example, appears to be a particularly
powerful moderator of genetic influences on externalizing behaviors. Harden, Hill,
Turkheimer, and Emery (2008) investigated the effect of best-friend substance use
on the substance use of adolescents, finding evidence of a triggering effect such that
adolescents at high genetic risk for substance abuse were found to be much more
sensitive to the substance use behavior of their best friends than adolescents who
were at low genetic risk. The heritability of conduct problems is similarly greater
among adolescents who have multiple delinquent peers than among those who have
few (Beaver, DeLisi, Wright, & Vaughn, 2009). The magnitude of genetic influ
ences on conduct problems and antisocial behavior also appears to be modulated by
aspects of parenting. Punitive disciplinary practices by fathers, negative and less
warm parenting styles, and relationship problems with either parent, for example,
all appear to have triggering effects on genetic risk for externalizing psychopathol
ogy (Button, Lau, Maughan, & Eley, 2008; Feinberg, Button, Neiderhiser, Reiss, &
Hetherington, 2007; Hicks, South, DiRago, Iacono, & McGue, 2009). Other factors,
such as poor academic achievement and engagement and stressful life events, also
appear to more strongly contribute to externalizing behaviors among those at higher
versus lower genetic risk (Hicks etal., 2009). These studies all tend to converge on
similar findings, indicating that a genetic vulnerability for externalizing behaviors
can be triggered by adverse environmental factors, such as delinquent peers, paren
tal relationship problems, or a stressful environment.
Candidate-Gene Environment Approach
The latent-variable gene environment interaction approach neither postulates nor

identifies the specific genetic markers or systems involved in the interaction. As a
consequence, this approach provides limited biological insights, even if it can help
to identify the relevant environmental factors and characterize the situations in
which those factors modulate genetic influences. In contrast, the candidate-gene
environment (cGE) approach, which involves testing for an interaction between a
specific genotyped variant and a specific environment, is biologically informed
(Moffitt, Caspi, & Rutter, 2005). The landmark studies by Caspi and colleagues are
examples of cGE studies. The first of these concerned the origins of violence in
males and involved a functional variant in the promoter region of the MAO-A gene,
a target of considerable research on the genetics of aggression, on the X chromo

some and a history of childhood maltreatment (Caspi etal., 2002). Consistent with
a diathesisstress conceptualization, these researchers reported that those with the
high-risk MAO-A genotype were more sensitive to being maltreated than those with
the low-risk genotype. The second Caspi etal. (2003) cGE study was also consis
tent with a diathesisstress conceptualization. In this case, the researchers were
interested in the origins of depression and targeted a functional polymorphism in the
promoter region of the serotonin transporter gene, which has been hypothesized to
be linked to depression, and the experience of psychological stress. Again, individu
als with the high-risk genotype were found to be more sensitive to environmental
exposurethe genotypic effect on depression risk was minimal in a low-stress envi
ronment but large in a high-stress environment.
There are several excellent recent reviews of cGE research in behavioral and
psychiatric genetics, including those by Dick (2011), Manuck and McCaffery (2014),
and Duncan, Pollastri, and Smoller (2014). As with the latent-variable approach,
evidence for both the diathesisstress and social control forms of GE can be found
in these reviews. In addition to the two Caspi etal. studies mentioned previously,
there are many reports of cGE that are consistent with a diathesisstress formula
tion. These include, for example, reports that prenatal smoking interacts with a
genetic variant in the dopamine transporter gene in predicting symptoms of ADHD
(Becker, El-Faddagh, Schmidt, Esser, & Laucht, 2008), maternal insensitivity inter
acts with a genetic variant in the dopamine D4 receptor gene in predicting external
izing symptoms (Bakermans-Kranenburg & van Ijzendoorn, 2006), and children
who carry a high-risk allele at the dopamine D4 locus are more sensitive to the
effects of poor parenting than children who carry a low-risk allele (Sheese, Voelker,
Rothbart, & Posner, 2007). Although there are only a few relevant studies, there are
also cGE reports consistent with a social control form of interaction. For example,
Dick and colleagues (2009) reported that the association of variation in the GABRA2
gene with externalizing behavior was diminished among adolescents whose parents
closely monitored their behavior. In a subsequent study with the same adolescent
sample, Dick etal. (2011) reported that the association of variants in the CHRM2
gene with externalizing behavior was reduced by close parental monitoring.
Issues withtheCandidate GE Approach
Even if the Caspi etal. studies did not directly lead to the explosion of interest in
GE research, their cGE approach has served as a model for many behavioral
scientists. Within the past few years, however, several critical reviews of the cGE
literature in psychiatry and psychology have been published (Duncan & Keller,
2011; Duncan etal., 2006, 2014; Munaf, Zammit, & Flint, 2014), with some critics
going so far as to conclude that methodological flaws preclude drawing any firm
conclusions from existing cGE research in the behavioral sciences. The limitations
of the cGE approach, at least as applied in the behavioral sciences, can be seen in
the status of the influential Caspi et al. studies. Despite more than 10 years of
research and scores of attempts at replication, the field continues to debate the valid
ity of these findings (Munaf etal., 2014; Rutter, 2014). Several methodological
limitations of cGE research have been identified.
Power
Before discussing concerns surrounding the power of cGE studies, it is informative

to consider the current status of research aimed at identifying genetic main effects.
For nearly 20years, a candidate-gene study was the major approach for identifying
specific genetic associations. We now know that the overwhelming majority of these
studies were grossly underpowered and that despite considerable effort the candi
date-gene approach yielded few replicable findings (Ioannidis, 2005). Colhoun,
McKeigue, and Smith (2003) estimated that 95% of published reports of significant
associations in candidate-gene studies were likely false positives. The simulation
studies by Sullivan (2007) provide some insight as to how such a high rate of false
positives might be achieved. Simulating data under various realistic scenarios where
there was no genetic association, Sullivan (2007) shows that there is a greater than
95% chance of a publishable result (i.e., an association at p<.05) through multiple
statistical testing that is neither reported nor corrected for in the analysis.
The poor record of candidate-gene studies has resulted in the genome-wide asso
ciation study (GWAS) becoming the preferred method for identifying genetic asso
ciations in human genetics. In a GWAS, 500,000 or more single nucleotide
polymorphisms (SNPs), selected because they span the genome rather than have
any presumed biological relevance, are each tested for association with the target
phenotype. Despite the need for large samples to account for the multiple testing
burden, GWAS has in a relatively short period of time produced thousands of reli
able genetic associations for multiple complex phenotypes (Visscher, Brown,
McCarthy, & Yang, 2012). Of interest, despite large GWAS samples, associations
first established in candidate-gene studies have typically not been supported in
GWAS (Bosker et al., 2011; Olfson & Bierut, 2012; Siontis, Patsopoulos, &
Ioannidis, 2010), providing further evidence that the existing candidate-gene litera
ture is likely plagued with false-positive reports.
GWAS has revealed two key features of the genetic architecture of non-Mendelian
(i.e., complex) phenotypes that we believe have implications for GE research. First,
the number of genetic variants that underlies the heritability of any complex human
phenotype is likely to be very large, in the thousands if not tens of thousands. Second,
and almost as a corollary to the first, the contribution of any specific genetic variant
to phenotypic variance is likely to be very small. Based on an analysis of early GWAS
findings, Park etal. (2010) concluded that any single SNP association will typically
account for much less than .5% of the genetic variance, with SNPs that account for
as much as .5% of the variance being rare and those accounting for as much as 1.0%
being exceedingly rare. Based on GWAS published subsequent to their analysis, the
Table 3.1 Effect sizes reported in large-scale genome-wide association studies

% of phenotype
variance accounted
for by
Biometric # All
heritability Significant significant Average
Phenotype Study (%) GWAS N SNPS SNPS (%) SNP (%)
BMI Speliotes 5080 249,796 32 1.5 .05
etal.
(2010)
Height Wood etal. 7085 253,288 423 15.9 .04
(2014)
Serum urate Kottgen 4070 143,160 26 7.0 .27
etal.
(2013)
Total Willer 5075 188,577 157 15.0 .10
cholesterol etal.
(2013)
C-reactive Dehghan 2540 82,725 18 5.0 .28
protein etal.
(2011)
projections made by Park etal. (2010) may have been, if anything, a bit optimistic.
Table 3.1 gives characteristics of SNP discoveries in several large GWAS.The aver
age SNP effect is quite small, averaging as little as .04% of the variance for height.
And although some variants accounting for as much as 1% of variance have been
identified (e.g., for C-reactive protein and serum urate), no SNP accounting for as
much as .5% of variance has been found for either height or body mass index (BMI).
Because it is easy to describe scenarios where there is no relationship between
main and interaction effects, it is reasonable to ask whether GWAS findings have
any implications for GE research. We believe there are several. First, a cGE study
is a simple extension of the logic of a candidate-gene test for main effects, and if
nothing else the poor track record of the latter should lead us to be very cautious in
judging the reliability of findings from the former. Second, if for any given pheno
type there are hundreds maybe thousands of relevant genetic variants, then there
may be a similar large number of individual GE effects, no one of which has a
large effect on phenotype. Large-scale investigations with nonbehavioral pheno
types support the conclusion that the magnitude of any single specific GE contri
bution is unlikely to be large. For example, in a sample of nearly 80,000 women,
Barrdahl etal. (2014) found no evidence of an interaction between a set of repli
cated genetic and a set of well-established environmental risk factors for breast
cancer. Epidemiologists have an old rule of thumb that tests for interaction require
four times the sample size for comparably powered tests for main effects (Smith &
Day, 1984). Although this simple rule cannot possibly cover all possible scenarios,
it certainly suggests that power may be a concern in cGE research given the large
sample sizes considered necessary for tests of genetic main effects.
Power in a cGE study is a complex function of the distribution and magnitude

of the genetic main effect, the distribution and magnitude of the environmental main
effect, and the form of the GE. To illustrate the issues surrounding power, we
derived the power function under a single scenario, which is in several ways advan
tageous for detecting GE.Specifically, we modeled environmental exposure at two
equally probable levels accounting for 3% to 12% of the phenotypic variance,
genetic background as additive with an allele frequency of .50, and GE as a pure
interaction. In a pure interaction, there is an effect of genotype at one of the levels
of environmental exposure but not the other. The results of these power calculations
are shown in Fig. 3.3.
Duncan and Keller (2011) computed a median sample size of 345 for cGE stud
ies in psychiatry. At that sample size and for the example illustrated in the figure,
adequate power would exist to detect an interaction in which the genetic effect
accounted for 8% of phenotypic variance at one level of the environment and 0% of
variance at the other. Alternatively, to have power of at least 80% to detect an inter
action where there was no genetic effect at one level of environmental exposure but
a genetic effect accounting for 1% of phenotypic variance at the other level (i.e., at
the very upper end of the distribution of individual genetic effects) would require a
sample size of approximately 3000 individuals. In their review of existing cGE
research in psychiatry, Duncan and Keller (2011) come to a conclusion similar to
that reached by Colhoun etal. (2003) in their review of main effects research, the
Fig. 3.3 Sample size needed to achieve power of at least 80% to detect a pure GE at =.05.
Pure GE was modeled in terms of two environments where there is no genotypic effect in one
environment but a genotypic effect in the second environment that accounts for r2G(E) of the pheno
typic variance. The vertical dashed line is at an r2G(E)=.01, which Park etal. (2010) reported to be
at the highest end of the distribution of genetic main effects reported in GWAS; the horizontal
dashed line is at N=345, which Duncan and Keller (2011) reported is the median sample size in
existing candidate-geneenvironment research in psychiatry
vast majority of published studies are underpowered and consequently most of the
reported findings are likely false-positive results.
Measurement Error
Measurement error in any of the components of a cGE study (i.e., phenotype,

environmental exposure, and genotype) can attenuate the magnitude of the interac
tion effect and result in a loss of statistical power. Since interaction models involve
fitting a compound function of G and E, measurement error becomes a particular
concern when testing for GE. Wong, Day, Luan, Chan, and Wareham (2003)
showed how errors in the measurement of both exposure and outcome can substan
tially reduce the sensitivity of cGE analyses. For example, they show that the sam
ple size would need to be more than doubled to compensate for the loss of power
that results from a decrease of as little as 10% in the reliability of both the exposure
and outcome measures.
The rate of genotyping error is generally low, especially for SNPs (Ragoussis,
2009), which might lead one to conclude that errors in genotyping may not be of
much concern in cGE analyses. Markers on most standard genotype arrays were,
however, selected because they could predict (i.e., tag through linkage disequilib
rium) nearby markers and not because they were thought to be causal variants
(Zondervan & Cardon, 2004). Thus many of the genetic variants that are included in
GE research may only be correlated with the actual causal variants. The lack of
perfect correlation between genotyped marker and actual causal variant is analo
gous to measurement error and can attenuate effects in a GE analysis (Hein,
Beckmann, & Chang-Claude, 2008). Significantly, while measurement error gener
ally biases effects toward the null, Dudbridge and Fletcher (2014) have shown that
misclassification of the genotype (as occurs when a tagging marker rather than the
causal variant is used) can lead to spurious findings of GE when there is gene
environment correlation.
GE is Not Scale Invariant
A GE is the result of applying a particular statistical model to a set of data. As

such, it is not an inherent property of the individuals being studied but rather derives
meaning only in the context of the specific statistical model fit (Clayton & McKeigue,
2001). Figure 3.4 illustrates the nature of this issue when the phenotype is categori
cal (e.g., disease status). In this case, there is evidence for an interaction on an addi
tive scale (i.e., the difference in disease prevalence between those that are and are
not exposed depends on genotype) but no evidence for interaction on a multiplica
tive scale (i.e., prevalence among those who are exposed is always three times the
rate among those who are not exposed regardless of genotype.)
0.4
2 Risk
Exposure to Environmental Alleles
# Risk Risk
Alleles 0.3
No Yes
1 Risk
Allele
0 .05 .15 0.2
0 Risk
Alleles
1 .075 .225
0.1
2 .1125 .3375
0
No Yes
Environmental Exposure
Fig. 3.4 Data illustrating an interaction on an additive but not a multiplicative scale. On an addi
tive scale, the effect of environmental exposure is to increase disease risk by .10 (i.e., .15.05)
among those with no risk alleles but .225 among those with two risk alleles, suggesting greater
vulnerability of the latter genotype. On a multiplicative scale, however, environmental exposure
increases disease risk by a multiplicative factor of 3 independent of genotype and each allele
increases disease risk by a factor of 1.5 independent of environment
The parallel issue with a quantitative phenotype concerns whether evidence for
interaction would be eliminated through transformation of the phenotypic measure
ment scale. For example, evidence for an interaction on the raw phenotypic scale
may be eliminated if the phenotype was log transformed (a situation analogous to
fitting an additive versus multiplicative model to disease prevalence.) Eaves (Eaves,
2006; Eaves & Verhulst, 2014) has written extensively on the problem of scaling
when testing for GE, showing that completely additive data can lead to the false
finding of an interaction when data are analyzed under the standard assumption of a
normal distribution.
There is one form of interaction which cannot be fully eliminated through scale
transformation. In a disordinal (sometimes called cross-over or qualitative) interac
tion, there is a reversal of the effect of one variable as a function of level on the other
(Thompson, 1991). Figure 3.5 provides an illustration of a disordinal cGE.The asso
ciation of genotype with disease risk is reversed in an environment characterized by
high support versus an environment characterized by low support. Rather than being
characterized along a continuum of risk (cf., the diathesisstress model), genotypes
are characterized as differentially sensitive to environmental influence. In this illustra
tion, in the low-support environment, the low-sensitive genotype experiences the low
est risk and the high-sensitive genotype the highest risk of disease. This pattern is
reversed in the high-support environment, where the high-sensitive genotype experi
ences the lowest risk of disease. Although an intriguing idea, the extent to which there
is reliable evidence for disordinal interactions in behavioral or psychiatric genetics
continues to be debated (Belsky etal., 2009), with recent evidence questioning the
robustness of at least some previous findings in this area (Belsky etal., 2015).
Genotype
Low-Sensitivity
Medium-Sensitivity
Disease Risk
High-Sensitivity
High Low
Level of Environmental Support
Fig. 3.5 Example of a disordinal (cross-over, qualitative) interaction. In this form of GE, the
association of genotype with disease risk is reversed at the two extremes of environmental expo
sure. The evidence supporting a disordinal interaction cannot be completely eliminated through a
monotonic transformation of the phenotypic scale
Addressing theLimitations ofGE Research
Increasing Sample Size
Duncan and Keller (2011) identified small sample size as probably the single great
est impediment to progress in cGE research in the behavioral sciences. Even
though there is general recognition that larger samples are needed to test for interac
tions as compared to main effects, it is very rare for a cGE study to have a sample
that exceeds 1000, and a sample of that size would be considered inadequate for
testing main effects given what we have learned from GWAS over the past 5 years.
Amassing a sample large enough to undertake large-scale genomics research is no
longer within the capabilities of a single investigator or research lab, leading human
geneticists to establish multi-investigator consortia. In the behavioral sciences, the
most prominent among these include the Psychiatric Genomics Consortium
(Sullivan, 2010), the ENIGMA neuroimaging consortium (Thompson etal., 2014),
and the Social Science Genetics Association Consortium (Rietveld etal., 2013).
Although genomics consortia were initially organized around testing for genetic
main effects, they could in principle be extended for the purpose of testing for
GE. Data harmonization, however, represents a major challenge to consortia
research on GE. Imputation provides an effective framework for harmonizing
genetic data (Marchini & Howie, 2010), and although assessment approaches may
differ, the standardized criteria in the DSM or ICD often provide a workable frame
work for harmonizing a disease phenotype. Unfortunately, until recently there have
been limited attempts to standardize the assessment of environmental risk (Hamilton

etal., 2011) as a way to address concerns as to whether post hoc harmonization of
disparate measures increases measurement error (Fortier et al., 2011). Moreover,
harmonization across multiple studies often results in the richness that characterizes
the environmental assessment in any given study being sacrificed in order to achieve
a measure that is comparable across multiple studies. In two recent large-scale con
sortia investigations of GE for obesity, for example, environmental exposure was
reduced to a dichotomy, physically active versus inactive (Ahmad et al., 2013;
Kilpelinen etal., 2011). Fortunately, the growing interest in pooling data has moti
vated several attempts to develop standardized assessment protocols, such as the
PhenX project (Hamilton etal., 2011).
Alternative Approaches toAssessment
The sensitivity of tests for GE could be improved by increasing the accuracy of

measurement, although there is often a trade-off between achieving a large sample
versus having in-depth assessments. Nonetheless, based on their simulations, Wong
etal. (2003) concluded that the sensitivity of tests for GE could in certain situa
tions benefit more from improved measurement than from enlarging sample size.
Because of the high accuracy of genotyping and improved methods of imputation,
there probably is little loss in power due to errors in genotyping, even when using a
proxy rather than the true causal variant. Consequently, there is greater need to
improve measurement of the phenotype and environmental exposure than genotype.
Highly reliable assessment typically requires in-person testing, however, which can
be expensive and difficult to implement in large-scale genomics studies. Adding
further complexity is that for many psychiatric disorders the relevant environmental
exposure plays out over a lifetime rather than being contemporaneous with the time
of assessment (Sameroff, Seifer, & Bartko, 1997). Online methods of assessment,
which are increasingly used in epidemiology, have the potential to address some of
the measurement problems inherent to GE research without necessarily sacrificing
sample size (van Gelder, Bretveld, & Roeleveld, 2010).
It is interesting to contrast the general difficulty human behavioral geneticists
have had in reliably identifying GE effects with the widespread evidence for GE
that has been reported in research with experimental organisms (Wahlsten, 1990). A
major difference between the two research literatures concerns control over envi
ronmental exposure, which does not typically exist in studies with humans but does
in research on model organisms. When given control, experimentalists typically
maximize the extremes of environmental exposure in their study, and it is at the
most extreme levels of environmental exposure where GE is likely to be most
evident (Henderson, 1990). Consistent with this explanation for why the two GE
literatures diverge is the observation that in nonhuman species the evidence for GE
is greater in laboratory studies, where experimental manipulation is possible, than
in the wild, where it is not (Charmantier & Garant, 2005). Regardless, several inves
tigators have sought to increase the sensitivity of GE research with humans by
preferentially sampling individuals at extreme levels of environmental exposure by,

for example, sampling children with a documented history of child maltreatment
(Kaufman etal., 2006).
Endophenotypes represent another approach for trying to increase the sensitivity
of GE research. An endophenotype is a phenotype that is intermediate between
primary gene product and the observed clinical phenotype (Gottesman & Gould,
2003). In psychiatric genetics, endophenotypes are often derived from a brain imag
ing or brain wave assessments and so are expensive to obtain. Nonetheless, because
endophenotypes are more proximal to gene function, they are believed to greatly
increase the sensitivity of tests for both genetic main and interaction effects (Hyde,
Bogdan, & Hariri, 2011). An example that has been identified as a model of the neu
rogenetic approach to GE involves differential amygdala reaction to presentation of
an aversive face and the serotonin transporter 5-HTTLPR polymorphism (Caspi,
Hariri, Holmes, Uher, & Moffitt, 2010). In the first meta-analysis of research in this
area, 14 studies yielded a statistically significant pooled standardized effect size of
d=.63in favor of a GE (Munaf, Brown, & Hariri, 2008). A subsequent meta-
analysis of 34 samples gave an average Hedges effect size of g=.21, which continued
to be consistent with the original GE but was now just significant at p=.05 (Murphy
etal., 2013). The most recent meta-analysis was published in 2014, and reported a
pooled meta-analytic effect of g=.20, which was no longer statistically significant
(Bastiaansen etal., 2014). The declining effect size with increasing data may reflect
that early published studies in this area reported false-positive findings. If there is an
interaction effect involving amygdala function and 5-HTTLPR, it is likely substan
tially smaller in magnitude than what was reported in the original investigation.
Leveraging GWAS Findings
There are several ways by which findings from GWAS can be used to facilitate GE
research. The variants identified in large-scale GWAS can be used as candidate geno
types in GE analysis. There are, however, two caveats to this approach. First, it
assumes that the genetic variants that contribute to GE will generate a main effect
that can be identified in GWAS.With the exception of some types of disordinal inter
actions, the existence of a GE typically results in a genetic association, so that the
first assumption is generally justified. Second, we know that the magnitudes of
GWAS main effects are typically quite small. If interaction effects are usually smaller
than main effects, as many epidemiologists believe, then it might take truly massive
samples to detect interaction effects based on GWAS results for individual variants.
Research with a variant in the FTO gene (i.e., rs9939609) that has been identified
by GWAS as a risk factor for obesity provides a helpful illustration (Frayling etal.,
2007). Not unexpectedly, the FTO variant has a small effect on obesity risk, account
ing for less than .3% of the phenotypic variance in BMI.A meta-analysis in a sam
ple of more than 200,000 individuals found that the effect of rs9939609 on obesity
risk was 27% lower among individuals who were physically active versus those who
were not, a GE (Kilpelinen etal., 2011). Genetic moderation of pharmacologic
treatment for nicotine dependence provides one of the few applications of the use of
GWAS signals in GE research with a psychiatric phenotype. GWAS of smoking
has identified genetic variants in the nicotinic receptor gene cluster on chromosome
15 (CHRNA5CHRNA3CHRNB4) as being associated with developing nicotine
dependence among ever smokers (Bierut, Johnson, & Saccone, 2014). In a sample
of approximately 1000 smokers randomized to a pharmacological intervention ver
sus placebo, Chen et al. (2012) reported that genetic variation in the CHRNA5
CHRNA3CHRNB4 predicted response to treatment, a GE that would have
clinical application if it proved replicable in larger samples.
An alternative to using single GWAS variants is to aggregate all known genetic
variants into a polygenic risk score (PRS) for GE analysis. A PRS is computed by
taking a linear composite of SNP effects with each SNP weighted by the magnitude
of its association with phenotype (e.g., by the regression coefficient). A PRS conse
quently provides a composite index of genetic risk, although because most of the
heritability is missing these scores currently do a relatively poor job in forecasting
individual risk. In a sample of more than 110,000 individuals of European ancestry,
Ahmad etal. (2013) reported that the association of the PRS with BMI was reduced
among individuals who are physically active, a GE that is consistent with earlier
research using the single FTO variant. A subsequent study of approximately 3000
Han Chinese reported a similar attenuation of the PRS effect on BMI among indi
viduals who were physically active (Zhu etal., 2014). Again, there has been limited
application of this approach to investigating GE for behavioral phenotypes. A
recent preliminary report by Salvatore etal. (2014) reported that the strength of the
association of a PRS with alcohol problems was amplified under conditions of high
peer deviance or low parental monitoring. Researchers have begun to discuss how
gene- or pathway-based approaches could provide alternatives to the PRS for aggre
gating multiple genetic signals for GE research (Winham & Biernacka, 2013).
Finally, rather than limiting GE investigations to those genetic variants found to
be associated with phenotype in GWAS, some have proposed systematically search
ing the whole genome for evidence of GE.In a geneenvironment-wide interac
tion studies (abbreviated GEWIS and pronounced GE-Whiz), each of the variants
genotyped in a GWAS would be separately tested in a GE analysis (Thomas,
Lewinger, Murcray, & Gauderman, 2012). Given the large number of individual
GE effects that would be tested and the consequent need for massive samples,
most of the discussion surrounding GEWIS concerns the need for efficient research
designs and the approach has seen limited application, including in the behavioral
sciences (Winham & Biernacka, 2013).
Prospects forDetecting GE inIntervention Research
Individualized or genomic medicine is premised on the existence of GE, specifi

cally the assumption that genetic factors contribute to differential responses to treat
ment so that to be maximally effective interventions need to be tailored to genotype
(Manolio etal., 2013). Building the evidentiary database needed to support genomic
medicine approaches will be challenging. Consortia will be needed to achieve the
requisite large samples and pooling of data will be more effective if standardized
assessments are agreed to a priori than if data harmonization is attempted a poste
riori. One significant advantage of GE research in an intervention context is that by
having experimental control investigators can maximize differences in exposure in
a way that increases the likelihood of observing robust GE effects, much as is the
case in research with model organisms (Henderson, 1990).
Although there have been some initial attempts to identify gene intervention
interactions in the psychological literature, most of the relevant studies have had
relatively small samples with findings that have not as yet been subject to replica
tion (Brody et al., 2013). Previous experience with candidate-gene research sug
gests that it is too early to draw conclusions from this work. Although there is a
somewhat larger literature investigating gene intervention interactions involving
pharmacologic treatments, even here there is not sufficient data to draw firm conclu
sions (Hamilton, 2015). With respect to the pharmacologic treatment of depression,
for example, the clinical utility of genetic variation in the cytochrome p450 system
remains unclear (Perlis, 2014), there is not strong evidence that the 5-HTTLPR vari
ant predicts response to treatment (Crawford, Lewis, Lewis, & Munafo, 2013), and
the largest GWAS of treatment response (N=2256) reported no significant effects
(Uher, GENDEP Investigators, MARS Investigators, & STAR*D Investigators,
2013). Although the current status of gene environment interaction research may
seem disappointing, it is important to recognize that so did the initial GWAS reports
for psychiatric phenotypes. We have learned much over the past 10 years from
candidate-gene and GWAS research, which should help with designing better gene
intervention research in the future.
References
Ahmad, S., Rukh, G., Varga, T.V., Ali, A., Kurbasic, A., Shungin, D., DIRECT Consortium.
(2013). Gene physical activity interactions in obesity: Combined analysis of 111,421 individu
als of European Ancestry. PLoS Genetics, 9(7), e1003607. doi:10.1371/journal.pgen.1003607.
Anastasi, A. (1958). Heredity, environment, and the question "How?". Psychological Review,
65(4), 197208.
Anda, R.F., Felitti, V.J., Bremner, J.D., Walker, J.D., Whitfield, C., Perry, B.D., Giles, W.H.
(2006). The enduring effects of abuse and related adverse experiences in childhoodA conver
gence of evidence from neurobiology and epidemiology. [Review]. European Archives of
Psychiatry and Clinical Neuroscience, 256(3), 174186. doi:10.1007/s00406-005-0624-4.
Bakermans-Kranenburg, M.J., & van Ijzendoorn, M.H. (2006). Geneenvironment interaction of
the dopamine D4 receptor (DRD4) and observed maternal insensitivity predicting externalizing
behavior in preschoolers. Developmental Psychobiology, 48(5), 406409. doi:10.1002/
dev.20152.
Barrdahl, M., Canzian, F., Joshi, A.D., Travis, R.C., Chang-Claude, J., Auer, P.L., Campa, D.
(2014). Post-GWAS geneenvironment interplay in breast cancer: Results from the Breast and
Prostate Cancer Cohort Consortium and a meta-analysis on 79000 women. Human Molecular
Genetics, 23(19), 52605270. doi:10.1093/hmg/ddu223.
Bastiaansen, J.A., Servaas, M.N., Marsman, J.B. C., Ormel, J., Nolte, I.M., Riese, H., & Aleman,
A. (2014). Filling the gap: Relationship between the serotonin-transporter-linked poly
morphic region and Amygdala activation. Psychological Science, 25(11), 20582066.
doi:10.1177/0956797614548877.
Baugh, C.M., Robbins, C.A., Stern, R.A., & McKee, A.C. (2014). Current understanding of
chronic traumatic encephalopathy. [Review]. Current Treatment Options in Neurology, 16(9),
306. doi:10.1007/s11940-014-0306-5.
Beaver, K.M., DeLisi, M., Wright, J.P., & Vaughn, M.G. (2009). Geneenvironment interplay
and delinquent involvement evidence of direct, indirect, and interactive effects. Journal of
Adolescent Research, 24(2), 147168. doi:10.1177/0743558408329952.
Becker, K., El-Faddagh, M., Schmidt, M.H., Esser, G., & Laucht, M. (2008). Interaction of dopa
mine transporter genotype with prenatal smoke exposure on ADHD symptoms. Journal of
Pediatrics, 152(2), 263269. doi:10.1016/j.jpeds.2007.07.004.
Belsky, J., Jonassaint, C., Pluess, M., Stanton, M., Brummett, B., & Williams, R. (2009). Vulnerability
genes or plasticity genes? Molecular Psychiatry, 14(8), 746754. doi:10.1038/mp.2009.44.
Belsky, J., Newman, D.A., Widaman, K.F., Rodkin, P., Pluess, M., Fraley, R.C., Roisman,
G.I. (2015). Differential susceptibility to effects of maternal sensitivity? A study of candidate
plasticity genes. Development and Psychopathology, 27(3), 725746.
Berrington de Gonzales, A., & Cox, D.R. (2007). Interpretation of interaction: A review. Annals
of Applied Statistics, 1, 371385.
Bierut, L.J., Johnson, E.O., & Saccone, N.L. (2014). A glimpse into the futurePersonalized
medicine for smoking cessation. [Review]. Neuropharmacology, 76, 592599. doi:10.1016/j.
neuropharm.2013.09.009.
Boardman, J.D. (2009). State-level moderation of genetic tendencies to smoke. American Journal
of Public Health, 99(3), 480486. doi:10.2105/ajph.2008.134932.
Boardman, J.D., Blalock, C.L., & Pampel, F.C. (2010). Trends in the genetic influences on smok
ing. Journal of Health and Social Behavior, 51(1), 108123. doi:10.1177/0022146509361195.
Boffetta, P., Winn, D.M., Ioannidis, J.P., Thomas, D.C., Little, J., Smith, G.D., Khoury, M.J.
(2012). Recommendations and proposed guidelines for assessing the cumulative evidence on
joint effects of genes and environments on cancer occurrence in humans. International Journal
of Epidemiology, 41(3), 686704. doi:10.1093/ije/dys010.
Bosker, F.J., Hartman, C.A., Nolte, I.M., Prins, B.P., Terpstra, P., Posthuma, D., Nolen, W.A.
(2011). Poor replication of candidate genes for major depressive disorder using genome-wide
association data. Molecular Psychiatry, 16(5), 516532. doi:10.1038/mp.2010.38.
Brody, G.H., Beach, S.R. H., Hill, K.G., Howe, G.W., Prado, G., & Fullerton, S.M. (2013).
Using genetically informed, randomized prevention trials to test etiological hypotheses about
child and adolescent drug use and psychopathology. American Journal of Public Health, 103,
S19S24. doi:10.2105/ajph.2012.301080.
Button, T.M. M., Lau, J.Y. F., Maughan, B., & Eley, T.C. (2008). Parental punitive discipline,
negative life events and geneenvironment interplay in the development of externalizing
behavior. Psychological Medicine, 38(1), 2939. doi:10.1017/s0033291707001328.
Button, T.M. M., Hewitt, J.K., Rhee, S.H., Corley, R.P., & Stallings, M.C. (2010). The moderat
ing effect of religiosity on the genetic variance of problem alcohol use. Alcoholism: Clinical
and Experimental Research, 34(9), 16191624. doi:10.1111/j.1530-0277.2010.01247.x.
Caspi, A., McClay, J., Moffitt, T.E., Mill, J., Martin, J., Craig, I.W., Poulton, R. (2002). Role
of genotype in the cycle of violence in maltreated children. Science, 297, 851854.
Caspi, A., Sugden, K., Moffitt, T.E., Taylor, A., Craig, I., Harrington, H.L., Poulton, R. (2003).
Influence of life stress on depression: Moderation by polymorphism in the 5-HTT gene.
Science, 301, 386389.
Caspi, A., Hariri, A.R., Holmes, A., Uher, R., & Moffitt, T.E. (2010). Genetic sensitivity to the
environment: the case of the serotonin transporter gene and its implications for studying com
plex diseases and traits. American Journal of Psychiatry, 167(5), 509527. doi:10.1176/appi.
ajp.2010.09101452.
Chan, I. S., & Ginsburg, G. S. (2011). Personalized medicine: Progress and promise. In
A.Chakravarti & E.Green (Eds.), Annual review of genomics and human genetics (Vol. 12,
pp.217244). Palo Alto, CA: Annual Reviews.
Charmantier, A., & Garant, D. (2005). Environmental quality and evolutionary potential: Lessons
from wild populations. [Review]. Proceedings of the Royal Society B: Biological Sciences,
272(1571), 14151425. doi:10.1098/rspb.2005.3117.
Chen, L.S., Baker, T.B., Piper, M.E., Breslau, N., Cannon, D.S., Doheny, K.F., Bierut, L.J.
(2012). Interplay of genetic risk factors (CHRNA5-CHRNA3-CHRNB4) and cessation treat
ments in smoking cessation success. American Journal of Psychiatry, 169(7), 735742.
doi:10.1176/appi.ajp.2012.11101545.
Clayton, D., & McKeigue, P.M. (2001). Epidemiological methods for studying genes and environ
mental factors in complex diseases. [Review]. Lancet, 358(9290), 13561360. doi:10.1016/
s0140-6736(01)06418-2.
Colhoun, H. M., McKeigue, P. M., & Smith, G. D. (2003). Problems of reporting genetic
associations with complex outcomes. [Review]. Lancet, 361(9360), 865872. doi:10.1016/
s0140-6736(03)12715-8.
Crawford, A.A., Lewis, G., Lewis, S.J., & Munafo, M.R. (2013). Systematic review and meta-
analysis of serotonin transporter genotype and discontinuation from antidepressant treatment.
[Review]. European Neuropsychopharmacology, 23(10), 11431150. doi:10.1016/j.euro
neuro.2012.12.001.
De Gonzalez, A.B., & Cox, D.R. (2007). Interpretation of interaction: A review. [Review]. Annals
of Applied Statistics, 1(2), 371385. doi:10.1214/07-aoas124.
Dehghan, A., Dupuis, J., Barbalic, M., Bis, J.C., Eiriksdottir, G., Lu, C., Chasman, D.I. (2011).
Meta-analysis of genome-wide association studies in > 80 000 subjects identifies multiple loci for
C-reactive protein levels. Circulation, 123(7), 731U151. doi:10.1161/circulationaha.110.948570.
Dick, D.M. (2011). Geneenvironment interaction in psychological traits and disorders. Annual
Review of Clinical Psychology, 7, 383409. doi:10.1146/annurev-clinpsy-032210-104518.
Dick, D.M., Viken, R., Purcell, S., Kaprio, J., Pulkkinen, L., & Rose, R.J. (2007). Parental moni
toring moderates the importance of genetic and environmental influences on adolescent smok
ing. Journal of Abnormal Psychology, 116(1), 213218.
Dick, D.M., Latendresse, S.J., Lansford, J.E., Budde, J.P., Goate, A., Dodge, K.A., Bates, J.E.
(2009). Role of GABRA2in trajectories of externalizing behavior across development and evi
dence of moderation by parental monitoring. Archives of General Psychiatry, 66(6), 649657.
Dick, D.M., Meyers, J.L., Latendresse, S.J., Creemers, H.E., Lansford, J.E., Pettit, G.S.,
Huizink, A.C. (2011). CHRM2, parental monitoring, and adolescent externalizing behavior:
Evidence for geneenvironment interaction. Psychological Science, 22(4), 481489.
doi:10.1177/0956797611403318.
Dudbridge, F., & Fletcher, O. (2014). Geneenvironment dependence creates spurious geneenvi
ronment interaction. American Journal of Human Genetics, 95(3), 301307. doi:10.1016/j.
ajhg.2014.07.014.
Duncan, A.E., Scherrer, J., Fu, Q., Bucholz, K.K., Heath, A.C., True, W.R., Jacob, T. (2006).
Exposure to paternal alcoholism does not predict development of alcohol-use disorders in off
spring: Evidence from an offspring-of-twins study. Journal of Studies on Alcohol, 67(5),
649656.
Duncan, L.E., & Keller, M.C. (2011). A critical review of the first 10years of candidate gene-by-
environment interaction research in psychiatry. [Review]. American Journal of Psychiatry,
168(10), 10411049. doi:10.1176/appi.ajp.2011.11020191.
Duncan, L.E., Pollastri, A.R., & Smoller, J.W. (2014). Mind the gap why many geneticists and
psychological scientists have discrepant views about geneenvironment interaction (GE)
research. American Psychologist, 69(3), 249268. doi:10.1037/a0036320.
Eaves, L., & Verhulst, B. (2014). Problems and pit-falls in testing for G E and epistasis in candi
date gene studies of human behavior. Behavior Genetics, 44(6), 578590. doi:10.1007/
s10519-014-9674-6.
Eaves, L.J. (2006). Genotype environment interaction in psychopathology: Fact or artifact? Twin
Research and Human Genetics, 9(1), 18. doi:10.1375/183242706776403073.
Feinberg, M.E., Button, T.M. M., Neiderhiser, J.M., Reiss, D., & Hetherington, E.M. (2007).
Parenting and adolescent antisocial behavior and depressionEvidence of genotype parent
ing environment interaction. Archives of General Psychiatry, 64(4), 457465. doi:10.1001/
archpsyc.64.4.457.
Fisher, R.A. (1918). The correlation between relatives on the supposition of Mendelian inheri
tance. Transactions of the Royal Society of Edinburgh, 52, 399433.
Fortier, I., Doiron, D., Little, J., Ferretti, V., LHeureux, F., Stolk, R. P., International
Harmonization. (2011). Is rigorous retrospective harmonization possible? Application of the
DataSHaPER approach across 53 large studies. International Journal of Epidemiology, 40(5),
13141328. doi:10.1093/ije/dyr106.
Frayling, T.M., Timpson, N.J., Weedon, M.N., Zeggini, E., Freathy, R.M., Lindgren, C.M.,
Wellcome Trust Case Control Consortium. (2007). A common variant in the FTO gene is asso
ciated with body mass index and predisposes to childhood and adult obesity. Science,
316(5826), 889894. doi:10.1126/science.1141634.
van Gelder, M., Bretveld, R.W., & Roeleveld, N. (2010). Web-based Questionnaires: The future in
epidemiology? American Journal of Epidemiology, 172(11), 12921298. doi:10.1093/aje/
kwq291.
Gottesman, I.I., & Gould, T.D. (2003). The endophenotype concept in psychiatry: Etymology and
strategic intentions. American Journal of Psychiatry, 160(4), 636645.
Gottlieb, G. (1995). Some conceptual deficiencies in developmental behavior genetics. Human
Griffiths, P.E., & Tabery, J.(2008). Behavioral genetics and development: Historical and concep
tual causes of controversy. [Review]. New Ideas in Psychology, 26(3), 332352. doi:10.1016/j.
newideapsych.2007.07.016.
Hamilton, C.M., Strader, L.C., Pratt, J.G., Maiese, D., Hendershot, T., Kwok, R.K., Haines,
J. (2011). The PhenX toolkit: Get the most from your measures. American Journal of
Epidemiology, 174(3), 253260. doi:10.1093/aje/kwr193.
Hamilton, S. P. (2015). The promise of psychiatric pharmacogenomics. [Review]. Biological
Psychiatry, 77(1), 2935. doi:10.1016/j.biopsych.2014.09.009.
Harden, K.P., Hill, J.E., Turkheimer, E., & Emery, R.E. (2008). Geneenvironment correlation
and interaction in peer effects on adolescent alcohol and tobacco use. Behavior Genetics, 38(4),
339347. doi:10.1007/s10519-008-9202-7.
Hein, R., Beckmann, L., & Chang-Claude, J.(2008). Sample size requirements for indirect asso
ciation studies of geneenvironment interactions (G E). Genetic Epidemiology, 32(3), 235
245. doi:10.1002/gepi.20298.
Henderson, N.D. (1990). Why do geneenvironment interactions appear more often in laboratory
animal studies than in human behavioral genetic research? Behavioral and Brain Sciences,
13(1), 136137.
Hicks, B.M., South, S.C., DiRago, A.C., Iacono, W.G., & McGue, M. (2009). Environmental
adversity and increasing genetic risk for externalizing disorders. Archives of General
Psychiatry, 66(6), 640648.
Hunter, D.J. (2005). Geneenvironment interactions in human diseases. [Review]. Nature Reviews
Genetics, 6(4), 287298. doi:10.1038/nrg1578.
Hyde, L.W., Bogdan, R., & Hariri, A.R. (2011). Understanding risk for psychopathology through
imaging geneenvironment interactions. [Review]. Trends in Cognitive Sciences, 15(9), 417
427. doi:10.1016/j.tics.2011.07.001.
Ioannidis, J.P. A. (2005). Why most published research findings are false. PLoS Medicine, 2(8),
696701. doi:10.1371/journal.pmed.0020124.
Jordan, B. D., Relkin, N. R., Ravdin, L. D., Jacobs, A. R., Bennett, A., & Gandy, S. (1997).
Apolipoprotein E epsilon4 associated with chronic traumatic brain injury in boxing [see com
ments]. JAMA, 278(2), 136140.
Kaufman, J., Yang, B. Z., Douglas-Palumberi, H., Grasso, D., Lipschitz, D., Houshyar, S.,
Gelernter, J.(2006). Brain-derived neurotrophic factor-5-HTTLPR gene interactions and envi
ronmental modifiers of depression in children. Biological Psychiatry, 59(8), 673680.
doi:10.1016/j.biopsych.2005.10.026.
Kilpelinen, T.O., Qi, L., Brage, S., Sharp, S.J., Sonestedt, E., Demerath, E., Loos, R.J. F.
(2011). Physical activity attenuates the influence of FTO variants on obesity risk: A meta-
analysis of 218,166 adults and 19,268 children. PLoS Medicine, 8(11), e1001116. doi:10.1371/
journal.pmed.1001116.
Koopmans, J.R., Slutske, W.S., van Baal, G.C., & Boomsma, D.I. (1999). The influence of reli
gion on alcohol use initiation: Evidence for genotype environment interaction. Behavior
Genetics, 29, 445453.
Kottgen, A., Albrecht, E., Teumer, A., Vitart, V., Krumsiek, J., Hundertmark, C., MAGIC
Consortium. (2013). Genome-wide association analyses identify 18 new loci associated with
serum urate concentrations. Nature Genetics, 45(2), 145154. doi:10.1038/ng.2500.
Kutner, K.C., Erlanger, D.M., Tsai, J., Jordan, B., & Relkin, N.R. (2000). Lower cognitive per
formance of older football players possessing apolipoprotein E epsilon 4. Neurosurgery, 47(3),
651657. doi:10.1097/00006123-200009000-00026.
Legrand, L.N., Keyes, M., McGue, M., Iacono, W.G., & Krueger, R.F. (2008). Rural environ
ments reduce the genetic influence on adolescent substance use and rule-breaking behavior.
Psychological Medicine, 38(9), 13411350. doi:10.1017/s0033291707001596.
Manolio, T.A., Chisholm, R.L., Ozenberger, B., Roden, D.M., Williams, M.S., Wilson, R.,
Ginsburg, G. S. (2013). Implementing genomic medicine in the clinic: The future is here.
[Review]. Genetics in Medicine, 15(4), 258267. doi:10.1038/gim.2012.157.
Manuck, S.B., & McCaffery, J.M. (2014). Geneenvironment interaction. [Review; Book Chapter].
Annual Review of Psychology, 65, 4170. doi:10.1146/annurev-psych-010213-115100.
Marchini, J., & Howie, B. (2010). Genotype imputation for genome-wide association studies.
[Review]. Nature Reviews Genetics, 11(7), 499511. doi:10.1038/nrg2796.
Moffitt, T.E., Caspi, A., & Rutter, M. (2005). Strategy for investigating interactions between mea
sured genes and measured environments. Archives of General Psychiatry, 62, 473481.
Moffitt, T.E., Caspi, A., & Rutter, M. (2006). Measured geneenvironment interactions in psycho
pathology concepts, research strategies, and implications for research, intervention, and public
understanding of genetics. [Review]. Perspectives on Psychological Science, 1(1), 527.
doi:10.1111/j.1745-6916.2006.00002.x.
Munaf, M.R., Brown, S.M., & Hariri, A.R. (2008). Serotonin transporter (5-HTTLPR) genotype
and amygdala activation: A meta-analysis. Biological Psychiatry, 63(9), 852857. doi:10.1016/j.
biopsych.2007.08.016.
Munaf, M. R., Zammit, S., & Flint, J. (2014). Practitioner review: A critical perspective on
geneenvironment interaction modelsWhat impact should they have on clinical perceptions
and practice? [Review]. Journal of Child Psychology and Psychiatry, 55(10), 10921101.
doi:10.1111/jcpp.12261.
Murphy, S.E., Norbury, R., Godlewska, B.R., Cowen, P.J., Mannie, Z.M., Harmer, C.J., & Munafo,
M.R. (2013). The effect of the serotonin transporter polymorphism (5-HTTLPR) on amygdala
function: a meta-analysis. Molecular Psychiatry, 18(4), 512520. doi:10.1038/mp.2012.19.
Offit, K. (2011). Personalized medicine: New genomics, old lessons. [Review]. Human Genetics,
130(1), 314. doi:10.1007/s00439-011-1028-3.
Olfson, E., & Bierut, L.J. (2012). Convergence of genome-wide association and candidate gene
studies for alcoholism. Alcoholism: Clinical and Experimental Research, 36(12), 20862094.
doi:10.1111/j.1530-0277.2012.01843.x.
Park, J.H., Wacholder, S., Gail, M.H., Peters, U., Jacobs, K.B., Chanock, S.J., & Chatterjee, N.
(2010). Estimation of effect size distribution from genome-wide association studies and impli
cations for future discoveries. Nature Genetics, 42(7), 570U139. doi:10.1038/ng.610.
Perlis, R.H. (2014). Pharmacogenomic testing and personalized treatment of depression. [Review].
Clinical Chemistry, 60(1), 5359. doi:10.1373/clinchem.2013.204446.
Purcell, S. (2002). Variance components models for geneenvironment interaction in twin analy
sis. Twin Research, 6, 554571.
Ragoussis, J.(2009). Genotyping technologies for genetic research. Annual review of genomics
and human genetics (Vol. 10, pp.117133). Palo Alto, CA: Annual Reviews.
Rietveld, C. A., Medland, S. E., Derringer, J., Yang, J., Esko, T., Martin, N. W., LifeLines
Cohort Study. (2013). GWAS of 126,559 individuals identifies genetic variants associated with
educational attainment. Science, 340(6139), 14671471. doi:10.1126/science.1235488.
Rose, R.J., Dick, D.M., Viken, R.J., & Kaprio, J.(2001). Geneenvironment interaction in pat
terns of adolescent drinking: Regional residency moderates longitudinal influences on alcohol
use. Alcoholism: Clinical and Experimental Research, 25(5), 637643.
Rutter, M. (2014). Commentary: G E in child psychiatry and psychology: a broadening of the
scope of enquiry as prompted by Munafo etal. (2014). [Editorial Material]. Journal of Child
Psychology and Psychiatry, 55(10), 11021104. doi:10.1111/jcpp.12309.
Rutter, M., & Silberg, J.(2002). Geneenvironmental interplay in relation to emotional and behav
ioral disturbance. Annual Review of Psychology, 53, 463490.
Salvatore, J.E., Aliev, F., Edwards, A.C., Evans, D.M., Macleod, J., Hickman, M., Dick, D.M.
(2014). Polygenic scores predict alcohol problems in an independent sample and show modera
tion by the environment. [Meeting Abstract]. Genes, 5, 330346.
Sameroff, A.J., Seifer, R., & Bartko, W.T. (1997). Environmental perspectives on adaptation dur
ing childhood and adolescence. In S.S. Luthar, J.A. Brrack, D.Cicchetti, & J.R. Weisz (Eds.),
Developmental psychopathology: Perspectives on adjustment, risk and disorder (pp.507526).
Cambridge, England: Cambridge University Press.
Shanahan, M. J., & Hofer, S. M. (2005). Social context in geneenvironment interactions:
Retrospect and prospect. Journal of Gerontology, 60B, 6576.
Sheese, B.E., Voelker, P.M., Rothbart, M.K., & Posner, M.I. (2007). Parenting quality interacts
with genetic variation in dopamine receptor D4 to influence temperament in early childhood.
Development and Psychopathology, 19(4), 10391046. doi:10.1017/s0954579407000521.
Siontis, K.C. M., Patsopoulos, N.A., & Ioannidis, J.P. A. (2010). Replication of past candidate
loci for common diseases and phenotypes in 100 genome-wide association studies. European
Journal of Human Genetics, 18(7), 832837. doi:10.1038/ejhg.2010.26.
Smith, P.G., & Day, N.E. (1984). The design of case-control studies: The influence of confound
ing and interactions. International Journal of Epidemiology, 13(3), 356365. doi:10.1093/
ije/13.3.356.
Speliotes, E.K., Willer, C.J., Berndt, S.I., Monda, K.L., Thorleifsson, G., Jackson, A.U.,
MAGIC. (2010). Association analyses of 249,796 individuals reveal 18 new loci associated
with body mass index. Nature Genetics, 42(11), 937948. doi:10.1038/ng.686
Sullivan, P.F. (2007). Spurious genetic associations. Biological Psychiatry, 61(10), 11211126.
doi:10.1016/j.biopsych.2006.11.010.
Sullivan, P. F. (2010). The Psychiatric GWAS Consortium: Big science comes to psychiatry.
[Editorial Material]. Neuron, 68(2), 182186. doi:10.1016/j.neuron.2010.10.003.
Tabery, J.(2008). R.A. Fisher, Lancelot Hogben, and the origin(s) of genotypeenvironment inter
action. Journal of the History of Biology, 41(4), 717761. doi:10.1007/s10739-008-9155-y.
Tabery, J.(2014). Beyond versus: The struggle to understand the interaction of nature and nurture.
Cambridge: MIT Press.
Teasdale, G.M., Nicoll, J.A. R., Murray, G., & Fiddes, M. (1997). Association of apolipoprotein
E polymorphism with outcome after head injury. [Article]. Lancet, 350(9084), 10691071.
doi:10.1016/s0140-6736(97)04318-3.
Thomas, D.C., Lewinger, J.P., Murcray, C.E., & Gauderman, W.J. (2012). Invited Commentary:
GE-Whiz! Ratcheting geneenvironment studies up to the whole genome and the whole expo
some. [Editorial Material]. American Journal of Epidemiology, 175(3), 203207. doi:10.1093/
aje/kwr365.
Thompson, P.M., Stein, J.L., Medland, S.E., Hibar, D.P., Vasquez, A.A., Renteria, M.E.,
Saguenay Youth Study Group. (2014). The ENIGMA Consortium: Large-scale collaborative
analyses of neuroimaging and genetic data. Brain Imaging and Behavior, 8(2), 153182.
doi:10.1007/s11682-013-9269-5.
Thompson, W.D. (1991). Effect modification and the limits of biological inference from epide
miological data. [Editorial Material]. Journal of Clinical Epidemiology, 44(3), 221232.
doi:10.1016/0895-4356(91)90033-6.
Turkheimer, E. (2000). Three laws of behavior genetics and what they mean. Current Directions in
Psychological Science, 9, 160164.
Turkheimer, E., Haley, A., Waldron, M., DOnofrio, B., & Gottesman, I.I. (2003). Socioeconomic
status modifies heritability of IQ in young children. Psychological Science, 14, 623628.
Uher, R., GENDEP Investigators, MARS Investigators, & STAR*D Investigators. (2013).
Common genetic variation and antidepressant efficacy in major depressive disorder: A meta-
analysis of three genome-wide pharmacogenetic studies. American Journal of Psychiatry,
170(2), 207217. doi:10.1176/appi.ajp.2012.12020237.
Visscher, P.M., Hill, W.G., & Wray, N.R. (2008). Heritability in the genomics eraConcepts and
misconceptions. [Review]. Nature Reviews Genetics, 9(4), 255266. doi:10.1038/nrg2322.
Visscher, P.M., Brown, M.A., McCarthy, M.I., & Yang, J.(2012). Five years of GWAS discovery.
[Review]. American Journal of Human Genetics, 90(1), 724. doi:10.1016/j.ajhg.2011.11.029.
Wahlsten, D. (1990). Insensitivity of the analysis of variance to heredityenvironment interaction.
Behavioral and Brain Sciences, 13(1), 109120.
Willer, C.J., Schmidt, E.M., Sengupta, S., Peloso, G.M., Gustafsson, S., Kanoni, S., Global
Lipids Genetics Consortium. (2013). Discovery and refinement of loci associated with lipid
levels. Nature Genetics, 45(11), 12741283. doi:10.1038/ng.2797.
Winham, S.J., & Biernacka, J.M. (2013). Geneenvironment interactions in genome-wide asso
ciation studies: Current approaches and new directions. Journal of Child Psychology and
Wong, M.Y., Day, N.E., Luan, J.A., Chan, K.P., & Wareham, N.J. (2003). The detection of
geneenvironment interaction for continuous traits: Should we deal with measurement error by
bigger studies or better measurement? International Journal of Epidemiology, 32(1), 5157.
doi:10.1093/ije/dyg002.
Wood, A.R., Esko, T., Yang, J., Vedantam, S., Pers, T.H., Gustafsson, S., LifeLines Cohort
Study. (2014). Defining the role of common variation in the genomic and biological architec
ture of adult human height. Nature Genetics, 46(11), 11731186. doi:10.1038/ng.3097.
Zhu, J.W., Loos, R.J. F., Lu, L., Zong, G., Gan, W., Ye, X.W., Lin, X. (2014). Associations of
genetic risk score with obesity and related traits and the modifying effect of physical activity in
a Chinese Han population. PLoS One, 9(3), e91442. doi:10.1371/journal.pone.0091442.
Zondervan, K.T., & Cardon, L.R. (2004). The complex interplay among factors that influence
allelic association. Nature Reviews Genetics, 5, 89100.
Chapter 4
Epigenetics andtheBiology ofGene
Environment Interactions
MichaelJ.Meaney
The modern genomics era initially emphasized studies of genotypephenotype rela-

tions in search of genetic variants associated with disease states. Such studies were
based, in part, on the assumption that variation in genetic sequence would faithfully
translate into altered expression structure or function of cellular proteins, and thus
contribute directly to disease states. The hunt has yielded mixed results. The Psychiatric
Genetics Consortium mobilized meta-analyses of genome-wide association studies
(GWAS) revealing a number of statistically reliable sequence polymorphisms for
schizophrenia, many (e.g. GRIN2B) of which are biologically very plausible (Gratten,
Wray, Keller, & Visscher, 2014). The same analyses for major depression or attention-
deficit hyperactivity disorder (ADHD) yielded few such hits. However, the merits of
the GWASs goes well beyond that of identifying candidate genomic risk variants.
Approaches that are informed by evolutionary theory suggested that genetic variation
would be more closely aligned with that occurring in specific traits that then deter-
mine vulnerability for specific disorders. There is a compelling argument that the
clinical relevance of heritable, sequence-based variations in nucleotide sequences is
best studied in relation to specific endophenotypes (Meyer-Lindenberg & Weinberger,
2006) that associate with disease, rather than directly with diseases states. And of
course genes most directly code for RNAs, not directly for traits, and certainly not
diseases or their symptoms. Most scientists, including those in genetics, appreciated
that the characterization of the human genome and the description of sequence vari-
ants with haplotype mapping would launch a remarkable era in modern biology
whereby genetic effects could be examined at the level of molecular mechanism as
M.J. Meaney (*)

Department of Psychiatry, Ludmer Centre for Neuroinformatics and Mental Health and
Sackler Program for Epigenetics and Psychobiology at McGill University, Douglas
University Mental Health Institute, McGill University,
6875 LaSalle, Montreal, QC, Canada, H4H 1R3
e-mail: Michael.meaney@mcgill.ca

DOI10.1007/978-3-319-49227-8_4
60 M.J. Meaney
opposed to statistical inference based on estimates of heritability. GWAS is a start-

ing point in the hunt for the biological pathways that shape vulnerability for psycho-
pathology. In addition, the broader integration of genetics into the biological and
social sciences has provided a greater breadth to the study of genotypephenotype
relations. The findings from GWAS provide an evidence-based framework for studies
of genotypephenotype variation: how and under what circumstances do risk alleles
influence mental function?
The most relevant conceptual advance for the purpose of this chapter is research
on the variability in genotypephenotype relations that actually provide a deeper
level of appreciation of genetic influences. These approaches examine genetic varia-
tion as moderators of environmental influences, often within the context of the
analysis of gene x environment interdependence. This approach derives from the
realization that environmental signals operate in a cellular context through biological
pathways, which are subject to the influence of function genomic variants (Meaney,
2010). It is actually difficult to imagine how environmental effects on neural devel-
opment and function could emerge independent of functional, genetic polymor-
phisms. The now classic studies of Caspi and colleagues provided a catalyst for a
cavalcade of studies examining the role of candidate genetic variants as moderators
of environmental conditions. These studies focused on functional variants in the pro-
moter region of the SLC6A4 gene that encodes for the serotonin transporter (Lesch
etal., 1996), the MAOA gene that codes for monoamine oxidase or the DRD4 gene
that codes for the dopamine 4 receptor. These variants, and many others, appear to
moderate the influence of clinically relevant environmental events (i.e. conditions
that predict health outcomes) on a wide range of phenotypic outcomes. These gene x
environment studies provide potential explanations for the impressive variability in
health outcomes associated with even severe and profoundly influential forms of
early experience, such as severe childhood maltreatment.
Such studies are not without critique (e.g. Risch etal., 2009; Duncan & Keller,
2011). Insufficient statistical power and replication has commonly been an issue. Of
course these are the same issues that plague all forms of genotypephenotype stud-
ies. Meta-analyses of the more common G x E effects suggest that there is indeed
evidence for genotypic moderation of environmental influences (e.g. Karg et al.,
2011; Uher & McGuffin, 2010). The computational challenges for researchers exam-
ining gene x environment interactions are very real and there are emerging approaches
that permit data reduction strategies that would facilitate the search for gene x envi-
ronment interactions within the context of GWA studies (Thomas, 2010; Wang, Li,
& Bucan, 2007). In this chapter, I explore two additional issues associated with gene
x environment studies. The first issue concerns the issue of biological plausibility:
how might environmental factors interact directly with the genome to produce phe-
notypic variation. What are the biological bases for gene x environment interactions?
In the course of this discussion, the text also considers the newly advanced idea that
specific genetic variants operate as plasticity genes that regulated the sensitivity of
the organism for environmental conditions or context. The argument here is that
studies of epigenetics provide a basis for understanding gene x environment interac-
tions: epigenetics describes the mechanisms by which environmental signals might
directly alter genome structure and function. The concept of plasticity genes is an
4 Epigenetics andtheBiology ofGene Environment Interactions 61
enriching concept that is consistent with the biology of gene x environment interac-
tions. However, emerging evidence suggests that gene x environment interactions,
much like genetic polymorphisms, operate in a tissue-specific manner. Genetically
defined differential susceptibility to environmental conditions is domain specific.
Epigenetics andtheBiological Basis ofG E Interactions
Variation in genotypephenotype relations are an essential feature of development in

any multicellular organism. Each human embodies evidence for this point as normal
human development is a process that creates over 200 different cell types, all from a
common genetic template. Barring the occasional somatic mutation, the hepatocytes
of your liver contain the same nuclear DNA as the neurons of your CNS or the fibro-
blasts of the skin. All show distinct patterns of specialization that associate with
variation in genomic function: the same genome operates differently in hepatocytes
than in neurons. Genes that are commonly expressed in one cell type may be silenced
in another. That portion of the actively transcribed genome, or transcriptome, defines
cell function. In essence, development produces cell-type specific transcriptomes.
And this variation is explained by the epigenetic signals that regulate genomic func-
tion and which differ from cell type to cell type.
The term epigenetics derives from the Greek epi meaning upon and genetics.
Epigenetic signals are chemical modifications to chromatin that regulate genomic
transcription. Figure 4.1 depicts the nucleosome structure of chromatin, which is
comprised of ~146 base pairs of DNA wrapped around a histone octamer (Luger,
Mader, Richmond, Sargent, & Richmond, 1997; Turner, 2001). There is commonly
a closed nucleosome state that is formed by the physiochemical relation between the
histone proteins and its accompanying DNA.This restrictive configuration is main-
tained, in part, by electrostatic bonds between the positively charged histones and the
negatively charged DNA. The closed configuration constrains transcription factor
binding and gene expression. Epigenetic marks are chemical modifications to either
the histone proteins, often at amino acid residues at the tail regions, or to the DNA
itself. These modifications determine chromatin configuration. A closed nucleosome
disfavours transcription factor binding and therefore associates with transcriptional
silencing. An increase in transcription factor binding to DNA and the subsequent
activation of gene expression commonly requires epigenetic modifications to chro-
matin that relax the histoneDNA bonds, open chromatin and favour transcription
factor binding and genomic transcription.
Histone Modifications
Dynamic alteration of chromatin structure is achieved through covalent modifica-

tions to the histone proteins at the amino acids that form the histone protein tails
(Fig. 4.1). There are several examples of such modifications including, but not
62 M.J. Meaney
Fig. 4.1 Crystallographic image (Luger etal., 1997) of the nucleosome showing 146bp wrapped
around a histone complex that is comprised of histone 2A, 2B, 3 and 4 proteins. The tight configu-
ration is maintained, in part, by electrostatic bonds. Modifications, such as acetylation, to the his-
tone regulate transcription factor binding and occur primarily at the histone tails protruding out of
the nuclesome (pictured is the blue tail of histone 3). The amino acids chains that comprise the
histone tails are also major targets for more stable modifications such as methylation
limited to, acetylation, phosphorylation, methylation and ubiquitylation (Maze,

Noh, & Allis, 2013). Specific enzymes catalyse these modifications to regulate the
local chemical properties at specific amino acids (Grunstein, 1997; Hake & Allis,
2006; Jenuwein & Allis, 2001). For example, histone acetyltransferase transfers
an acetyl group onto specific lysines on the histone tails, which diminishes the
positive charge, loosening the relation between the histones and DNA, opening
the chromatin and improving the ability of transcription factors to access DNA
sites. Histone acetylation is associated with active gene transcription. Proteins
known to be associated with transcriptional activation (e.g. transcriptional co-
factors) are commonly identified as histone acetyltransferases. Histone deacety-
lases (HDACs) serve as functional antagonists of the histone acetyltransferases.
HDACs remove acetyl groups and prevent further acetylation, thus maintaining a
closed chromatin structure, decreasing transcription factor and gene expression.
Histone modifications gate transcription factor binding to DNA.
Histone acetylation and deacetylation are dynamic processes that are regulated by
multiple signals, including additional epigenetic modifications. One such modification
is that of histone methylation that also occurs at multiple amino acid sites in the histone
tails and is a more stable modification (Kouzarides, 2007). Histone methylation does
not directly alter chromatin, but rather attracts protein complexes that then mediate
either the closure or opening of chromatin. The various histone methylation marks
often serve to attract histone acetyltransferases, thus favouring transcription, or HDACs,
which produce transcriptional silencing. Histone methylation signals are highly spe-
cific and can serve as platforms for complexes that either open chromatin and thus
favour transcriptional activity or bias towards closed chromatin and transcriptional
silencing. Methylation or demethylation at individual sites is catalysed by specific
enzymes regulated by intracellular signalling pathways that are sensitive to environ-
mental conditions (Kouzarides, 2007). Importantly, because of the more stable nature
of histone methylation, this class of epigenetic modifications can serve to maintain
environmental effects on transcription (Feng & Nestler, 2013; Maze & Nestler, 2011).
DNA Methylation
Epigenetic mechanisms include chemical modifications directly at the level of the

DNA. The classic epigenetic alteration is DNA methylation, which involves the
addition of a methyl group (CH3) onto cytosines predominately bound to guanines
(CpGs) in the DNA (Bird, 1986, 2002; Razin & Riggs, 1980; although see Lister
etal., 2013). DNA methylation in gene regulatory regions is typically, but not uni-
versally, associated with transcriptional repression. The relationship between DNA
methylation and transcription in regions such as the gene body can be more com-
plex (Maunakea etal., 2010; Shenker & Flanagan, 2012) and is commonly positive
(Fig. 4.2). The mechanism by which DNA methylation in regions such as the gene
bodies alters transcription is largely unknown.
Regulatory (non-coding) region Coding region
11 14 15 16 17 18 19 110 111 2 9
GR Promoter 17 Sequence
..tggg16cg gggg17cgggag.
NGFI-A
Fig. 4.2 A representative DNA methylation profile derived from the Illumina HumanMethylation
450K Bead Chip array, which interrogates methylation levels (expressed as beta values) across
multiple individual CpG or CpH sites per gene. This profile is typical of an actively expressed
gene, which shows low levels of DNA methylation in regulatory regions such as those upstream
of the transcriptional start site (TSS), with higher levels in the region of the gene body. A region
surrounding a CTCF/Sp1-binding site is also shown. Such regions commonly show reduced levels
of DNA methylation
64 M.J. Meaney
The repressive effect of DNA methylation on gene transcription occurs through

multiple pathways (Bird, 2002). First, wide swaths of densely methylated DNA com-
pact chromatin and preclude transcription factor binding to DNA sites, thus silencing
transcription. The second manner is subtler and probably more prevalent in regions
with more dynamic variations in gene transcription, such as the brain. In this case,
the presence of the methyl group attracts methylated-DNA-binding proteins (Klose
& Bird, 2006) that, in turn, attract a cluster of proteins that form a repressor complex,
which includes active mediators of gene silencing. The HDACs are the critical com-
ponent of the DNA methylation-associated repressor complexes. Thus, the effect
ohoferf DNA methylation is mediated by HDAC-induced transcriptional repression.
DNA demethylation associates with a return to an open chromatin state that occurs
subsequent to specific histone modifications (e.g. Weaver etal., 2007).
DNA methylation and demethylation are active and dynamic processes.
Dynamic methylation is regulated by DNA-methyltransferases (and see later).
The process of active demethylation is less clear. Candidate mechanisms for DNA
demethylation (Bhutani etal., 2011) include Growth Arrest and DNA Damage 45
alpha and beta (GADD45a/b), as well as oxidation of methylated cytosines to an
alternative methylation form, 5-hydroxymethylcytosine (5hmC), followed by
nucleotide excision (Hackett etal., 2013; Ma etal., 2009). The important point for
consideration is that the enzymes that catalyse DNA methylation as well as those
implicated in demethylation are expression in neurons and glia throughout life.
Thus, CNS cells bear the capacity for remodelling DNA methylation at any phase
of the life cycle. This is likely to be a process that is essential for neuronal func-
tion. The brain shows very limited turnover of cells, and thus variation in neuronal
phenotype, such as that occurring in response to experience, must occur within an
existing population of cells. As such, brain cells are constantly exhibiting varia-
tion in function that reflects differential transcription of the genome.
In addition to its potential role in DNA demethylation, 5hmC is a putative epigen-
etic modification with the potential to bind methylated-DNA-binding proteins and
regulate gene transcription (Branco, Ficz, & Reik, 2012; Melln, Ayata, Dewell,
Kriaucionis, & Heintz, 2012). This epigenetic modification may also be of relevance
for neurodevelopment given its intense enrichment in the brain (Jin, Wu, Li, &
Pfeifer, 2011; Kriaucionis & Heintz, 2009), association with synaptic genes and
dynamic regulation in early development (Ruzov etal., 2011; Szulwach etal., 2011).
Interestingly, 5hmC is particularly enriched in brain relative to other tissues. 5hmC
appears to be mainly euchromatic, suggesting association with actively expressed
genes. Both TET proteins, which catalyse 5hmC, and 5hmC are enriched within gene
bodies, transcription start sites and promoters relative to other regions of the genome.
There is clearly much to be learned about 5hmC, as well as other forms of methylated
DNA, but perhaps the most important point here is that DNA methylation varies both
in its absence and presence at specific sites across the genome, as well as in its
chemical forms.
The reader should be cautioned that the review provided earlier is far from com-
plete. There are multiple histone modifications not discussed, as well as variant forms
of the histone proteins themselves, which are dynamically regulated and functionally
important for transcription. Likewise, there are critical roles for multiple species of
non-coding RNAs, which often work in concert with other epigenetic marks to regu-
late transcription. Moreover, the relationship between specific epigenetic marks and
transcription can vary depending upon context (Meaney & Ferguson-Smith, 2010).
Collectively these modifications alter the structure and chemical properties of the
DNA, and thus gene expression. As such, modifications to the DNA and its chromatin
environment can be considered as an additional layer of information that is contained
within the genome. Furthermore, unlike the underlying DNA sequence, which remains
static across development, epigenetic modifications are dynamically regulated and
responsive to changes in the environment. This is illustrated using the example of
DNA methylation; however, it should be noted that chromatin modifications and his-
tone variants also show dynamic change across development.
Epigenetics andDevelopment
Epigenetics modifications stably alter the structure and chemical properties of the
DNA, and thus gene expression. Essentially, epigenetic modifications produce mul-
tiple functional genomes from a common DNA sequence template. This feature is
critical for understanding variation in genotypephenotype relations, the simplest
example of which is the ability to provide multiple cell types in complex organisms.
Humans, for example, produce over 200 cell types, each of which differs in func-
tion, despite being derived from a common genetic sequence. Thus, over the course
of cell differentiation, each cell type assumes a tissue-specific epigenome that
guides gene expression in a manner that defines the function of that cell type (e.g.
Fan etal., 2005). DNA methylation is considered as a mechanism for the genomic
silencing that underlies the cell specialization. The DNA methylation pattern of a
hepatocyte, for example, is notably different from that of a neuron. DNA methyla-
tion is thus a mechanism for the genomic silencing that underlies the cell specializa-
tion. Interestingly, in the case of neuronal differentiation, DNA methylation appears
to be an end-stage feature of cell specialization that consolidates cell fate (Mohn
etal., 2008). Such events occur early in development and are considered to be highly
stable, such that de-differentiation (whereby a cell loses is specialization) is rare,
and often associated with organ dysfunction and pathology (Hansen etal., 2011;
Irizarry etal., 2009). Indeed a considerable portion of the methylome in any tissue
is largely invariant across individuals and can actually be used as a signature for cell
typing. The tissue-specific nature of DNA methylation signals complicates the study
of epigenomics in human neural function as the relevant tissue is available only in
the context of postmortem studies or with biopsies.
Unlike the underlying DNA sequence, which remains static across develop-
ment, a certain portion of the epigenome is dynamically regulated by environmen-
tal conditions. Thus, epigenetic signals within the same tissues come to differ
between monozygotic twins and are a plausible mechanism for discordance in phe-
notypic outcomes (Kaminsky etal., 2009). Predictably, monozygotic twins appear
66 M.J. Meaney
to become increasingly different in DNA methylation patterns with age. Dynamic

changes in DNA methylation are of considerable interest for medical research,
where more subtle variations in differentiation that occur in later periods of devel-
opment or even in the fully mature brain that underlie inter-individual variation in
cellular function and thus health outcomes. For issues of mental health, the issue is
why neurons in one individual function differently from those of another, or how
neurons might dynamically alter functional properties in relation to experience
(i.e. activity-dependent neuronal plasticity). There is now considerable evidence
from neuroscience and other fields, including immunology and endocrinology, that
the state of DNA methylation at specific genomic sites is indeed dynamic even in
adult animals (Bird, 2007; Jirtle & Skinner, 2007; Meaney & Szyf, 2005) such that
alterations in DNA methylation are candidate mechanisms for experience-driven
plasticity in neural function. Thus, DNA methylation patterns are actively modified
in mature (i.e. fully differentiated) cells, including, and perhaps especially neu-
rons, and that such modifications can occur in animals in response to cellular sig-
nals driven by environmental events (Jirtle & Skinner, 2007; Meaney & Szyf,
2005; Sweatt, 2009). Variations in DNA methylation in multiple cell types are
associated with nutritional states (Cooney, Dave, & Wolff, 2002; Waterland &
Jirtle, 2003; Waterland, Lin, Smith, & Jirtle, 2006; Whitelaw & Whitelaw, 2006),
in antigen-induced maturation of lymphocytes (Bruniquel & Schwartz, 2003;
Murayama etal., 2006) and neurons (e.g. Champagne, 2008; Champagne etal.,
2006; Lubin, Roth, & Sweatt, 2008; Martinowich et al., 2003; Sweatt, 2009).
Similarly, well-studied behavioural adaptations to environmental signals, such as
phenotypic variation in the honey bee, are associated with dynamic change in neu-
ral DNA methylation profiles (Herb et al., 2012). There is also compelling evi-
dence for associations between genome-wide variation in DNA methylation and
clinically relevant environmental conditions such as socio-economic status (Lam
etal., 2012) and childhood maltreatment (Mehta etal., 2013).
Epigenetics andtheBiology ofG E Interactions
Phenotypic variation across all life forms emerges as a function of the interaction
between the genome and environmental conditions. The relation between genotype
and environment is best described in the concept of a norm of reaction, whereby
genotype determines a range of phenotypic outcomes across various environmental
conditions. Genotype may be considered as an influence that determines the range
of environmentally induced, phenotypic outcomes. This concept is critical for our
understanding of the genetic and environmental influences that regulate develop-
ment. It is particularly important that the relative importance of genes and envi-
ronment is not a unitary value and may vary greatly depending on the environmental
context within which the genome operates. Studies of heritability carried out in a
single environment cannot accurately estimate the Norm of Reaction, and often may
not be predictive of phenotypic outcomes in a different environment. Estimates of
the heritability of IQ, for example, vary depending upon socio-economic context,
with low estimates of heritability among those living in impoverished conditions,
and substantially higher estimates amongst those reared in more favourable condi-
tions (Turkheimer, Haley, Waldron, DOnofrio, & Gottesman, 2003). We will later
return to this same point in understanding environmentally induced variation across
the epigenome. Indeed, the greatest strength of heritability estimates for any spe-
cific phenotypic outcome may derive from the study of their variation over time and
context. Such studies clearly reflect geneenvironment interdependency (Sokolowski
& Wahlsten, 2001).
Apart from the dynamic interplay between environmental signals and the genome,
studies of development are replete with examples of the environmental programming
of gene expression. Such studies commonly report that environment conditions in
early life associate with gene expression and biological function in adulthood, and
thus well beyond the duration of the relevant environmental condition. The persis-
tence of the environmental effect is the critical feature. Prenatal nutrient deprivation
or enhanced exposure to hormonal signals associated with stress stably alters predic-
tion patterns of gene expression in the liver and other sites associated with metabo-
lism (Bateson etal., 2004; Gluckman & Hanson, 2004, 2007; Jirtle & Skinner, 2007;
Meaney, Szyf, & Seckl, 2007; Seckl & Holmes, 2007). A major question is that of
precisely how environmental conditions might stably affect gene expression? Studies
from rodent models suggest that such effects are mediated by enduring epigenetic
modifications (an epigenetic memory of environmental conditions during develop-
ment) that subsequently alter gene expression and cellular function
We explored this issue using a rodent model to examine the enduring effects of
variation in maternal care on multiple phenotypic outcomes (Champagne et al.,
2003). There are marked individual differences in the frequency with which lactat-
ing rats lick/grooming (LG) their offspring. Pup LG is a major source of tactile
stimulation for the neonatal rat that regulates endocrine and cardiovascular function
in the pup (Hofer, 1984; Levine, 1994; Schanberg, Evoniuk, & Kuhn, 1984). The
tactile stimulation derived from pup LG increases levels of growth hormone and
decreases that of adrenal glucocorticoids. These effects promote somatic growth.
The individual differences in maternal LG are stable across time and even across
multiple litters, providing an opportunity to associate variations in a specific form
of maternal care with developmental outcomes.
The resulting studies reveal pronounced effects of maternal care on behavioural
and endocrine responses to stress in the offspring, as well as a range of associated
phenotypes, including defensive responses to threat, female reproductive behaviours
and metabolic function. The male or female adult offspring of High LG mothers
show more modest behavioural and endocrine responses to stress compared to ani-
mals reared by Low LG mothers (Liu etal., 1997; Caldji, Sharma, Francis, Plotsky,
& Meaney, 1998; Francis, Diorio, Liu, & Meaney, 1999; Weaver etal., 2004; Menard,
Champagne, & Meaney, 2004; Toki, Imanaka, Yamamoto, Yamawaki, & Honma,
2007). Specifically, the offspring of High LG mothers show reduced fearfulness and
more modest hypothalamic-pituitary-adrenal (HPA) responses to acute stress. Cross-
fostering studies, where pups born to High LG mothers are fostered at birth to Low
68 M.J. Meaney
LG mothers (and vice versa), or neonatal handling studies, which significantly

increase maternal LG, reveal a direct relationship between maternal care and the
postnatal development of individual differences in behavioural and HPA responses to
stress (Francis et al., 1999; Weaver et al., 2004; Caldji et al., 2000, 2003). In the
cross-fostering studies, the rearing mother determined the phenotype of the off-
spring. Handling-induced increases in maternal LG reverse the phenotypes associ-
ated with decreased maternal licking (Francis etal., 1999). Thus, variations within a
normal range of parental care can dramatically alter phenotypic development in the
rat. It is nevertheless interesting to note that some of the phenotypic differences asso-
ciated with variations in pup LG are only partly reversed with postnatal cross-foster-
ing or neonatal handling. These findings suggest possible in utero influences that
have been confirmed at least with respect to the development of individual differ-
ences in hypothalamic-pituitary-ovarian function and sexual behaviour in the female
offspring (Borrow, Levy, Soehngen, & Cameron, 2013; Cameron etal., 2008) and
metabolic function (Kappeler, Silveria, & Meaney, unpublished).
The differences in the HPA response to stress in the offspring of High and Low
LG mothers are mediated by a maternal effect on the expression of the glucocorticoid
receptor gene (Nr3c1) in the hippocampus (Liu et al., 1997; Weaver et al., 2004,
2007). The glucocorticoid receptor (GR) is a ligand-gated nuclear receptor, which
when bound by glucocorticoids is activated and translocated to the cell nucleus
where it functions as a transcription factor that regulates gene expression. GR activa-
tion in the hippocampus associates with the activation of a negative feedback signal
that regulates corticotrophin-releasing factor (CRF) expression in the hypothalamus.
Since CRF serves to activate the pituitary-adrenal stress response, negative feedback
regulation serves to moderate the magnitude of the stress response. The offspring of
High, compared with Low LG mothers, show increased hippocampal GR expression,
more efficient negative feedback regulation of CRF, reduced hypothalamic CRF
expression and more modest HPA responses to stress. Moreover, variations in the
frequency of pup LG towards individual pups of the same mother are significantly
correlated with hippocampal GR expression in adulthood (van Hasselt etal., 2012).
Finally, studies directly examining the effects of the tactile stimulation associated
with pup LG show that animals brushed for 15min/day exhibit increased hippo-
campal GR expression (Hellstrom, Dhir, Diorio, & Meaney, 2012). These findings,
as well as those from studies that directly manipulate the frequency of pup LG by the
dam reveal a direct relation between maternal care and the phenotypic development
of the offspring.
The Nr3c1 gene includes a variable exon 1 region which contains multiple pro-
moter sequences, each of which serves to activate GR expression (Fig. 4.3a;
McCormick etal., 2000). Later studies revealed that the enduring effects of mater-
nal care associate with altered levels of DNA methylation of a specific promoter
sequence, the exon 17 promoter, and increased exon 17-induced GR expression
(Hellstrom etal., 2012; Weaver etal., 2004, 2007). The difference in DNA methyla-
tion at the exon 17 promoter occurred in a region that binds the transcription factor,
nerve-growth factor-induced factor A (NGFI-A). The binding of NGFI-A to the
exon 17 promoter activates GR expression through the exon 17 promoter (Weaver
a
Regulatory (non-coding) region Coding region
11 14 15 16 17 18 19 110 111 2 9
GR Promoter 17 Sequence
..tggg16cg gggg17cgggag.
NGFI-A
High LG Low LG
5-HT
Hippocampus 5-HT7 rec Hippocampus
cAMP (-)
(-) (-) (-)
PKA
Hypothalamus Hypothalamus
CBP
CRF NGFI-A CRF
Pituitary Pituitary
GR gene
ACTH
ACTH
Adrenals
Glucocorticoids
Adrenals
Glucocorticoids
Fig. 4.3(a) A schema describing the organization of the rat glucocorticoid receptor gene includ-
ing nine exon regions. Exons 29 participate in the coding of the glucocorticoid receptor protein.
Exon 1 is comprised of multiple regions each of which is capable of activating gene transcription
(i.e. promoter sequences). The various exon 1 promoters actions are tissue specific, with evidence
suggesting that certain promoters are more active in areas such as liver or thymus, and others more
active in brain (e.g. exon 17; based on McCormick etal., 2000 and see Turner & Muller for com-
parable data in humans). The consensus binding site for NGFI-A lying within the exon 17 promoter
is highlighted. (b) Model depicting the effect of maternal care on hippocampal GR expression
(centre panel), highlighting the activation of 5-HT-induced increases in NGFI-A/CBP complexes
that bind the exon 17 GR promoter. The left and right hand panels show the influence of GR
expression on HPA responses to stress, with the relative increase in hippocampal GR in the adult
offspring of High LG mothers associated with greater feedback inhibition of hypothalamic CRF
and more modest responses to stress
70 M.J. Meaney
etal., 2007). Thus, in the adult offspring of High LG mothers, there are reduced
levels of level of DNA methylation at the NGFI-A-binding site and increased GR
expression.
The differences in DNA methylation at the exon 17 promoter site are regulated by
a cascade of events that are induced by maternal LG (Fig. 4.3b; Hellstrom et al.,
2012; Meaney et al., 2000; Weaver et al., 2007). Briefly, the tactile stimulation
derived by LG (1) increases thyroid hormone signalling and serotonin activity in the
hippocampus; (2) serotonin increases protein kinase A/cAMP signalling through a
5-HT7 receptor, which (3) then activates the expression of NGFI-A, as well as the
CREB-binding protein (CBP) and specific protein 1 (Sp1) in hippocampal neurons
and the binding of the transcriptional signals to the exon 17 sequence. The binding of
NGFI-A, CBP or Sp1 to the exon 17 promoter sequence is actively regulated by
motherpup interactions, such that there is increased NGFI-A bound to the exon 17
promoter immediately following a nursing bout, but not at a period that follows
25min without motherpup contact (Hellstrom etal., 2012). Likewise, brushing
pups also increased NGFI-A expression and NGFI-A binding to the exon 17 GR gene
promoter (Hellstrom etal., 2012). Moreover, the binding of NGFI-A is obligatory for
the alteration in DNA methylation (Weaver etal., 2007) and invitro over-expression
of NGFI-A leads to a demethylation of the exon 17 promoter. These findings reveal
that environmentally induced activation of intracellular signalling pathways can
associate with active remodelling of methylation states. Comparable findings emerge
from studies of environmental enrichment. Peripubertal enrichment associates with
altered levels of DNA methylation at multiple sites across the genome and is reliant
upon an enrichment-induced increase in CREB (Guo etal., 2011), a transcription
factor that is highly expressed in brain and linked to synaptic plasticity.
Hippocampal NGFI-A levels can initiate the remodelling of DNA methylation.
The NGFI-A transcription factor binds to multiple sites across the genome, suggesting
that other NGFI-A-sensitive regions should show a maternal effect on DNA methyla-
tion and gene expression. Zhang and colleagues showed that the hippocampal expres-
sion of the GAD1 gene that encodes for glutamic acid decarboxylase, an enzyme in
the production of the neurotransmitter GABA is increased in the adult offspring of
High LG mothers (Zhang, Hellstrom, et al., 2010). This effect is associated with
altered DNA methylation of an NGFI-A response element in a manner comparable
to that for the glucocorticoid receptor gene. Moreover, as with the effect on the GR,
an invitro increase in NGFI-A expression mimics the effects of increased pup LG.
As discussed earlier, 5-hydroxymethocytosine is an alternative form of DNA
methylation that, while less prominent that 5-methyl-cytosine, is highly enriched in
brain (Kriaucionis & Heintz, 2009; Tahiliani etal., 2009; Ito etal., 2010; Guo etal.,
2011). The ten-eleven translocation (TET) family of enzymes converts 5-methylcy-
tosine to 5-hydroxymethylcytosine. We (Zhang, Labont, Wen, Turecki, & Meaney,
2013) used methylation-dependent immunoprecipitation to show that the difference
in DNA methylation at the exon 17 GR promoter between the offspring of High and
Low LG mothers exists in the form of 5-hydroxymethylcytosine.
The link between DNA methylation and gene transcription is, in part, determined
by the capacity of DNA methylation to regulate histone modifications, especially
histone acetylation. As discussed earlier, histone acetylation opens chromatin and

increases the access of transcription factors to their DNA-binding sites. Acetylation
commonly occurs at lysine residues, such as the H3K9, and is catalysed by histone
acetyltransferases and reversed by histone deacetylases (HDACs). Cytosine meth-
ylation attracts repressor complexes comprised of HDACs, which remove acetyl
groups from histone tails and prevent subsequent acetylation (Shahbazian &
Grunstein, 2007), such that DNA methylation and histone acetylation are inversely
related. Consistent with the differences in DNA methylation, there is increased
H3K9ac of the exon 17 GR promoter in hippocampus from the adult offspring of
High compared with Low LG mothers (Hellstrom etal., 2012; Weaver etal., 2004,
2007; Zhang etal., 2013). This pattern is similar to maternal effects on hippocampal
GAD1 or Grm1 expression; in each case decreased DNA methylation within
promoter regions associates with increases in both H3K9ac and gene transcription.
Additional histone modifications, notably histone methylation, influence transcrip-
tion through indirect pathways and are catalysed by distinct histone methyltransfer-
ases and reversed by histone demethylases (Berger, 2007; Bernstein et al., 2005;
Kouzarides, 2007; Ruthenburg, Allis, & Wysocka, 2007; Taverna etal., 2007). This
process provides a signalling pathway that begins with the activation of the intracel-
lular signals that activate the individual methylating or demethylating enzymes pro-
ducing a specific epigenetic profile on the histone tails. And histone modifications
co-vary. One example is that of H3K9ac and H3K4me, which are marks generally
present at actively transcribed genomic regions ( Kouzarides, 2007; Ruthenburg
etal., 2007). We find increased H3K9ac and H3K4me3 at both regions of the exon
17 GR promoter and the levels of these individual marks are very highly correlated
(Zhang etal., 2013). H3K4me appears to protect CpG islands against methylation
(Ooi etal., 2007; Thomson etal., 2010) that sustains a negative correlation between
H3K4me and CpG methylation. The exon 17 GR promoter in the hippocampal sam-
ples from the offspring of High compared to Low LG mothers shows decreased
level of DNA methylation associated and increased level of H3K4m3 (Zhang etal.,
2013). H3K4me3 targets the NURF chromatin remodelling factor and the Yng1
protein complex to genes thus increasing the level of histone acetylation and tran-
scriptional activation, and explaining the tight correlation between the levels of
H3K4m3 and H3K9ac. These DNA methylation-dependent histone modifications
are critical for the effects of maternal care on GR expression and HPA responses to
stress. The effect of CpG methylation on gene expression is, in part, mediated by the
recruitment of HDAC-containing repressor complexes. HDAC inhibits histone acet-
ylation and transcription factor binding and liberates the expression of genes from
methylation-induced repression. HDAC inhibition also reverses the maternal effects
on hippocampal GR expression (Weaver etal., 2004). Chronic, central infusion of
adult offspring of Low LG mothers with the broad spectrum HDAC inhibitor,
trichostatin A (TSA; Weaver etal., 2004; Weaver, Meaney, & Szyf, 2006), signifi-
cantly increased H3K9 acetylation, NGFI-A binding to the GR-17 promoter and GR
expression to levels comparable to those observed in the offspring of High LG
mothers. TSA infusion also eliminated the effect of maternal care on HPA responses
to acute stress. These results suggest a direct relation between maternal care, histone
72 M.J. Meaney
acetylation, DNA methylation of the GR-17 promoter, GR expression and HPA

responses to stress.
These studies reflect the capacity for dynamic variation in DNA methylation
states in adult brain. Importantly, TSA treatment not only increases GR expression,
and thus alters HPA responses to stress, but it also associates with a demethylation
of the exon 17 GR promoter (Weaver etal., 2004). TSA treatment does not directly
later DNA methylation. Rather, it prevents the DNA methylation-induced repres-
sion of transcription through HDAC occupancy. The subsequent increase in tran-
scriptional activity drives an alteration in DNA methylation. These findings suggest
that DNA methylation states that are established during development are potentially
reversible. This hypothesis remains largely untested and is a serious gap in our
understanding of the role of epigenetic mechanisms in mediating the influence of
environmental conditions on brain function. The implications for intervention pro-
grammes are obvious. Indeed, there are studies with human populations (described
later) that reflect the continued dynamic state of DNA methylation. Relevant animal
models of phenotypic variation associated with environmental experience in adult-
hood, such as environmental enrichment, are ideal for studies on the mechanisms of
epigenetic plasticity over the lifespan.
Variations in parentoffspring interactions epigenetically programme hippo-
campal GR expression and thus HPA response to stress. Subsequent studies reveal
effects of early experience on multiple components of the HPA axis. In each case,
there is evidence for stable epigenetic programming. Prolonged periods of maternal
separation in the mouse alter the methylation state of the promoter for the arginine
vasopressin gene (AVP), increasing hypothalamic AVP synthesis and HPA responses
to stress (Murgatroyd etal., 2009). This epigenetic programming of AVP expression
in the paraventricular nucleus involves Ca(++)/calmodulin kinase-mediated
phosphorylation of the methyl-CpG-binding domain protein MeCP2 leading to dis-
sociation from its DNA-binding site and derepression of AVP gene transcription.
The reduced occupancy of MeCP2 during this early stage of life facilitates the
development of hypomethylation at the AVP enhancer, which sustains the dere-
pressed state of the AVP gene. Thus, as described earlier, environmental conditions
alter intracellular signalling pathways that then produce conditions for the remodel-
ling of otherwise chemically stable epigenetic marks.
Environmental conditions that increased the frequency of pup LG in the rat are
associated with decreased paraventricular CRF expression (Liu etal., 1997; Plotsky
et al., 2005). Baram and colleagues (Avishai-Eliner, Eghbal-Ahmadi, Bar-El, &
Baram, 2001) showed that this maternally regulated decrease in CRF expression is
associated with active repression of the CRF through increased NRSF binding to a
21bp sequence within the regulatory region of the Crh gene. In addition to active
crh repression, NRSF might also initiate chromatin modification. Korosi et al.
(2010) then showed that the number of excitatory synapses and the frequency of
miniature excitatory synaptic currents onto CRF neurons, as were the levels of the
glutamate vesicular transporter vGlut2, were reduced as a function of augmented
maternal care. A study using a procedure that disrupts the quality of maternal care
in the mouse and enhances CRF expression (Rice, Sandman, Lenjavi, & Baram,
2008) showed enhanced glutamatergic transmission to hypothalamic CRF neurons

in the offspring (Korosi etal., 2010). Since neuronal activity can influence NRSF
expression, the stable alteration in excitatory input together with later GR-mediated
negative feedback inhibition from the hippocampus might sustain the effect of
maternal care on hypothalamic CRF expression and HPA responses to stress.
Maternal regulation of HPA function also includes effects at the level of the pitu-
itary (Wu, Daniel, Almeida, & Spengler, 2014). Maternal separation of neonatal
mice produces an enduring hypomethylation of the POMC gene, which encodes for
proopiomelanocortin from which ACTH is derived, increased POMC mRNA
expression and increased basal and CRF-induced levels of ACTH. Subsequent
invitro studies showed that methylation of this region of the promoter produced a
MeCP2-mediated repression of POMC expression. These findings reveal that the
quality of postnatal maternal care epigenetically programmes gene expression at
multiple levels of the HPA axis to regulate both basal and stress-induced activity.
pigenetic Regulation ofGR Expression andFunction

E
inHumans
We translated the findings from the rodent epigenetics programme using human
hippocampus obtained from the Qubec Suicide Brain Bank (www.douglas.qc.ca/
suicide). Childhood adversity, including childhood sexual and physical abuse, as
well as parental neglect is common among suicide victims. We (McGowan etal.,
2009; Labont etal., 2013) found decreased hippocampal GR expression in samples
from suicide completers with histories of childhood maltreatment compared with
controls (sudden, involuntary fatalities) with data obtained from validated forensic
interviews. There were no significant correlations between psychopathology, nota-
bly depression and substance disorders, and hippocampal GR expression. Rather
decreased hippocampal GR expression associated with childhood maltreatment;
there were no differences in hippocampal GR expression in samples from suicides
negative for a history of childhood maltreatment.
Splice variant analysis revealed decreased expression of non-coding exons 1B, 1C,
1F and 1H in suicides with a history of childhood maltreatment compared with both
controls and suicides without a history of maltreatment that correlated with differen-
tial DNA methylation patterns between groups in the corresponding exon 1 variant
promoters. The exon 1F sequence is of particular interest as it is the orthologue of the
rat exon 17, is highly expressed in brain and contains an NGFI-A-binding site (Turner
& Muller, 2005; McGowan etal., 2009). The exon 1F sequence shows increased DNA
methylation and decreased NGFI-A binding in samples from suicide victims with a
history of maltreatment. Decreased expression levels of GR exon 1B, 1C, and 1H tran-
scripts were also associated with alterations in methylation of the respective sequences,
with particular sites significantly correlated with expression levels (LaBont et al.,
2013). Interestingly, differential methylation was associated with putative transcrip-
tion factor-binding sites, implicating intracellular signalling pathways.
74 M.J. Meaney
Inter-individual differences in DNA methylation can be tissue specific and cell

type specific. Nevertheless, there are now a number of reports of associations between
the quality of childhood experience and the methylation status of the exon 1F NR3C1
gene promoter in peripheral cells. Perroud etal. (2011, 2014) used peripheral blood
lymphocytes to show that childhood maltreatment associates with an increased level
of exon 1F methylation and, importantly, that the methylation status of the promoter
was closely correlated with both the frequency and severity of maltreatment (also see
Romens, McDonald, Svaren, & Pollak, 2015). The offspring of mothers exposed to
intimate partner violence also show an increased level of methylation of the exon 1F
NR3C1 gene promoter in blood cells (Radtke etal., 2011). Interestingly, Tyrka, Price,
Marsit, Walters, and Carpenter (2012) reported that the increased methylation of the
exon 1F NR3C1 gene promoter in leukocytes that associated with disruption of normal
parentoffspring interactions or maltreatment was linked to an attenuated cortisol
responses to the Dex/CRH test. In this study, and others (e.g. Melas etal., 2013) child-
hood parental loss was associated with increased methylation of the exon 1F NR3C1
gene promoter (note Melas etal. used salivary DNA, which is primarily of leukocyte
origin). Taken together these studies provide support for the association between the
quality of childhood experience and methylation of the NR3C1 gene. DNA methyla-
tion patterns are highly tissue specific. Indeed, variation in methylation patterns forms
a basis for stable, cell differentiation. Nevertheless, the association between childhood
adversity and the methylation state of the exon 1F GR promoter has consistently been
observed in tissue of non-neural origin (Turecki & Meaney, 2016).
The ability to detect the epigenetic signatures of early social adversity in samples
readily accessible from living human subjects creates opportunities for translational
research. Nanni et al. (2012) identified childhood maltreatment as a significant
moderator of the treatment response to antidepressant medications. While these
findings suggest that a developmental history of childhood adversity might inform
clinical treatment, self-reports of adversity, especially abuse or extreme neglect, in
early childhood are often unreliable. However, an epigenetic mark that reflects the
impact of the adversity on the individual could be of value in predicting clinical
outcomes. We (Yehuda etal., 2013; Yehuda, Daskalakis, etal., 2014; Yehuda, Flory,
etal., 2014) provided preliminary evidence for an association between the methyla-
tion state of the exon 17 GR promoter and treatment (psychotherapy) outcome with
PTSD patients. We found that methylation of the exon 17 GR promoter was corre-
lated with HPA function and, although methylation did not vary as a function of
treatment, this measure did significantly correlate with treatment outcome. These
are very preliminary studies, but suggest that epigenetic marks that reliably reflect
developmental history might be of value as predictors of treatment responsivity.
Environmental Regulation oftheHuman Methylome
The emergence of DNA methylation arrays (Dedeurwaerder et al., 2011) permits

analyses of the impact of environmental conditions on DNA methylation across the
genome. The term genome-wide should not imply complete coverage across the
genome, which remains the domain of NextGen sequencing, but rather a sampling of
candidate methylation sites, normally CpG dinucleotides, that results in about 23%
coverage. Earlier analyses of samples from adolescents enrolled in the English and
Romanian Adoptees study (Nelson etal., 2007; Rutter, Kumsta, Schlotz, & Sonuga-
Barke, 2012) and the Bucharest Early Intervention Project (BEIP) suggest a linger-
ing epigenetic signature associated with the deprivation of parental care in early
childhood (Naumova et al., 2012). The results suggested that ~.5% of CpG sites
sampled on this array show differential methylation, with evidence for a shift towards
hypermethylation of target genes associated with immune response and cellular
signalling in the institutionalized group. Additional genome-wide analyses of DNA
methylation reveal the persisting effect of early adversity on the methylome (Essex
et al., 2013; Labonte et al., 2012; Mehta et al., 2013). Maternal stress, especially
early in infancy, is predictive of DNA methylation profiles assessed in buccal epithe-
lial cells at 15years of age (Essex etal., 2013). Binder and colleagues show that
childhood adversity associates with coordinated epigenetic and transcriptional
changes in peripheral blood cells from adults with PTSD (Mehta et al., 2013).
Interestingly, despite a similar clinical presentation for both groups (i.e. PTSD) early
life adversity was associated with an almost unique transcriptional profile, relative to
PTSD patients without early life exposure, suggesting differential regulation of gene
transcription in as function of childhood adversity.
he Epigenome asaFunction ofGene Environment

T
Interactions: Allele-Specific Methylation
Epigenetic modifications modify the relation between genetic polymorphism and

gene transcription. However, epigenetic states and DNA sequence are interdependent
processes. Genotype is an important determinant of the epigenetic landscape (Deaton
& Bird, 2011; Lienert etal., 2011). Variation in DNA methylation also reflects genetic
relatedness (Gertz etal., 2011) and examples of allele-specific methylation, where the
methylation state at a particular genomic region is associated with sequence variation.
There are widespread effects of genotype on DNA methylation in the human brain
(Zhang, Cheng, etal., 2010) with some evidence that genotype-methylation associa-
tions are conserved across peripheral and neural tissues (Docherty etal., 2012; Klengel
etal., 2013), permitting the study of such associations in peripheral tissue.
There are multiple ways in which DNA sequence variation can influence DNA
methylation. There are disrupting influences whereby genetic polymorphisms
such as single nucleotide polymorphisms (SNPs), variable number tandem repeats
and copy number variants may introduce or remove CpGs thereby altering potential
sites for methylation. Genotypic variation may also act in trans to influence that
activity cell signalling pathways that trigger changes in DNA methylation. DNA
methylation is influenced by transcription factor binding and genetic polymor-
phisms that modulate transcription factor binding may thus influence DNA meth-
ylation events. Increased transcription factor binding can protect CpG islands from
DNA methylation and vice versa (Stadler etal., 2011; Lienert etal., 2011; Macleod
76 M.J. Meaney
etal., 1994). Sequence variants may act in cis to alter transcription factor binding,
thus influencing the methylation patterns at neighbouring regions. Both gene pro-
moter and more distal regulatory regions appear to be influenced by transcription
factor binding (Schubeler, 2012). These processes underlie the association between
genotype and DNA methylation.
We (Teh etal., 2014) used DNA obtained from umbilical chords obtained in a
longitudinal, birth cohort study and surveyed the genome for sequence polymor-
phisms and DNA methylation across the genome using the HumanMethylation
450K Bead Chip (450K array). Genomic DNA was obtained from umbilical chord
samples for which there was both full genotyping of the mother and child, as well
as extensive measures of the maternal environment and phenotype. We first estab-
lished those regions of the genome that showed marked inter-individual variation in
DNA methylation. Such variably methylated regions (VMRs) were defined at 1kb
regions with >2 CpGs showing more that 15% inter-individual variation. There
were 1423 such VMRs and each was regressed against all the SNPs covered in the
genotyping array to identify the strongest SNPVMR association. We used these
SNPVMR associations as a proxy for a genetic effect on methylation. Multiple
regression models were used to examine whether inter-individual variation for each
of the VMRs was best explained by the genetic factor (the SNPVMR association),
an environmental factor (e.g. maternal depression, maternal BMI, parity, gestational
age, birth weight, etc.) or an interaction between gene and environment. Twenty-
five percent of the VMRs were best explained by a genetic effect, 75% were best
explained by a gene x environment interaction model and none were best explained
by an environmental factor alone (note the use of the term best, which implies that
the strongest model, but not that that model accounted for all of the variance). When
this multi-ethnic sample was reduced to include only one ethnic group, the percent-
age of VMRs best explained by genetic variation fell to 16%. The most important
point is the degree to which variation across the human methylome is determined by
gene x environment interactions.
These processes are apparent in studies of candidate genomic regions. Philibert
and colleagues used lymphocytes to show that the methylation of the SLC6A4
promoter region, so commonly studied in G x E studies, is determined by an inter-
action between childhood adversity and the SLC6A4 genotype. Childhood adver-
sity associates with hypermethylation of CpGs in proximity to the transcriptional
start site (Beach et al., 2010, 2011), with some evidence of an association with
SLC6A4 transcription (Vijayendran, Beach, Plume, Brody, & Philibert, 2012).
Hariri and colleagues (Nikolova etal., 2014) found that methylation of the SLC6A4
proximal promoter region associates with threat-related amygdala reactivity in
functional magnetic resonance imaging (fMRI) studies, with hypermethylation
predicting greater amygdala reactivity, an effect that parallels that observed for
childhood adversity. Moreover, methylation of the CpG site most strongly corre-
lated with amygdala reactivity was also correlated to serotonin transporter mRNA
levels in postmortem amygdala tissue from an independent cohort. These findings
are consistent with the idea that the methylation status of the SLC6A4 promoter
emerges as a function of gene x environment interactions and is associated with
5-HTT expression and neural function. Likewise, stress-related alterations to

methylation levels of the COMT gene associate with prefrontal function (Ursini
etal., 2011).
These findings beg the question of the mechanisms underlying gene x environ-
ment effects on DNA methylation. The most sophisticated model for such effects
derives from the studies of Binder and colleagues on the FKBP5 gene (Klengel &
Binder, 2015; Klengel etal., 2013). FKBP5 encodes FKBP5, which is a glucocorti-
coid receptor (GR) chaperone protein that is induced by increased levels of GR acti-
vation and binding following stress. FKBP5 then binds to the GR complex, reduces
the affinity of GR for glucocorticoids, thus limiting GR activation and translocation
of GR to the nucleus, thus inhibiting GR signalling at the genomic level. A functional
sequence variant in FKBP5 moderates the risk for mood disorders associated with
childhood adversity. This polymorphism alters GR binding to FKBP5 and the capac-
ity for GR-induced increases in FKBP5 expression. The allele associated with stron-
ger FKBP5 induction produces GR resistance, with presumable GR feedback deficits,
prolonged cortisol response following stress and increased risk for psychopathology.
Exposure to childhood adversity leads to allele-specific epigenetic changes with a
decrease in DNA methylation in the region of the FKBP5 gene that binds GR, but
only in carriers of the risk allele. This demethylation enhances GR binding and
FKBP5 induction following GR exposure (Klengel etal., 2013). Interestingly, there
is evidence that increased GR binding directly alters methylation of the FKBP5 gene.
GR activation with a selective agonist in a neuronal progenitor cell line leads to a
demethylation in exactly the same CpG dinucleotides that are hypomethylated in
DNA from peripheral blood in trauma-exposed risk allele carriers. These CpGs are
located either within or between GR-binding sites. In sum, the difference in methyla-
tion occurs because of an interaction between adversity-related increases in GR sig-
nalling at the level of the FKBP5 gene and the allelic variation that moderates
GR-binding capacity to this same region.
Gene x Environment Effects: Plasticity Genes?
The initial interpretation of the G x E data suggested that certain alleles conferred
resilience, such that the effects of endangering environmental conditions were largely
limited to those with risk or susceptibility variants. The Caspi studies (Caspi,
Hariri, Holmes, Uher, & Moffitt, 2010) with the serotonin-transporter-linked poly-
morphic region (5-HTTLPR) of the SLC6A4 gene, those victims of childhood mal-
treatment that were homozygous for the longer (l) version of the promoter, were
statistically at no greater risk for depression in adulthood than were individuals not
exposed to maltreatment. Similar studies showed a comparable effect of genotype.
The two functional alleles, long (L) and short (S), result from a 43bp insertion/dele-
tion in the promoter region of 5-HTT.The S, as opposed to the L allele, associates
with reduced invitro basal transcription of 5-HTT mRNA suggesting alterations in
5-HT signalling.
78 M.J. Meaney
The impact of severe childhood maltreatment on mental health is moderated by

variants in genes that encode for RNA or protein products involved in the regulation
of stress responses (Binder etal., 2008; Bradley & Corwyn, 2008; Caspi etal., 2010;
Heim & Nemeroff, 2009; Xie etal., 2010) and associate with sensitivity to adversity.
However, many of these variants also associate with enhanced susceptibility to more
favourable environmental conditions. Thus, both the serotonin-transporter-linked
polymorphic region (5-HTTLPR) of the SLC6A4 gene and the Val66Met polymor-
phism of the BDNF gene associate with an increased risk for affective disorders under
adverse conditions, but also with a decreased risk under more propitious setting
(Casey etal., 2009; Chen, Li, & McGue, 2013; Devlin, Brain, Austin, & Oberlander,
2010; Gatt et al., 2009; Hilt, Sander, Nolen-Hoeksema, & Simen, 2007; Pluess &
Belsky, 2011; Pluess et al., 2011; Suzuki et al., 2012; Taylor et al., 2010; Uher &
McGuffin, 2008; Wagner, Baskaya, Dahmen, Lieb, & Tadic, 2010). These findings led
to the consideration of plasticity genes that associate with differential susceptibility
to environmental context, rather than with risk, per se (Belsky, Bakermans-Kranenburg,
& Van IJzendoorn, 2007; Belsky et al., 2009; Boyce & Ellis, 2005; Ellis, Boyce,
Belsky, Bakermans-Kranenburg, & van IJzendoorn, 2011). This conceptualization of
genetic influences would explain previously identified examples of individual differ-
ences in biological sensitivity to context; differences nicely captured in the depiction
of orchid and dandelion children. Orchids flourish under appropriate conditions,
but perish in harsh circumstance. Dandelions, as well known to anyone with land-
scaped properties, are blithely unaffected by circumstance.
These findings have the potential for considerable impact on public policy.
Indeed, one implication is that we consider psychosocial interventions in the same
manner as we might for any treatment in our attempts to use genotyping to formu-
late effective, personalized medicine. This, in my mind, is a laudable objective, and
genetic information will at some level need to be integrated in a manner in which we
target individuals to treatments. Advances in many other fields of clinical medicine
rely on either direct (e.g. genetic sequence) or assumed (e.g. family history) mea-
sures of heritable genetic variation to more effectively treat patients. There is, how-
ever, a critical difference in the approach used in cancer and heart disease from that
advocated in mental health on the basis of plasticity genes. Studies of tumours or
coronary heart disease focus on specific outcomes. The concept of plasticity or
susceptibility genes is blithely free of such constraint, preferring to consider the
effects of genetic variants as determining universal susceptibility or resistance to
environmental conditions regardless of the outcome (or the environment for that
matter): hence the term plasticity gene.
A major limitation of G x E studies is that they examine only a specific outcome
to a certain environmental condition; there is little within any existing datasets to
suggest that what appears to the apparently resistance variant might influence
other phenotypic outcomes. What are needed are studies that simultaneously mea-
sure (and report) multiple phenotypic outcomes as a function of a genomic variant
in interaction with a specific environmental condition. Neuroimaging studies pro-
vide an ideal opportunity for such analyses since measures of structure, organiza-
tion and activity of multiple brain structures are captured at a single time point.
Recent examples of such studies question the concept of universally operative plas-
ticity genes (Li etal., 2015; Qiu etal., 2013).
Li etal. (2015) pursued the influence of gene x environment interaction effects
across the human methylome focusing on the putative plasticity gene, BDNF, which
encodes for brain-derived neurotropic factor. There is a common SNP (rs6265) in the
human BDNF gene (BDNF Val66Met; Egan etal., 2003). The BDNF gene is an ideal
candidate for individual differences in sensitivity to context since BDNF is widely
expressed throughout the CNS and obligatory for multiple forms of experience-
dependent synaptic plasticity. The Val66Met variant leads to a change from valine
(Val) to methionine (Met) at amino acid position 66 within the pro-domain of
BDNF. Approximately 30% of the Caucasian population are carriers of the Met
allele, with ~4% being homozygous Met/Met, with a substantially higher frequency
of the Met allele in Asian samples (Petryshen etal., 2010). The Met allele leads to a
reduction in the activity-dependent release of BDNF (Chen etal., 2005, 2006; Egan
etal., 2003) and is associated with hippocampal-dependent memory function (Hariri
etal., 2003) and multiple neuropsychiatric disorders (Duman & Monteggia, 2006;
Groves, 2007; Rybakowski, 2008). The BDNF Val66Met polymorphism also associ-
ates with direct measures of structural plasticity in the cortex during learning and
memory (Kleim etal., 2006; McHughen etal., 2010; Wang etal., 2014).
Li etal. examined the association of antenatal maternal anxiety with DNA meth-
ylation profiles as a function of the Val66Met BDNF genotype using genomic DNA
obtained from umbilical chords with the same 450K array described earlier. There
was a greater influence of antenatal maternal anxiety on the neonatal epigenome
among Met/Met compared to Val/Val carriers, with a trend for a Met allele dosage
effect. This finding is consistent with previous studies revealing differential sensitivity
in carriers of the BDNF Met/Met compared to the Val/Val genotype. Several popula-
tion-based studies examined the interaction between the BDNF Val66Met polymor-
phism and early adversity on adult depression (Aguilera etal., 2009; Carver, Johnson,
Joormann, Lemoult, & Cuccaro, 2011; Chen etal., 2013; Gatt etal., 2009; Wichers
etal., 2008). The results of these studies suggest a greater impact of early adversity on
risk for affective disorders in BDNF Met allele carriers than among those with the Val/
Val genotype. Likewise, mice carrying the Met/Met variant show increased stress-
induced anxiety-like behaviours (Chen et al., 2006). The Met allele also increases
sensitivity to both positive and negative familial influences (Hayden et al., 2010).
Likewise, children adopted from orphanages with Val/Met or Met/Met genotypes
exhibited fewer attention regulatory problems than Val/Val genotypes when adopted
very early and more symptoms when adoption occurred later in development (Gunnar
etal., 2012). In sum, the data suggest greater sensitivity to the effects of antenatal
maternal anxiety on the human methylome in Met/Met individuals.
The Li et al. study included neuroimaging data collected shortly after birth.
Imaging studies with this cohort (Qiu etal., 2013; Rifkin-Graboi etal., 2013) reveal
that antenatal maternal emotional well-being predicts variation in volume and
microstructure brain regions associated with emotional function (Etkin & Wager,
2007; Meaney, LeDoux, & Leibowitz, 2008). This imaging dataset was then anal-
ysed in relation to the differential methylation of CpG sites as a function of maternal
80 M.J. Meaney
anxiety. A disproportionate number of variably methylated CpGs in the Met/Met

group associated with right amygdala volume. In contrast, there was an even more
disproportionate number of affected CpGs in the Val/Val group that associated with
left hippocampal volume. These findings suggest that the effects of BDNF genotype
were both quantitative (i.e. there were more CpGs affected by maternal mood in
Met/Met carriers) and qualitative. The CpGs affected by maternal mood in the Val/
Val group differed from those in the Met/Met groups and associated with hippocam-
pal development; those affected in the Met/Met group were more commonly associ-
ated with amygdala structure.
In contrast to these findings of increased evidence for differential sensitivity
(Belsky et al., 2009; Boyce & Ellis, 2005) among Met/Met carriers, there is also
evidence for greater environmental plasticity among those bearing the Val/Val BDNF
Val66Met genotype. Human neuroimaging research demonstrates that Met allele
carriers show a deficit in fear conditioning and impaired aversive memory acquisi-
tion relative to Val/Val carriers (Hajcak etal., 2009; Lonsdorf etal., 2010). These
findings suggest increased plasticity among Val/Val carriers. Imaging studies also
provide evidence for increased experience-dependent neuroplasticity among Val/Val
carriers of the BDNF Val66Met polymorphism. These studies used neuroimaging to
show increased training-dependent plasticity in the motor cortex (Kleim etal., 2006;
McHughen etal., 2010), revealing increased plasticity in Val/Val subjects. There is
also evidence of greater stress-induced HPA activation (Alexander et al., 2010;
Shalev etal., 2009) among Val/Val carriers. Likewise, bdnf Val/Val mice demonstrate
a greater stress-induced reduction in social activity compared with Met/Met mice
(Krishnan etal., 2007). These findings suggest that the influence of the bdnf geno-
type in moderating the impact of the environmental condition varies as a function of
the outcome under study.
The Li etal. findings provide a potential explanation for the inconsistent plastic-
ity gene findings in studies comparing Met/Met and Val/Val carriers. There are
studies that associate the Met/Met variant of the BDNF Val66Met polymorphism
with either anxiety disorders or with anxiety-related endophenotypes, such as harm
avoidance (Jiang etal., 2005; Montag, Basten, Stelzel, Fiebach, & Reuter, 2010).
Thus, transgenic BDNF Met/Met mice manifest increased anxious-like behaviours
(Chen etal., 2006). Li etal. (2015) report a significantly greater number of variably
methylated CpGs associated with right amygdala volume among newborns with the
Met/Met compared to the Val/Val BDNF genotype. These findings are consistent
with earlier reports suggesting that carriers of the Met variant showed stronger
amygdala activation in the right hemisphere in response to emotional stimuli com-
pared to neutral stimuli (Lau etal., 2010; Montag, Reuter, Newport, Elger, & Weber,
2008). There is also evidence for a selective association between measures of activ-
ity in the right amygdala and conditions of anxiety (Etkin & Wager, 2007).
Individuals at risk for anxiety disorders show increased activity in the amygdala
(Stein, Simmons, Feinstein, & Paulus, 2007) as do patients with anxiety disorders
(Rauch, Savage, Alpert, Fischman, & Jenike, 1997; Wright, Martis, McMullin,
Shin, & Rauch, 2003), and there is evidence for selective associations between anxi-
ety and the right insula, which is closely connected to the right amygdala. Buss etal.
(2012), Paulus, Rogalsky, Simmons, Feinstein, and Stein (2003) and Wright etal.
(2003) found that antenatal maternal cortisol levels, presumably reflecting greater
maternal distress, were selectively associated with increased right amygdala volume
in the offspring during childhood. The disproportionate number of CpGs that co-
vary with antenatal maternal anxiety and right amygdala volume in Met/Met infants
is consistent with the association of this brain region with anxiety-like states.
Li etal. also report a disproportionate number of variably methylated CpGs that
co-vary selectively with left hippocampal volume in Val/Val infants. The studies
reviewed earlier, including those involving fear conditioning, examine forms of plas-
ticity (e.g. learning and memory) that are hippocampal dependent. The brain-region-
specific associations of CpGs (right amygdala vs left hippocampus) in the Met/Met
and Val are consistent with the idea that the BDNF genotype of the offspring deter-
mines the nature of the effect of maternal anxiety and other uterine environments, on
the epigenome and neurodevelopment of the offspring. The confusion in the litera-
ture concerning the apparent differential susceptibility of the Met/Met and Val/Val
genotypes occurs because of the differential neural structures under study. Li etal.
(2015) suggest that the evidence for greater sensitivity to the effect of antenatal
maternal anxiety at the level of the epigenome amongst among Met/Met or Val/Val
carriers was dependent upon whether we considered variation in DNA methylation
that also co-varied with hippocampal or amygdala volume at birth. Thus, the differ-
ence in sensitivity to context associated with the Val66Met polymorphism is likely to
be dependent upon the specific neural function under study: we should not expect
that variants in genes that mediate neuroplasticity will necessarily reveal uniform
differences in plasticity regardless of the nature of the functional outcome and
underlying neural systems.
An obvious question concerns the biology underlying the potential diversity of
effects associated with the Val66Met BDNF polymorphism. BDNF acts as a trophic
factor that moderates activity-dependent synaptic strength throughout the CNS.Thus,
the specificity of the effect of maternal anxiety is unlikely to be associated with factors
that underlie BDNF signalling only. However, BDNF acts within the synaptic context
in concert with multiple neurotransmitter systems such that the specificity of a BDNF
effect is defined by signalling partners that are commonly classical neurotransmitter
systems. Therefore, the impact of BDNF polymorphisms is moderated by variants of
genes that moderate local neurotransmitter signalling. Hnnerkopf etal. (2007) report
an interaction effect of BDNF Val66Met and a variation on the dopamine transporter
gene (DAT) on harm avoidance. Other studies report interaction effects of BDNF
Val66Met and the SLC6A4 polymorphism on serotonin-dependent emotional states
such as depression (Kim etal., 2007; Martinowich & Lu, 2008), obsessivecompul-
sive disorder (Wendland, Kruse, Cromer, & Murphy, 2007) and a three-way interac-
tion effect of BDNF Val66Met, the SLC6A4 polymorphism, which affects the
transcription of the serotonin transporter mRNA, and child adversity on depressive
symptoms (Wichers et al., 2008). Since synaptic plasticity occurs as a function of
multiple, interacting signals, each potentially affected by numerous sequence variants,
regional variation in the influence of the Val66Met polymorphism might be expected.
Li etal. (2015) reported that the Met/Met variant was often associated with differen-
82 M.J. Meaney
tial DNA methylation in genes associated with dopamine signalling, while the Val/Val
variant was more so associated with genes that encode for products that directly regu-
late synaptic strength (e.g. NGF). These findings suggest a qualitatively different
effect of antenatal maternal anxiety based on BDNF genotype.
An additional publication from this same cohort provides even more direct evi-
dence for this point. Qiu etal. (2013) examined the influence of antenatal maternal
anxiety of cortical structure as a function of COMT genotype. The catechol-O-
methyltransferase (COMT) gene, located on chromosome 22q11.2, regulates cate-
cholamine signalling in the prefrontal cortex and is implicated in anxiety, attention
and stress responsivity. Qiu etal. examined whether individual SNPs of the COMT
gene and their haplotypes moderate the association between antenatal maternal
anxiety and in utero cortical development. Neonatal cortical morphology was char-
acterized using cortical thickness in an MRI study. Individual COMT SNPs (val-
158met, rs737865 and rs165599) modulated the association between antenatal
maternal anxiety and the prefrontal and parietal cortical thickness of neonates.
Haplotype trend regression analysis revealed that the A-val-G (AGG) of rs737865-
val158met-rs165599 haplotype probabilities modulated positive associations of
antenatal maternal anxiety with cortical thickness in right ventrolateral prefrontal
cortex, and right superior parietal cortex and precuneus. In contrast, the G-met-A
(GAA) haplotype probabilities modulated negative associations of antenatal mater-
nal anxiety with cortical thickness in bilateral precentral gyrus and dorsolateral pre-
frontal cortex. The results suggest that the association between maternal anxiety and
in utero neurodevelopment is modified through complex genetic variation in COMT.
Such genetic moderation may explain, in part, the variation in phenotypic outcomes
in offspring associated with maternal emotional well-being.
Implications forPrevention Programmes
Childhood maltreatment, as well as other forms of developmental adversity (e.g.

poverty, poor birth outcomes) predict an increased risk for mental disorders and
impaired capacity. However, studies of gene x environment interactions reveal
remarkable variation in developmental outcomes even in cases of extreme adversity.
These same studies also show that the relevant genotypes associate not with the risk
for mental disorders per se, but rather with individual differences in the sensitivity
of the individual to environmental conditions. Such plasticity genes commonly
involve variants that influence stress reactivity and thus the degree to which indi-
viduals are affected by environmental adversity. These findings reflect the reality of
childhood adversity: amongst more sensitive individuals, even seemingly moder-
ate forms of childhood adversity reflected in measures of the quality of parental
care predict the risk for poor mental health and quality of life. It thus appears that
neither measures of genotype nor of childhood adversity alone can effectively cap-
ture the vulnerability of individuals. We suggest that the most effective measures of
vulnerability in specific functional domains will be epigenetic states that reflect the
relevant gene x environment interactions and represent the stable impact of the
adversity on the individual. Epigenetic measures might thus constitute a biological
registry of the impact of the environmental adversity at the level of the individual.
The wide variability in mental health outcomes associated with early life adver-
sity suggests that global measures derived from population studies are not ideal in
defining vulnerability at the level of the individual child. Genetic variants, and espe-
cially variants in genes implicated in stress responses, can moderate the impact of
adversity rendering individuals more or less sensitive. Moreover, in more reactive
children even seemingly milder forms of adversity may compromise development
and function. The interactive nature of genetic and environmental influences implies
that measures of childhood adversity alone can effectively capture the vulnerability
of individuals. Advances in intervention programmes that effectively targeted the
most vulnerable children must assess the impact of adversity at the level of the indi-
vidual child. Epigenetic measures, particular the chemically more stable marks,
such as DNA methylation, may ultimately serve to meet this challenge.
Conclusions
Studies of the determinants of variation across the human epigenome reveal strong
evidence for G x E interaction effects. These results are consistent with those using
rodent models. Environmental enrichment alters DNA methylation across the
genome in hippocampal tissue and the effect is moderated by genotype (Guo etal.,
2011). Environmental conditions alter cellular activity through specific signalling
pathways that are subject to the influence of genomic variation. Polymorphisms that
influence the efficacy of a signalling pathway can be expected to alter the influence
of environmental conditions that operate through that pathway, and less so pathways
unaffected by the same polymorphisms. Signalling pathways vary across brain
regions, and thus we can expect that the influence of genetic polymorphisms, and
thus G x E interactions, will vary across brain regions. In the studies described ear-
lier, the gene x environment interaction effects operated in a brain region-specific
manner. Thus, the difference in apparent sensitivity to context associated with the
Val66Met BDNF polymorphism or the COMT haplotypes was dependent upon the
specific neural system. We should not expect that variants in genes that mediate
neuroplasticity necessarily reveal uniform differences in plasticity regardless of
the nature of the functional outcome and underlying neural systems.
There are very real issues confronting the study of G x E interactions. These issues
will require a concerted effort to increase statistical power in a manner similar to that
achieved by the Psychiatric Genetics Consortium in the study of genomic variants and
psychiatric disorders. The issue of biological plausibility, however, no longer seems in
question. The study of epigenetics provides a powerful model by which we can under-
stand how it is that environmental conditions, including social context, can biologi-
cally interact with the genome at the level of both structure and function. The reality
of G x E interactions should provide the necessary incentive for the big data
approaches that will be necessary to resolve the issues of replicability.
84 M.J. Meaney
References
Aguilera, M., Arias, B., Wichers, M., Barrantes-Vidal, N., Moya, J., Villa, H., Fananas, L. (2009).
Early adversity and 5-HTT/BDNF genes: New evidence of gene-environment interactions on
depressive symptoms in a general population. Psychological Medicine, 39, 14251432.
Alexander, N., Osinsky, R., Schmitz, A., Mueller, E., Kuepper, Y., & Hennig, J. (2010). The BDNF
Val66Met polymorphism affects HPA-axis reactivity to acute stress. Psychoneuroendocrinology,
35, 949953.
Avishai-Eliner, S. G. E., Eghbal-Ahmadi, M., Bar-El, Y., & Baram, T. Z. (2001). Altered regulation
of gene and protein expression of hypothalamicpituitaryadrenal axis components in an
immature rat model of chronic stress. Journal of Neuroendocrinology, 13, 799807.
Bateson, P., Barker, D., Clutton-Brock, T., Deb, D., DUdine, B., Foley, R. A., Sultan, S. E.
(2004). Developmental plasticity and human health. Nature, 430, 419421.
Beach, S. R., Brody, G. H., Todorov, A. A., Gunter, T. D., & Philibert, R. A. (2011). Methylation
at 5Htt mediates the impact of child sex abuse on womens antisocial behavior: An examination
of the Iowa adoptee sample. Psychosomatic Medicine, 73, 8387.
Beach SRH, Brody GGH, Todorov AA, Gunter TD, Philibert RA (2010) Methylation at SLC6A4
is linked to Family History of Child Abuse: An Examination of the Iowa Adoptee Sample. Am
J Med Genet N Neuropsyciatr Genet 153B:710713.
Belsky, J., Bakermans-Kranenburg, M. J., & Van IJzendoorn, M. H. (2007). For better and for
worse: Differential susceptibility to environmental influences. Current Directions in
Belsky, J., Jonassaint, C., Pluess, M., Stanton, M., Brummett, B., & Williams, R. (2009).
Vulnerability genes or plasticity genes? Molecular Psychiatry, 14, 746754.
Berger, S. L. (2007). The complex language of chromatin regulation during transcription. Nature,
447(7143), 407412.
Bernstein, B. E., Kamal, M., Lindblad-Toh, K., Bekiranov, S., Bailey, D. K., Huebert, D. J.,
Lander, E. S. (2005). Genomic maps and comparative analysis of histone modifications in
human and mouse. Cell, 120, 169181.
Bhutani, N., Burns, D. M., & Blau, H. M. (2011). DNA demethylation dynamics. Cell, 146,
866872.
Binder, E. B., Bradley, R. G., Liu, W., Epstein, M. P., Deveau, T. C., Mercer, K. B., Ressler, K.
J. (2008). Association of FKBP5 polymorphisms and childhood abuse with risk of posttrau-
matic stress disorder symptoms in adults. JAMA, 299, 12911305.
Bird, A. P. (1986). CpG-rich islands and the function of DNA methylation. Nature, 321,
209213.
Bird, A. P. (2002). DNA methylation patterns and epigenetic memory. Genes & Development, 16,
621.
Bird, A. P. (2007). Perceptions of epigenetics. Nature, 447, 396398.
Borrow, A. P., Levy, M. J., Soehngen, E. P., & Cameron, N. M. (2013). Perinatal testosterone expo-
sure and maternal care effects on the female rats development and sexual behaviour. Journal
of Neuroendocrinology, 25, 528536.
Boyce, W. T., & Ellis, B. J. (2005). Biological sensitivity to context: I. An evolutionary-develop-
mental theory of the origins and functions of stress reactivity. Development and Psychopathology,
17(2), 271301.
Bradley, R. H., & Corwyn, R. F. (2008). Infant temperament, parenting, and externalizing behavior
in first grade: A test of the differential susceptibility hypothesis. Journal of Child Psychology
and Psychiatry, 49, 124131.
Branco, M. R., Ficz, G., & Reik, W. (2012). Uncovering the role of 5 hydroxymethylcytosine in
the epigenome. Nature Reviews Genetics, 13, 713.
Bruniquel, D., & Schwartz, R. H. (2003). Selective, stable demethylation of the interleukin-2 gene
enhances transcription by an active process. Nature Immunology, 4, 235240.
Buss, C., Davis, E. P., Shahbaba, B., Pruessner, J. C., Head, K., & Sandman, C. A. (2012). Maternal
cortisol over the course of pregnancy and subsequent child amygdala and hippocampus vol-
umes and affective problems. Proceedings of the National Academy of Sciences of the United
States of America, 109, E1312E1319.
Caldji, C., Diorio, J., & Meaney, M. J. (2000). Maternal care regulates the development of indi-
vidual differences in in stress reactivity through effects on GABAergic systems. Biological
Psychiatry, 48, 11641174.
Caldji, C., Diorio, J., & Meaney, M. J. (2003). Variations in maternal care alter GABAA receptor
subunit expression in brain regions associated with fear. Neuropsychopharmacology, 28,
150159.
Caldji, C. T. B., Sharma, S., Francis, D., Plotsky, P. M., & Meaney, M. J. (1998). Maternal care
during infancy regulates the development of neural systems mediating the expression of behav-
ioral fearfulness in adulthood in the rat. Proceedings of the National Academy of Sciences, 95,
53355340.
Cameron, N. M., Del Corpo, A., Diorio, J., McAllister, K., Sharma, S., & Meaney, M. J. (2008).
Maternal programming of sexual behavior and hypothalamicpituitarygonadal function in the
female rat. PLoS One, 3, e2210.
Carver, C. S., Johnson, S. L., Joormann, J., Lemoult, J., & Cuccaro, M. L. (2011). Childhood
adversity interacts separately with 5-HTTLPR and BDNF to predict lifetime depression diag-
nosis. Journal of Affective Disorders, 132, 8993.
Casey, B. J., Glatt, C. E., Tottenham, N., Soliman, F., Bath, K., Amso, D., Lee, F. S. (2009).
Brain-derived neurotrophic factor as a model system for examining gene by environment inter-
actions across development. Neuroscience, 164, 108120.
Caspi, A., Hariri, A. R., Holmes, A., Uher, R., & Moffitt, T. E. (2010). Genetic sensitivity to the
environment: The case of the serotonin transporter gene and its implications for studying com-
plex diseases and traits. The American Journal of Psychiatry, 167, 509527.
Champagne, F. A. (2008). Epigenetic mechanisms and the transgenerational effects of maternal
care. Frontiers in Neuroendocrinology, 29, 386397.
Champagne, F. A., Francis, D. D., Mar, A., & Meaney, M. J. (2003). Naturally-occurring variations
in maternal care in the rat as a mediating influence for the effects of environment on the devel-
opment of individual differences in stress reactivity. Physiology & Behavior, 79, 359371.
Champagne, F. A., Weaver, I. C., Diorio, J., Dymov, S., Szyf, M., & Meaney, M. J. (2006). Maternal
care regulates methylation of the estrogen receptor alpha 1b promoter and estrogen receptor
alpha expression in the medial preoptic area of female offspring. Endocrinology, 147,
29092915.
Chen, J., Li, X., & McGue, M. (2013). The interacting effect of the BDNF Val66Met polymor-
phism and stressful life events on adolescent depression is not an artifact of gene-environment
correlation: Evidence from a longitudinal twin study. Journal of Child Psychology and
Chen, Z. Y., Ieraci, A., Teng, H., Dall, H., Meng, C. X., Herrera, D. G., Lee, F. S. (2005). Sortilin
controls intracellular sorting of brain-derived neurotrophic factor to the regulated secretory
pathway. The Journal of Neuroscience, 25, 61566166.
Chen, Z. Y., Jing, D., Bath, K. G., Ieraci, A., Khan, T., Siao, C. J., Lee, F. S. (2006). Genetic
variant BDNF (Val66Met) polymorphism alters anxiety-related behavior. Science, 314,
140143.
Cooney, C. A., Dave, A. A., & Wolff, G. L. (2002). Maternal methyl supplements in mice affect
epigenetic variation and DNA methylation of offspring. Journal of Nutrition, 132, 2393S2400S.
Deaton, A. M., & Bird, A. (2011). Cpg islands and the regulation of transcription. Genes &
Development, 25, 10101022.
Dedeurwaerder, S., Defrance, M., Calonne, E., Denis, H., Sotiriou, C., & Fuks, F. (2011).
Evaluation of the infinium methylation 450K technology. Epigenomics, 3, 771784.
Devlin, A. M., Brain, U., Austin, J., & Oberlander, T. F. (2010). Prenatal exposure to maternal
depressed mood and the MTHFR C677T variant affect SLC6A4 methylation in infants at birth.
PLoS One, 5, e12201.
86 M.J. Meaney
Docherty, S., Davis, O. S., Haworth, C. M., Plomin, R., DSouza, U., & Mill, J. (2012). A genetic
association study of DNA methylation levels in the drd4 gene region finds associations with
nearby snps. Behavioral and Brain Functions, 8, 31.
Duman, R. S., & Monteggia, L. M. (2006). A neurotrophic model for stress-related mood disor-
ders. Biological Psychiatry, 59, 11161127.
Duncan, L. E., & Keller, M. C. (2011). A critical review of the first 10 years of candidate gene-by-
environment interaction research in psychiatry. The American Journal of Psychiatry, 168,
10411049.
Egan, M. F., Kojima, M., Callicott, J. H., Goldberg, T. E., Kolachana, B. S., Bertolino, A.,
Weinberger, D. R. (2003). The BDNF val66met polymorphism affects activity-dependent
secretion of BDNF and human memory and hippocampal function. Cell, 112, 257269.
Ellis, B. J., Boyce, W. T., Belsky, J., Bakermans-Kranenburg, M. J., & van IJzendoorn, M. H.
(2011). Differential susceptibility to the environment: An evolutionary-neurodevelopmental
theory. Development and Psychopathology, 23, 728.
Essex, M. J., Boyce, W. T., Hertzman, C., Lam, L. L., Armstrong, J. M., Neumann, S. M., & Kobor,
M. S. (2013). Epigenetic vestiges of early developmental adversity: Childhood stress exposure
and DNA methylation in adolescence. Child Development, 84, 5875.
Etkin, A., & Wager, T. D. (2007). Functional neuroimaging of anxiety: A meta-analysis of emo-
tional processing in PTSD, social anxiety disorder, and specific phobia. The American Journal
of Psychiatry, 164, 14761488.
Fan, G., Martinowich, K., Chin, M. H., et al. (2005). DNA methylation controls the timing of
astrogliogenesis through regulation of JAK-STAT signaling. Development, 132, 33453356.
Feng, J., & Nestler, E. J. (2013). Epigenetic mechanisms of drug addiction. Current Opinion in
Neurobiology, 23, 521528.
Francis, D., Diorio, J., Liu, D., & Meaney, M. J. (1999). Nongenomic transmission across genera-
tions of maternal behavior and stress responses in the rat. Science, 286, 11551158.
Gatt, J. M., Nemeroff, C. B., Dobson-Stone, C., Paul, R. H., Bryant, R. A., Schofield, P. R.,
Williams, L. M. (2009). Interactions between BDNF Val66Met polymorphism and early life
stress predict brain and arousal pathways to syndromal depression and anxiety. Molecular
Gertz, J., Varley, K. E., Reddy, T. E., Bowling, K. M., Pauli, F., Parker, S. L., Myers, R. M.
(2011). Analysis of DNA methylation in a three-generation family reveals widespread genetic
influence on epigenetic regulation. PLoS Genetics, 7, e1002228.
Gluckman, P. D., & Hanson, M. A. (2004). Living with the past: Evolution, development, and pat-
terns of disease. Science, 305, 17331736.
Gluckman, P. D., & Hanson, M. A. (2007). Developmental plasticity and human disease: Research
directions. Journal of Internal Medicine, 261, 461471.
Gratten, J., Wray, N. R., Keller, C. M., & Visscher, P. M. (2014). Large-scale genomics unveils the
genetic architecture of psychiatric disorders. Nature Neuroscience, 17, 782790.
Groves, J. O. (2007). Is it time to reassess the BDNF hypothesis of depression? Molecular
Psychiatry, 12(12), 10791088.
Grunstein, M. (1997). Histone acetylation in chromatin structure and transcription. Nature, 389,
349352.
Gunnar, M. R., Wenner, J. A., Thomas, K. M., Glatt, C. E., McKenna, M. C., & Clark, A. G. (2012).
The brain-derived neurotrophic factor Val66Met polymorphism moderates early deprivation
effects on attention problems. Development and Psychopathology, 24, 12151223.
Guo, J. U., Ma, D. K., Mo, H., Ball, M. P., Jang, M.-H., Bonaguidi, M. A., Song, H. (2011).
Neuronal activity modifies the DNA methylation landscape in the adult brain. Nature
Hackett, J. A., Sengupta, R., Zylicz, J. J., Murakami, K., Lee, C., Down, T. A., & Surani, M. A.
(2013). Germline DNA demethylation dynamics and imprint erasure through 5-hydroxymeth-
ylcytosine. Science, 339, 448452.
Hajcak, G., Castille, C., Olvet, D. M., Dunning, J. P., Roohi, J., & Hatchwell, E. (2009). Genetic
variation in brain-derived neurotrophic factor and human fear conditioning. Genes, Brain, and
Behavior, 8, 8085.
Hake, S. B., & Allis, D. C. (2006). Histone H3 variants and their potential role in indexing mam-
malian genomes: The H3 barcode hypothesis. Proceedings of the National Academy of
Sciences of the United States of America, 103, 64286435.
Hansen, K. D., Timp, W., Bravo, H. C., Sabunciyan, S., Langmead, B., McDonald, O. G.,
Feinberg, A. P. (2011). Increasedmethylation variation in epigenetic domains across cancer
types. Nature Genetics, 43, 768775.
Hariri, A. R., Goldberg, T. E., Mattay, V. S., Kolachana, B. S., Callicott, J. H., Egan, M. F., &
Weinberger, D. R. (2003). Brain-derived neurotrophic factor val66met polymorphism affects
human memory-related hippocampal activity and predicts memory performance. The Journal
of Neuroscience, 23, 66906694.
Hayden, E. P., Klein, D. N., Dougherty, L. R., Olino, T. M., Dyson, M. W., Durbin, C. E., Singh,
S. M. (2010). The role of brain-derived neurotrophic factor genotype, parental depression, and
relationship discord in predicting early-emerging negative emotionality. Psychological Science,
21, 16781685.
Heim, C., & Nemeroff, C. B. (2009). Neurobiology of posttraumatic stress disorder. CNS
Spectrums, 14(1 Suppl 1), 1324.
Hellstrom, I. C., Dhir, S., Diorio, J., Zhang, T. Y., & Meaney, M. J. (2012). Maternal licking regu-
lates hippocampal glucocorticoid receptor transcription through a thyroid hormone, serotonin,
NGFI-A signalling cascade. Philosophical Transactions of the Royal Society, B: Biological
Sciences, 367, 24952510.
Herb, B. R., Wolschin, F., Hansen, K. D., Aryee, M. J., Langmead, B., Irizarry, R., Feinberg, A.
P. (2012). Reversible switching between epigenetic states in honeybee behavioral subcastes.
Nature Neuroscience, 15, 13711373.
Hilt, L. M., Sander, L. C., Nolen-Hoeksema, S., & Simen, A. A. (2007). The BDNF Val66Met
polymorphism predicts rumination and depression differently in young adolescent girls and
their mothers. Neuroscience Letters, 429, 1216.
Hofer, M. A. (1984). Early stages in the organization of cardiovascular control. Proceedings of the
Society for Experimental Biology and Medicine, 175, 147157.
Hnnerkopf R, Strobel A, Gutknecht L, Brocke B, Lesch PK (2007) Interaction between BDNF
Val66Met and Dopamine Transporter Gene Variation Influences Anxiety-Related Traits.
Neuropsychopharmacology 32:25522560.
Irizarry, R. A., Ladd-Acosta, C., Wen, B., Wu, Z., Montano, C., Onyango, P., Feinberg, A. P.
(2009). The human colon cancer methylome shows similar hypo- and hypermethylation at
conserved tissue-specific CpG island shores. Nature Genetics, 41, 178186.
Ito, S. D., DAlessio, A. C., Taranova, O. V., Hong, K., Sowers, L. C., & Zhang, Y. (2010). Role of
Tet proteins in 5mC to 5hmC conversion, ES-cell self-renewal and inner cell mass specifica-
tion. Nature, 466, 11291133.
Jenuwein, T., & Allis, C. D. (2001). Translating the histone code. Science, 293, 10741080.
Jiang, X., Xu, K., Hoberman, J., Tian, F., Marko, A. J., Waheed, J. F., Lipsky, R. H. (2005).
BDNF variation and mood disorders: A novel functional promoter polymorphism and Val66Met
are associated with anxiety but have opposing effects. Neuropsychopharmacology, 30,
13531361.
Jin, S.-G., Wu, X., Li, A. X., & Pfeifer, G. P. (2011). Genomic mapping of 5-hydroxymethylcyto-
sine in the human brain. Nucleic Acids Research, 39, 50155024.
Jirtle, R. L., & Skinner, M. K. (2007). Environmental epigenomics and disease susceptibility.
Nature Reviews Genetics, 8, 253262.
Kaminsky, Z. A., Tang, T., Wang, S.-C., Ptak, C., Oh, G. H., Wong, A. H., Petronis, A. (2009).
DNA methylation profiles in monozygotic and dizygotic twins. Nature Genetics, 41,
240245.
Karg, K., Burmeister, M., Shedden, K. M., & Sen, S. (2011). The serotonin transporter promoter
variant (5-HTTLPR), stress, and depression meta-analysis revisited: Evidence of genetic mod-
eration. Archives of General Psychiatry, 68, 44454.
Kim, C. H., Cheon, K. A., Koo, M. S., Ryu, Y. H., Lee, J. D., Chang, J. W., & Lee, H. S. (2007).
Dopamine transporter density in the basal ganglia in obsessive-compulsive disorder, measured
88 M.J. Meaney
with [123I]IPT SPECT before and after treatment with serotonin reuptake inhibitors.
Neuropsychobiology, 55, 156162.
Kleim, J. A., Chan, S., Pringle, E., Schallert, K., Procaccio, V., Jimenez, R., & Cramer, S. C.
(2006). BDNF val66met polymorphism is associated with modified experience-dependent
plasticity in human motor cortex. Nature Neuroscience, 9, 735737.
Klengel, T., & Binder, E. B. (2015). FKBP5 allele-specific epigenetic modification in gene by
environment interaction. Neuropsychopharmacology, 40, 244246.
Klengel, T., Mehta, D., Anacker, C., Rex-Haffner, M., Pruessner, J. C., Pariante, C. M., Binder,
E. B. (2013). Allele-specific FKBP5 DNA demethylation mediates gene-childhood trauma
interactions. Nature Neuroscience, 16, 3341.
Klose, R. J., & Bird, A. P. (2006). Genomic DNA methylation: The mark and its mediators. Trends
in Biochemical Sciences, 31, 8997.
Korosi, A., Shanabrough, M., McClelland, S., Liu, Z. W., Borok, E., Gao, X. B., Baram, T. Z.
(2010). Early-life experience reduces excitation to stress-responsive hypothalamic neurons and
reprograms the expression of corticotropin-releasing hormone. The Journal of Neuroscience,
30, 703713.
Kouzarides, T. (2007). Chromatin modifications and their function. Cell, 128(4), 693705.
Kriaucionis, S., & Heintz, N. (2009). The nuclear DNA base 5-hydroxymethylcytosine is present
in Purkinje neurons and the brain. Science, 324, 929930.
Krishnan, V., Han, M. H., Graham, D. L., Berton, O., Renthal, W., Russo, S. J., Nestler, E. J.
(2007). Molecular adaptations underlying susceptibility and resistance to social defeat in brain
reward regions. Cell, 131, 391404.
Labonte, B., Yerko, V., Gross, J., Mechawar, N., Meaney, M. J., Szyf, M., & Turecki, G. (2012).
Differential glucocorticoid receptor exon 1(B), 1(C), and 1(H) expression and methylation in
suicide completers with a history of childhood abuse. Biological Psychiatry, 72, 4148.
Labont, B. S. M., Maussion, G., Lopez, J. P., Navarro-Snchez, L., Yerko, V., Mechawar, N.,
Turecki, G. (2013). Genome-wide methylation changes in the brains of suicide completers. The
American Journal of Psychiatry, 170(5), 511520.
Lam, L. L., Emberly, E., Fraser, H. B., Neumann, S. M., Chen, E., Miller, G. E., & Kobor, M. S.
(2012). Factors underlying variable DNA methylation in a human community cohort.
Proceedings of the National Academy of Sciences, 109(suppl 2), 1725317260.
Lau, J. Y., Goldman, D., Buzas, B., Hodgkinson, C., Leibenluft, E., Nelson, E., Ernst, M.
(2010). BDNF gene polymorphism (Val66Met) predicts amygdala and anterior hippocampus
responses to emotional faces in anxious and depressed adolescents. NeuroImage, 53,
952961.
Lesch, K. P., Bengel, D., Heils, A., Sabol, S. Z., Greenberg, B. D., Petri, S., et al. (1996). Association
of anxiety-related traits with a polymorphism in the serotonin transporter gene regulatory
region. Science, 274, 15271531.
Levine, S. (1994). The ontogeny of the hypothalamicpituitaryadrenal axis. The influence of
maternal factors. Annals of the New York Academy of Sciences, 746, 275288.
Li C, Pan H, Tuan TA, Teh AL, MacIsaac JL, Mah SM, McEwen LM, Li Y, Chen H, Broekman BF,
Buschdorf JP, Chong YS, Kwek K, Saw SM, Gluckman PD, Fortier MV, Rifkin-Graboi A,
Kobor MS, Qiu A, Meaney MJ, Holbrook JD, Gusto Study G (2015) Brain-derived neuro-
trophic factor (BDNF) Val66Met polymorphism influences the association of the methylome
with maternal anxiety and neonatal brain volumes. Dev Psychopathol 27:137150.
Lienert, F., Wirbelauer, C., Som, I., Dean, A., Mohn, F., & Schbeler, D. (2011). Identification of
genetic elements that autonomously determine DNA methylation states. Nature Genetics, 43,
10911097.
Lister, R., Mukamel, E. A., Nery, J. R., Urich, M., Puddifoot, C. A., Johnson, N. D., Ecker, J. R.
(2013). Global epigenomic reconfiguration during mammalian brain development. Science,
341, 1237905.
Liu, D., Diorio, J., Tannenbaum, B., Caldji, C., Francis, D., Freedman, A., Meaney, M. J.
(1997). Maternal care, hippocampal glucocorticoid receptors, and hypothalamicpituitary
adrenal responses to stress. Science, 277, 16591662.
Lonsdorf, T. B., Weike, A. I., Golkar, A., Schalling, M., Hamm, A. O., & Ohman, A. (2010).
Amygdala-dependent fear conditioning in humans is modulated by the BDNFval66met
polymorphism. Behavioral Neuroscience, 124, 915.
Lubin, F. D., Roth, T. L., & Sweatt, J. D. (2008). Epigenetic regulation of BDNF gene transcription
in the consolidation of fear memory. The Journal of Neuroscience, 28, 1057610586.
Luger, K., Mader, A. W., Richmond, R. K., Sargent, D. F., & Richmond, T. J. (1997). Crystal structure
of the nucleosome core particle at 2.8 A resolution. Nature, 389, 251260.
Ma, D. K., Jang, M. H., Guo, J. U., Kitabatake, Y., Chang, M. L., Pow-Anpongkul, N., Song, H.
(2009). Neuronal activity-induced Gadd45b promotes epigenetic DNA demethylation and
adult neurogenesis. Science, 323, 10741077.
Macleod, D., Charlton, J., Mullins, J., & Bird, A. P. (1994). Sp1 sites in the mouse aprt gene pro-
moter are required to prevent methylation of the CpG island. Genes & Development, 8,
22822292.
Martinowich, K., Hattori, D., Wu, H., Fouse, S., He, F., Hu, Y., Sun, Y. E. (2003). DNA methyl-
ation-related chromatin remodeling in activity-dependent Bdnf gene regulation. Science, 302,
890893.
Martinowich, K., & Lu, B. (2008). Interaction between BDNF and serotonin: Role in mood disor-
ders. Neuropsychopharmacology, 33, 7383.
Maunakea, A. K., Nagarajan, R. P., Bilenky, M., Ballinger, T. J., DSouza, C., Fouse, S. D.,
Costello, J. F. (2010). Conserved role of intragenic DNA methylation in regulating alternative
promoters. Nature, 466, 253257.
Maze, I., & Nestler, E. J. (2011). The epigenetic landscape of addiction. Annals of the New York
Academy of Sciences, 1216, 99113.
Maze, I., Noh, K.-M., & Allis, C. D. (2013). Histone regulation in the CNS: Basic principles of
epigenetic plasticity. Neuropsychopharmacology, 38, 322.
McCormick, J. A., Lyons, V., Jacobson, M. D., Noble, J., Diorio, J., Nyirenda, M., Chapman,
K. E. (2000). 5-heterogeneity of glucocorticoid receptor messenger RNA is tissue specific:
Differential regulation of variant transcripts by early-life events. Molecular Endocrinology, 14,
506517.
McGowan, P. O., Sasaki, A., Dymov, S., Turecki, G., Szyf, M., & Meaney, M. J. (2009). Epigenetic
regulation of the glucocorticoid receptor in human brain associates with childhood abuse.
McHughen, S. A., Rodriguez, P. F., Kleim, J. A., Kleim, E. D., Marchal Crespo, L., Procaccio, V.,
Cramer, S. C. (2010). BDNF val66met polymorphism influences motor system function in
the human brain. Cerebral Cortex, 20, 12541262.
Meaney, M. J. (2010). Epigenetics and the biological definition of gene environment interac-
tions. Child Development, 81, 4179.
Meaney, M. J., & Ferguson-Smith, A. (2010). Epigenomic regulation of the neural transcriptome:
The meaning of the marks. Nature Neuroscience, 13, 13131318.
Meaney, M. J., LeDoux, J. E. & Leibowitz, M. L. (2008). Neurobiology of anxiety disorders. In
Psychiatry (3rd ed., Vol. 1, pp. 317338).
Meaney, M. J., & Szyf, M. (2005). Maternal effects as a model for environmentally-dependent
chromatin plasticity. Trends in Neurosciences, 28, 456463.
Meaney, M. J., Szyf, M., & Seckl, J. R. (2007). Epigenetic mechanisms of perinatal programming
of hypothalamicpituitaryadrenal function and health. Trends in Molecular Medicine, 13,
269277.
Meaney, M. J. D. J., Francis, D., Weaver, S., Yau, J., Chapman, K., & Seckl, J. R. (2000). Postnatal
handling increases the expression of cAMP-inducible transcription factors in the rat hippocam-
pus: The effects of thyroid hormones and serotonin. The Journal of Neuroscience, 20,
39263935.
90 M.J. Meaney
Mehta, D., Klengel, T., Conneely, K. N., Smith, A. K., Altmann, A., Pace, T. W., Binder, E. B.
(2013). Childhood maltreatment is associated with distinct genomic and epigenetic profiles in
posttraumatic stress disorder. Proceedings of the National Academy of Sciences, 110,
83028307.
Melas, P. A., Wei, Y., Wong, C. C., Sjoholm, L. K., berg, E., Mill, J., Lavebratt, C. (2013).
Genetic and epigenetic associations of MAOA and NR3C1 with depression and childhood
adversities. The International Journal of Neuropsychopharmacology, 16, 15131528.
Melln, M., Ayata, P., Dewell, S., Kriaucionis, S., & Heintz, N. (2012). MeCP2 binds to 5hmC
enriched within active genes and accessible chromatin in the nervous system. Cell, 151,
14171430.
Menard, J., Champagne, D. L., & Meaney, M. J. (2004). Variations of maternal care differentially
influence fear reactivity and regional patterns of cFos immunoreactivity in response to the
shock-probe burying test. Neuroscience, 129, 297308.
Meyer-Lindenberg, A., & Weinberger, D. R. (2006). Intermediate phenotypes and genetic mecha-
nisms of psychiatric disorders. Nature Reviews. Neuroscience, 7, 818827.
Mohn, F., Weber, M., Rebhan, M., Roloff, T. C., Richter, J., Stadler, M. B., Schubeler, D.
(2008). Lineage-specific polycomb targets de novo DNA methylation define restriction and
potential of neuronal progenitors. Molecular Cell, 30, 755766.
Montag, C., Basten, U., Stelzel, C., Fiebach, C. J., & Reuter, M. (2010). The BDNF Val66Met
polymorphism and anxiety: Support for animal knock-in studies from a genetic association
study in humans. Psychiatry Research, 179, 8690.
Montag, C., Reuter, M., Newport, B., Elger, C., & Weber, B. (2008). The BDNF Val66Met poly-
morphism affects amygdala activity in response to emotional stimuli: Evidence from a genetic
imaging study. NeuroImage, 42, 15541559.
Murayama, A., Sakura, K., Nakama, M., Yasuzawa-Tanaka, K., Fujita, E., Tateishi, Y.,
Yanagisawa, J. (2006). A specific CpG site demethylation in the human interleukin 2 gene
promoter is an epigenetic memory. The EMBO Journal, 25, 10811092.
Murgatroyd, C. P. A., Wu, Y., Micale, V., Bockmhl, Y., Fischer, D., Holsboer, F., Spengler, D.
(2009). Dynamic DNA methylation programs persistent adverse effects of early-life stress.
Nanni, V., Uher, R., & Danese, A. (2012). Childhood maltreatment predicts unfavorable course of
illness and treatment outcome in depression: A met-analysis. The American Journal of
Naumova, O. Y., Lee, M., Koposov, R., Szyf, M., Dozier, M., & Grigorenko, E. L. (2012).
Differential patterns of whole-genome DNA methylation in institutionalized children and chil-
dren raised by their biological parents. Development and Psychopathology, 24, 143155.
Nelson, C. A., Zeanah, C. H., Fox, N. A., Marshall, P. J., Smyke, A. T., & Guthrie, D. (2007).
Cognitive recovery in socially deprived young children: The bucharest early intervention
project. Science, 318, 19371940.
Nikolova, Y., Koenen, K. C., Galea, S., Wang, C.-M., Seney, M. L., Sibille, E., Hariri, A. R.
(2014). Beyond genotype: Serotonin transporter epigenetic modification predicts human brain
function. Nature Neuroscience, 17, 11531155.
Ooi, S. K., Qiu, C., Bernstein, E., Li, K., Jia, D., Yang, Z., Bestor, T. H. (2007). DNMT3L con-
nects unmethylated lysine 4 of histone H3 to de novo methylation of DNA. Nature, 448,
714717.
Paulus, M. P., Rogalsky, C., Simmons, A., Feinstein, J. S., & Stein, M. B. (2003). Increased activa-
tion in the right insula during risk-taking decision making is related to harm avoidance and
neuroticism. NeuroImage, 19, 14391448.
Perroud, N., Dayer, A., Piguet, C., Nallet, A., Favre, S., Malafosse, A., Aubry, J. M. (2014).
Childhood maltreatment and methylation of the glucocorticoid receptor gene NR3C1 in bipolar
disorder. The British Journal of Psychiatry, 204, 3035.
Perroud, N., Paoloni-Giacobino, A., Prada, P., Oli, E., Salzmann, A., Nicastro, R., Malafosse,
A. (2011). Increased methylation of glucocorticoid receptor gene (NR3C1) in adults with a
history of childhood maltreatment: A link with the severity and type of trauma. Translational
Psychiatry, 1, e59.
Petryshen, T. L., Sabeti, P. C., Aldinger, K. A., Fry, B., Fan, J. B., Schaffner, S. F., Sklar, P.
(2010). Population genetic study of the brain-derived neurotrophic factor (BDNF) gene.
Molecular Psychiatry, 15, 810815.
Plotsky, P. M., Thrivikraman, K. V., Nemeroff, C. B., Caldji, C., Sharma, S., & Meaney, M. J.
(2005). Long-term consequences of neonatal rearing on central corticotropin-releasing factor
systems in adult male rat offspring. Neuropsychopharmacology, 30, 21922204.
Pluess, M., & Belsky, J. (2011). Prenatal programming of postnatal plasticity? Development and
Pluess, M., Velders, F. P., Belsky, J., Van IJzendoorn, M. H., Bakermans-Kranenburg, M. J., Jaddoe,
V. W., Tiemeier, H. (2011). Serotonin transporter polymorphism moderates effects of prena-
tal maternal anxiety on infant negative emotionality. Biological Psychiatry, 69, 520505.
Qiu, A., Rifkin-Graboi, A., Chen, H., Chong, Y. S., Kwek, K., Gluckman, P. D., Meaney, M. J.
(2013). Maternal anxiety and infants hippocampal development: Timing matters. Translational
Psychiatry, 3, e306.
Radtke, K. M., Ruf, M., Gunter, H. M., Dohrmann, K., Schauer, M., Elbert, T. (2011).
Transgenerational impact of intimate partner violence on methylation in the promoter of the
glucocorticoid receptor. Translational Psychiatry, 1, e21.
Rauch, S. L., Savage, C. R., Alpert, N. M., Fischman, A. J., & Jenike, M. A. (1997). The functional
neuroanatomy of anxiety: A study of three disorders using positron emission tomography and
symptom provocation. Biological Psychiatry, 42, 446452.
Razin, A., & Riggs, A. D. (1980). DNA methylation and gene function. Science, 210, 604610.
Rice, C. J., Sandman, C. A., Lenjavi, M. R., & Baram, T. Z. (2008). A novel mouse model for acute
and long-lasting consequences of early life stress. Endocrinology, 149, 48924900.
Rifkin-Graboi, A., Bai, J., Chen, H., Hameed, W. B., Sim, L. W., Tint, M. T., Qiu, A. (2013).
Prenatal maternal depression associates with microstructure of right Amygdala in neonates at
birth. Biological Psychiatry, 74, 837844.
Risch, N., Herrell, R., Lehner, T., Liang, K. Y., Eaves, L., Hoh, J., Merikangas, K. R. (2009).
Interaction between the serotonin transporter gene (5-HTTLPR), stressful life events, and risk
of depression: A meta-analysis. JAMA, 301, 24622471.
Romens, S. E., McDonald, J., Svaren, J., & Pollak, S. D. (2015). Associations between early life
stress and gene methylation in children. Child Development, 86, 303309.
Ruthenburg, A. J., Allis, C. D., & Wysocka, J. (2007). Methylation of lysine 4 on histone H3:
Intricacy of writing and reading a single epigenetic mark. Molecular Cell, 25, 1530.
Rutter, M., Kumsta, R., Schlotz, W., & Sonuga-Barke, E. (2012). Longitudinal studies using a
natural experiment design: The case of adoptees from romanian institutions. Journal of the
American Academy of Child and Adolescent Psychiatry, 51, 762770.
Ruzov, A., Tsenkina, Y., Serio, A., Dudnakova, T., Fletcher, J., Bai, Y., De Sousa, P. (2011).
Lineage-specific distribution of high levels of genomic 5-hydroxymethylcytosine in mamma-
lian development. Cell Research, 21, 13321342.
Rybakowski, J. K. (2008). BDNF gene: Functional Val66Met polymorphism in mood disorders
and schizophrenia. Pharmacogenomics, 9, 15891593.
Schanberg, S. M., Evoniuk, G., & Kuhn, C. M. (1984). Tactile and nutritional aspects of maternal
care: Specific regulators of neuroendocrine function and cellular development. Proceedings of
the Society for Experimental Biology and Medicine, 175, 135146.
Schubeler, D. (2012). Epigenetic islands in a genetic ocean. Science, 338, 756757.
Seckl, J. R., & Holmes, M. C. (2007). Mechanisms of disease: Glucocorticoids, their placental
metabolism and fetal programming of adult pathophysiology. Nature Clinical Practice.
Endocrinology & Metabolism, 3, 479488.
Shahbazian, M. D., & Grunstein, M. (2007). Functions of site-specific histone acetylation and
deacetylation. Annual Review of Biochemistry, 76, 75100.
Shalev, I., Lerer, E., Israel, S., Uzefovsky, F., Gritsenko, I., Mankuta, D., Kaitz, M. (2009).
BDNF Val66Met polymorphism is associated with HPA axis reactivity to psychological stress
characterized by genotype and gender interactions. Psychoneuroendocrinology, 34, 382388.
Shenker, N., & Flanagan, J. M. (2012). Intragenic DNA methylation: Implications of this epigen-
etic mechanism for cancer research. British Journal of Cancer, 106, 248253.
92 M.J. Meaney
Sokolowski, M. B., & Wahlsten, D. (2001). Geneenvironment interaction and complex behavior.
In S. O. Moldin (Ed.), Methods in genomic neuroscience (pp. 327). Boca Raton, FL: CRC.
Stadler, M. B., Murr, R., Burger, L., Ivanek, R., Lienert, F., Schler, A., Schbeler, D. (2011).
DNA-binding factors shape the mouse methylome at distal regulatory regions. Nature, 480,
490495.
Stein, M. B., Simmons, A. N., Feinstein, J. S., & Paulus, M. P. (2007). Increased amygdala and
insula activation during emotion processing in anxiety-prone subjects. The American Journal
of Psychiatry, 164, 318327.
Suzuki, A., Matsumoto, Y., Shibuya, N., Ryoichi, S., Kamata, M., Enokido, M., Otani, K.
(2012). Interaction effect between the BDNF Val66Met polymorphism and parental rearing for
interpersonal sensitivity in healthy subjects. Psychiatry Research, 2, 945948.
Sweatt, J. D. (2009). Experience-dependent epigenetic modifications in the CNS. Biological
Szulwach, K. E., Li, X., Li, Y., Song, C.-X., Wu, H., Dai, Q., Jin, P. (2011). 5-hmC-mediated
epigenetic dynamics during postnatal neurodevelopment and aging. Nature Neuroscience, 14,
16071616.
Tahiliani, M. K. K., Shen, Y., Pastor, W. A., Bandukwala, H., Brudno, Y., Agarwal, S., Aravind,
L. (2009). Conversion of 5-methylcytosine to 5-hydroxymethylcytosine in mammalian DNA
by MLL partner TET1. Science, 324, 930935.
Taverna, S. D., Ueberheide, B. M., Liu, Y., Tackett, A. J., Diaz, R. L., Shabanowitz, J., Allis, C.
D. (2007). Long-distance combinatorial linkage between methylation and acetylation on his-
tone H3 N termini. Proceedings of the National Academy of Sciences, 104, 20862091.
Taylor, R. C., Sanfilippo, A., McDermott, J. E., Baddeley, B., Riensche, R., Jensen, R.,
Pustejovsky, J. (2010). Enriching regulatory networks by bootstrap learning using optimised
GO-based gene similarity and gene links mined from PubMed abstracts. International Journal
of Computational Biology and Drug Design, 4(1), 5682.
Teh, A. L., Pan, H., Chen, L., Ong, M. L., Dogra, S., Wong, J., Holbrook, J. D. (2014). The
effect of genotype and in utero environment on inter-individual variation in neonate DNA
methylomes. Genome Research, 24, 10641074.
Thomas, D. (2010). Geneenvironment-wide association studies: Emerging approaches. Nature
Reviews Genetics, 11, 260272.
Thomson, J. P., Skene, P. J., Selfridge, J., Clouaire, T., Guy, J., Webb, S., Bird, A. (2010). CpG
islands influence chromatin structure via the CpG-binding protein Cfp1. Nature, 464,
10821086.
Toki, S. M. S., Imanaka, A., Yamamoto, S., Yamawaki, S., & Honma, K. (2007). Importance of
early lighting conditions in maternal care by dam as well as anxiety and memory later in life of
offspring. European Journal of Neuroscience, 25, 815829.
Turecki, G., & Meaney, M. J. (2016). Effects of the social environment and stress on glucocorti-
coid receptor gene methylation: A systematic review. Biological Psychiatry, 79, 8796.
Turkheimer, E., Haley, A., Waldron, M., DOnofrio, B., & Gottesman, I. I. (2003). Socioeconomic
status modifies heritability of IQ in young children. Psychological Science, 14, 623628.
Turner, B. M. (2001). Chromatin and gene regulation: Mechanisms in epigenetics. Cambridge,
MA: Blackwell Science.
Turner, J. D., & Muller, C. P. (2005). Structure of the glucocorticoid receptor (NR3C1) gene 5
untranslated region: Identification, and tissue distribution of multiple new human exon 1.
Journal of Molecular Endocrinology, 35, 283292.
Tyrka, A. R., Price, L. H., Marsit, C., Walters, O. C., & Carpenter, L. L. (2012). Childhood adver-
sity and epigenetic modulation of the leukocyte glucocorticoid receptor: Preliminary findings
in healthy adults. PLoS One, 7(1), e30148.
Uher, R., & McGuffin, P. (2008). The moderation by the serotonin transporter gene of environmen-
tal adversity in the aetiology of mental illness: Review and methodological analysis. Molecular
Psychiatry, 13(2), 131146.
Uher, R., & McGuffin, P. (2010). The moderation by the serotonin transporter gene of environmental
adversity in the etiology of depression: 2009 update. Molecular Psychiatry, 15, 1822.
Ursini, G., Bollati, V., Fazio, L., Porcelli, A., Iacovelli, L., Catalani, A., Bertolino, A. (2011).
Stress-related methylation of the catechol-O-methyltransferase Val158 allele predicts human
prefrontal cognition and activity. Journal of Neuroscience, 31, 66926698.
Van Hasselt, F. N., Cornelisse, S., Zhang, T. Y., Meaney, M. J., Velzing, E. H., Krugers, H. J., &
Jols, M. (2012). Adult hippocampal glucocorticoid receptor expression and dentate synaptic
plasticity correlate with maternal care received by individuals early in life. Hippocampus, 22,
255266.
Vijayendran, M., Beach, S., Plume, J. M., Brody, G., & Philibert, R. (2012). Effects of genotype
and child abuse on DNA methylation and gene expression at the serotonin transporter. Frontiers
in Psychiatry, 3.
Wagner, S., Baskaya, O., Dahmen, N., Lieb, K., & Tadic, A. (2010). Modulatory role of the brain-
derived neurotrophic factor Val66Met polymorphism on the effects of serious life events on
impulsive aggression in borderline personality disorder. Genes, Brain, and Behavior, 9,
97102.
Wang, C., Zhang, Y., Liu, B., Long, H., Yu, C., & Jiang, T. (2014). Dosage effects of BDNF
Val66Met polymorphism on cortical surface area and functional connectivity. The Journal of
Wang, K., Li, M., & Bucan, M. (2007). Pathway-based approaches for analysis of genomewide
association studies. American Journal of Human Genetics, 81, 12781283.
Waterland, R. A., & Jirtle, R. L. (2003). Transposable elements: Targets for early nutritional effects
on epigenetic gene regulation. Molecular and Cellular Biology, 23, 52935300.
Waterland, R. A., Lin, J. R., Smith, C. A., & Jirtle, R. L. (2006). Post-weaning diet affects genomic
imprinting at the insulin-like growth factor 2 (Igf2) locus. Human Molecular Genetics, 15,
705716.
Weaver, I. C., Cervoni, N., Champagne, F. A., DAlessio, A. C., Sharma, S., Seckl, J. R.,
Meaney, M. J. (2004). Epigenetic programming by maternal behavior. Nature Neuroscience, 7,
847854.
Weaver, I. C., DAlessio, A. C., Brown, S. E., Hellstrom, I. C., Dymov, S., Sharma, S., Meaney,
M. J. (2007). The transcription factor nerve growth factor-inducible protein a mediates epigen-
etic programming: Altering epigenetic marks by immediate-early genes. Journal of
Weaver, I. C., Meaney, M. J., & Szyf, M. (2006). Maternal care effects on the hippocampal tran-
scriptome and anxiety-mediated behaviors in the offspring that are reversible in adulthood.
Proceedings of the National Academy of Sciences of the United States of America, 103,
34803485.
Wendland, J. R., Kruse, M. R., Cromer, K. R., & Murphy, D. L. (2007). A large case-control study
of common functional SLC6A4 and BDNF variants in obsessive-compulsive disorder.
Neuropsychopharmacology, 32, 25432551.
Whitelaw, N. C., & Whitelaw, E. (2006). How lifetimes shape epigenotype within and across gen-
erations. Human Molecular Genetics, 15, R131R137.
Wichers, M., Kenis, G., Jacobs, N., Mengelers, R., Derom, C., Vlietinck, R., van Os, J. (2008).
The BDNF Val(66)Met 5-HTTLPR child adversity interaction and depressive symptoms:
An attempt at replication. American Journal of Medical Genetics. Part B, Neuropsychiatric
Genetics, 147B, 120123.
Wright, C. I., Martis, B., McMullin, K., Shin, L. M., & Rauch, S. L. (2003). Amygdala and insular
responses to emotionally valenced human faces in small animal specific phobia. Biological
Wu, Y. P. A., Daniel, G., Almeida, O. F., & Spengler, D. (2014). Early-life stress reduces DNA
methylation of the Pomc gene in male mice. Endocrinology, 155, 17511762.
Xie, P., Kranzler, H. R., Poling, J., Stein, M. B., Anton, R. F., Farrer, L. A., & Gelernter, J. (2010).
Interaction of FKBP5 with childhood adversity on risk for post-traumatic stress disorder.
Neuropsychopharmacology, 35, 16841692.
94 M.J. Meaney
Yehuda, R., Daskalakis, N. P., Desarnaud, F., Makotkine, I., Lehrner, A. L., Koch, E., Bierer, L.
M. (2013). Epigenetic biomarkers as predictors and correlates of symptom improvement
following psychotherapy in combat veterans with PTSD. Frontiers in Psychiatry, 4, 118.
Yehuda, R., Daskalakis, N. P., Lehrner, A., Desarnaud, F., Bader, H. N., Makotkine, I., & Meaney,
M. J. (2014). Influences of maternal and paternal PTSD on epigenetic regulation of the gluco-
corticoid receptor gene in Holocaust survivor offspring. The American Journal of Psychiatry,
171, 872880.
Yehuda, R., Flory, J. D., Bierer, L. M., Henn-Haase, C., Lehrner, A., Desarnaud, F., Meaney, M.
J. (2014). Lower methylation of glucocorticoid receptor gene promoter 1F in peripheral blood
of veterans suffering from post-traumatic stress disorder. Biological Psychiatry, 77, 356364.
Zhang, D., Cheng, L., Badner, J. A., Chen, C., Chen, Q., Luo, W., Liu, C. (2010). Genetic con-
trol of individual differences in gene-specific methylation in human brain. American Journal of
Human Genetics, 86, 411419.
Zhang, T. Y., Hellstrom, I. C., Bagot, R. C., Wen, X., Diorio, J., & Meaney, M. J. (2010). Maternal
care and DNA methylation of the a glutamic acid decarboxylase 1 promoter in rat hippocam-
pus. Journal of Neuroscience, 30, 1313013107.
Zhang, T. Y., Labont, B., Wen, X. L., Turecki, G., & Meaney, M. J. (2013). Epigenetic mecha-
nisms for the early environmental regulation of hippocampal glucocorticoid receptor gene
expression in rodents and humans. Neuropsychopharmacology, 38, 111123.
Chapter 5
Using Genetically Informed Designs
toUnderstand theEnvironment:
TheImportance ofFamily-Based Approaches
ChangLiu andJenaeM.Neiderhiser
There is a vast literature examining how rearing environmental factors, like parenting,
influence child development (e.g., Demo & Cox, 2000; Fletcher, Steinberg &
Williams-Wheeler, 2004). Until recently, few of these studies considered the pos-
sible role of genetic influences on the links between rearing environment and child
adjustment. Over the past few decades, however, many studies have reported that
both childrens and parents genes influence the childs rearing environment and
account, in part, for the associations between parenting and child behavioral out-
comes (e.g., Horwitz & Neiderhiser, 2015). This research has helped to provide new
insight into the important role of the child in helping to shape their own rearing
environment and has clarified when the effects of the rearing environment are purely
causal. Despite the critical role that family-based behavioral genetic studies have
played in understanding how rearing environment influences child outcomes, there
are some who have asked whether there is a continued need for family-based behav-
ioral genetic studies, especially when it is now possible to genotype individuals for
a relatively low cost (e.g., Charney, 2012). This chapter will first briefly describe the
logic of family-based behavioral genetic designs. This is followed by a summary of
the criticisms raised about family-based behavioral genetic designs, especially
about twin studies. Then, we directly address the concerns raised and clarify the
continued need for family-based behavioral genetic research. Finally, we discuss
how family-based behavioral genetic research informs our understanding of the
environment and can help guide research on intervention.
C. Liu J.M. Neiderhiser (*)

Department of Psychology, The Pennsylvania State University,
431 Moore Building, University Park, PA 16802-3106, USA
e-mail: CXL492@psu.edu; jenaemn@psu.edu

DOI10.1007/978-3-319-49227-8_5
96 C. Liu and J.M. Neiderhiser
Family-Based Behavioral Genetic Designs
By taking advantage of the natural variation in genetic relatedness among family

members, behavioral genetic designs attempt to disentangle genetic and environ-
mental contributions to behavior (see Plomin, DeFries, Knopik & Neiderhiser, 2013,
for a detailed discussion of behavioral genetic designs). The most commonly used
designs are twin studies, which examine identical (monozygotic: MZ) and fraternal
(dizygotic: DZ) twin pairs. For the purposes of behavioral genetic studies, MZ twins
are 100% genetically similar and DZ twins are 50% on average. Additional sibling
types can also be included, such as full siblings, who share 50% of their genes on
average, half siblings (25%), and genetically unrelated or adoptive siblings, sharing
none of their genes. By examining sibling types with different levels of genetic relat-
edness, genetic influences can be inferred based on twin and sibling similarity.
Behavioral genetic studies also examine environmental influences, including shared
environmentalnongenetic influences that account for sibling similarityand non-
shared environmentalnongenetic influences that account for sibling differences.
Thus, when twin or sibling pairs are reared in the same household genetic and envi-
ronmental influences can be estimated as follows, using just MZ and DZ twins as an
example. If MZ twins are twice as similar as DZ twins that suggests that heritable
influences are important for that construct. Any deviation of MZ twin similarity
from 1.0 is a direct indication of nonshared environmental influences. Finally, shared
environmental influences are indicated when DZ twins are more than half as similar
as MZ twins. The twin-only design is particularly powerful for detecting heritable
influences. Most twin studies that examine twins reared together from birth are less
powerful for detecting shared environmental influences, as there is no systematic
variation across twin types in potential shared environmental influences.
Other family-based designs used in behavioral genetics include parentoffspring
adoption designs and combinations of twin, family, and adoption designs. Parent
offspring adoption designs include children adopted at birth, their adoptive parents,
and their biological parents. Because the biological parents provide only their genes,
and in the case of the biological mother the prenatal environment, links between the
adopted child and their biological parent(s) are best explained as heritable. In con-
trast, the adoptive parents are genetically unrelated to the adopted child but provide
the rearing environment. Thus, links between the adopted child and their adoptive
parents are best explained as shared environmental. Combination designs include
the sibling adoption design, where two genetically unrelated siblings are reared
together; twin/adoption designs, where twins are separated and parented in different
households; children of twins designs, where children of adult twin parents are
examined with the twins; and various combinations and expansions of these designs.
Most noteworthy are sibling adoption designs and twin/adoption designs as these
designs, and parentoffspring adoption designs, complement twin studies by pro-
viding more power to estimate the effects of shared environment.
The role of family-based behavioral genetic designs for understanding the envi-
ronment. There are many ways that these studies can be used to advance our under-
standing of the critical role of the environment in shaping childrens behavioral
5 Genetically Informed Designs 97
outcomes. The ability to control for heritable influences is the most basic.
Specifically, if environmental influences are found even after heritable influences
are accounted for this provides strong support for the role of environmental factors.
This is different than finding effects of parenting on child behavioral outcomes in a
study of parents rearing their biological child(ren). Parents and children share genes
as well as environments, thus associations between parenting behavior and child
outcomes may be due to both heritable and environmental effects. By using designs
that allow genetic and environmental effects to be estimated this potential confound
can be accounted for.
Challenges toFamily-Based Behavioral Genetic Designs
Generally, two different but related types of concerns have been raised about
family-based behavioral genetic designs. The first set of criticisms is conceptual.
Specifically, there is some question about whether family-based behavioral genetic
designs are actually genetically informative, and therefore about whether they
provide information about genetic and environmental influences. Second, the meth-
odological strategies typically used in family-based behavioral genetic designs to
disentangle genetic and environmental influences have been questioned, especially
twin studies. We summarize these issues later.
Conceptual Issues
Genes and environments must make independent contributions to a behavioral phe-

notype in order for them to be disentangled using family-based behavioral genetic
designs. Some have argued that genes and environments work together in the devel-
opment of organisms in a way that makes it nearly impossible to disentangle them
(Burt & Simons, 2014; Gottlieb, 1998). This theoretical perspective then supports
the notion that genetically informed studies and designs are not actually genetically
informative (Partridge, 2005), because the contribution of genes and environments
on the organism cannot be quantitatively specified (Lerner, 2004; Lickliter, 2012),
thus making the partitioning of individual differences inaccurate.
A different position states that because estimating heritability does not explain
how exactly genes influence human behaviors it is not informative in human popu-
lations (Crusio, 2012). Specifically, in biometric models, heritability estimates do
not actually measure genes but are instead estimated based on the variance and
covariance of family members within families (Burt & Simons, 2014; Partridge,
2005). If a population is homogenous for genetic variability, the heritability estimate
is then zero. Similarly, if a population is largely heterogeneous for genetic variabil-
ity, the heritability estimate is relatively high. The magnitude of the heritability
estimated is largely dependent on the specific context and is variable from context
to context (or from population to population). Given the population specificity of

the heritability estimate, as it is a population-based statistic, estimating heritable
and environmental contributions to the total variance in a population provides little
information on the causal effects of genes and environments on complex human
behaviors (Turkheimer, 2011).
Finally, the third conceptual case often made for family-based behavioral genetic
designs is that the estimates of heritable and environmental influences do not reflect
the physical reality. In other words, genome sequence similarities among twin pairs,
as reflected by twin similarities, do not necessarily indicate the structural similari-
ties between proteins and gene functions among twins. Proponents of this position
state that the relationship between genotype and phenotype is not one to one, as
would be suggested by a typical twin design, but is instead ambiguous and does not
systematically vary as a function of twin genetic similarity (Partridge, 2005). The
reason for the ambiguous relationship between genotype and phenotype is that gene
expression is influenced by extracellular factors, such as behavioral and environ-
mental inputs, characterized by dynamic and integrative processes. This nonsystem-
atic variation in genotype as a function of various external factors means that
knowledge of the genome sequence itself provides little developmental information
about human behaviors.
Technical Issues
In addition to the conceptual issues described earlier, there are also more technical
issues, most related to the assumptions of family-based behavioral genetic designs.
Nonrandom mating, or assortative mating, is an assumption that has been frequently
raised as untenable for several decades. If individuals are more likely to choose a
partner similar to themselves, as is argued by critics of the assortative mating
assumption, then this assumption is not valid (Collins & Coltrane, 1985; Simons,
Stewart, Gordon & Conger, 2002). Violation of the assortative mating assumption
would result in an underestimate of heritability, however, because the genetic cor-
relation between DZ twin correlations would be greater than .50, thus decreasing
the difference between MZ and DZ twin pairs. It is important to note, however, that
the assortative mating assumption (or any other assumption) is only relevant for the
variables under study. In other words, if there is assortative mating for political
affiliation, but political affiliation is unrelated to the construct under study, the
impact on the findings would be minimal.
The notion of genetic additivity is another concept that has been criticized for
decades. Critics have argued that the genetic influences do not operate in an additive
fashion because of genetic interdependenciesthe effect of one allele is dependent
on the effect of other alleles (nonadditive genetic effects: dominance and epistasis).
The only relatives that are identical for nonadditive genetic effects are MZ twins,
while DZ twins share 25% of nonadditive genetic effects on average. Thus, when
genetic effects are partially nonadditive, the genetic difference between MZ and DZ
twin pairs is larger than expected 0.50, contributing to the overestimation of additive
genetic influences, at the expense of nonadditive genetic influences and shared
environmental influences (Grayson, 1989).
The equal environments assumption (EEA) has also been the focus of many criti-
cisms of family-based behavioral genetic designs, especially twin designs. The
EEA states that MZ and DZ twins do not experience systematically different envi-
ronments (Plomin, DeFries, Knopik & Neiderhiser, 2013). However, critics point
out that MZ twins have more similar environments compared to DZ twins (Burt &
Simons, 2014; Joseph, 2004; Rose, Lewontin & Kamin, 1984). In other words, MZ
twins are treated more similarly than DZ twins, because MZ twins look more alike
than DZ twins. There are, however, few empirical studies that support these argu-
ments. For example, one study found that MZ twins shared more socially-based
characteristics than DZ twins, like time spent together, shared friends, and physical
attractiveness (Horwitz, Videon, Schmitz & Davis, 2003). It should be noted that,
like the other assumptions, violation of the EEA is only a concern if the more simi-
lar environments of MZ twins than DZ twins are etiologically relevant to the
variables under study (trait-relevant EEA, Gottesman & Shields, 1966; Kendler,
Neale, Kessler, Heath & Eaves, 1993).
More recent concerns about the assumptions of family-based behavioral genetic
designs (mainly twin studies) have been based on findings from epigenetic studies
(Charney, 2012; Ho, 2013; Lerner, 2004; Lickliter, 2012) resulting in some calling
for the end of heritability estimates (Burt & Simons, 2014). These criticisms are
briefly summarized later. First, there is a claim that MZ twins are not genetically
identical, because their genomes differ in many ways (Charney, 2012; Ollikainen
etal., 2010). Thus, the larger phenotypic concordance among MZ twins compared
to DZ twins cannot be confidently ascribed to genetic concordance, leading to unre-
liable estimates of heritability and environmental influences in family-based behav-
ioral genetic designs. Second, proponents of this perspective state that the genetic
similarity of MZ and DZ twin pairs is not fixed but varies over time. The changes in
genetic overlap among MZ and DZ twin pairs over time are influenced by differ-
ences in the environmental and behavioral inputs. In other words, life experiences
cause genetic changes, and these changes occur differently for each member of the
twin pair, resulting in changes in the genetic similarity of the pairs over time. Thus,
the differences in genetic overlap within each twin pair are continuously increasing
as they age (Charney, 2012; Fraga etal., 2005). Finally, some have stated that pre-
natally, most DZ twins and singletons experience less stress than MZ twins, espe-
cially for those who share a single chorion. The greater concordance in prenatal
stress for MZ twins leads to greater concordance in behavioral phenotypes (Charney,
2012). Furthermore, critics adopt this assertion to support their notion that the EEA
is not held in the prenatal environment and the confounding effect due to prenatal
environments is not eliminated in the family-based behavioral genetic designs.
However, as noted before, the trait-relevant EEA that environments are only r elevant
to phenotypes under study should be applied here. In summary, these critics con-
clude that the violations of these assumptions may render the family-based behav-
ioral genetic designs unreliable and problematic, although they do not directly state
the expected direction of these effects as overestimating or underestimating herita-

bility and environmental influences.
he Continued Value ofFamily-Based Behavioral Genetic

T
Designs
Response toConceptual Criticisms
Studies using family-based behavioral genetic designs recognize the importance of

both genes and environments and many studies systematically study how genes and
environments work together sometimes referred to as geneenvironment inter-
play (Rutter, Moffitt & Caspi, 2006). What matters, however, for family-based
behavioral genetic studies is whether the total variance can be separated into genetic
and environmental components (Wright etal., 2015). The answer to this question is
yes and is supported in a large body of family-based behavioral genetic studies. As
Duncan (2014) notes, at the population level, disentangling genetic and environ-
mental sources of variance, and modeling GxEs in family-based behavioral genetic
studies are appropriate and applicable.
Although it is clear that the DNA sequence does not fully determine epigenetic
processes and gene expression, this does not mean we should abandon the long
existing family-based behavioral genetic approaches (McGue, Elkins, Walden &
Iacono, 2005) and embrace epigenetic studies. DNA sequence variations are impor-
tance and will continue to be important. These systematic variations explain why
MZ twins are more similar in complex traits than DZ twins in almost every twin
study, why children resemble their parents on many characteristics, and why child
genetically influenced characteristics can evoke parental behaviors (e.g., Polderman
etal., 2015; Horwitz & Neiderhiser, 2015). In addition, at least one study has found
that variation in DNA methylation is partially influenced by genetic factors (Heij
mans, Kremer, Tobi, Boomsma & Slagboom, 2007). Monks etal. (2004) reported
heritability estimate of differentially expressed genes with a median heritability of
.34 (ranging from .18 to 1.00), indicating the importance of genetic control in gene
expression. Both of these studies indicate that family-based behavioral genetic
designs continue to have an important role in helping to understand how both
genetic and environmental factors may influence variation in human populations.
It is also important to note that the findings from epigenetic studies on human
phenotypes are not entirely clear and some are contrary to empirical findings using
different strategies (Battaglia, 2012). For example, proponents of an epigenetic
approach would expect that the environmental and behavioral inputs over time
would contribute to increasing variation in gene expression, thus decreasing the
phenotypic concordance among MZ and DZ twin pairs. However, this prediction is
not consistent with the finding that for intelligence, the degree of similarity among
MZ twin pairs increases throughout the life span and the heritability of intelligence
continues to increase linearly with age (e.g., Deary, Spinath & Bates, 2006; Haworth
etal., 2010; Tucker-Drob & Briley, 2014). This inconsistency suggests that the pre-
diction deriving from epigenetic studies may not hold as a general principle across
human behaviors. In addition, findings from epigenetic studies are still controversial
and inconsistent, especially in humans (Battaglia, 2012; Moffitt & Beckley, 2015).
In sum, we wish to emphasize that family-based behavioral genetic approaches are
a promising way to study complex epigenetic effects and gene expression.
Response toTechnical Criticisms
A number of studies have addressed the technical criticisms by carefully examining

the adequacy of the assumptions that family-based behavioral genetic studies
(mainly twin studies) are based on and estimating the degree of bias when violating
these assumptions. Here we review some of the findings from these studies.
First, evidence suggests the existence of nonrandom mating across a variety of
physical and psychological characteristics. However, the correlation of physical
characteristics (e.g., height and weight) between spouses is relative low, around
0.20 and even lower for personality traits (Spuhler, 1968; Vandenberg, 1972). The
nonrandom mating for cognitive ability, education level, and externalizing prob-
lems (e.g., antisocial behavior, drug dependence) is moderate to high, and the
correlation between spouses is between 0.25 and 0.50 (Domingue, Fletcher, Conley
& Boardman, 2014; Jensen, 1978; Krueger, Moffitt, Caspi, Bleske & Silva, 1998;
Mare, 1991). Barnes and Boutwell (2013) have estimated the degree of bias when
the correlation between spouses ranges from .25 to .50 and concluded that nonran-
dom mating deflates heritability estimates and inflates shared environment esti-
mates. However, the degree of bias is not large, smaller than 10 percentage points
in their estimates. There are also a number of papers that have directly modeled
assortative mating. Although the effects are sometimes significant, the impact on
the estimates of genetic and environmental influences is minimal (e.g., Reynolds
etal., 2006).
Second, additive genetic influences are important in family-based behavioral
genetic studies, because they represent the resemblance between parents and their
offspring (Plomin etal., 2013). Evidence from human and animal genetic studies
across a wide range of traits suggests that additive genetic variance accounts for
80% of genetic variance on average and often explains close to 100% of the total
genetic variance (Hill, Goddard & Visscher, 2008). In other words, although domi-
nance and epistasis may operate in nonadditive ways to influence total genetic vari-
ance, in most cases simply estimating additive genetic variance will provide accurate
estimates of overall heritability. It is also possible to include nonadditive genetic
influences in models, although the patterns of twin and sibling correlations rarely
indicate that this is necessary.
Third, with regard to the validity of the EEA, Barnes and Boutwell (2013) have
reviewed the literature on the EEA, including empirical studies testing the validity
of EEA and a comprehensive overview of the validity of the EEA.They concluded
that 13 studies (21 percent) have claimed a violation of the EEA, although only 6 of
them tested the assumption. The few studies that have estimated the degree of bias
when violating the EEA found an average of one percentage point overestimate in
heritability. In contrast, most studies have found evidence supporting the validity of
the EEA (Borkenau, Riemann, Angleitner & Spinath, 2002; Cronk et al., 2002;
Hettema, Neale & Kendler, 1995; Kendler & Gardner, 1998; Kendler, Neale,
Kessler, Heath & Eaves, 1994; Klump, Holly, Iacono, McGue & Willson, 2000;
Plomin, Willerman & Loehlin, 1976; Scarr & Carter-Saltzman, 1979). More persua-
sive evidence derives from analyzing misclassified twin pairs. MZ twins are mis-
classified as DZ twins by their parents or by themselves and vice versa. Accumulative
empirical studies have shown that these misclassified twins are no more different
from the correctly classified twins, suggesting the validity of the EEA in family-
based behavioral genetic studies (Conley, Rauscher, Dawes, Magnusson & Siegal,
2013; Gunderson etal., 2006; Kendler etal., 1993).
Fourth, some findings have suggested that differences among the genomes of
MZ twin pairs are small and often negligible (Baranzini etal., 2010; van Dongen,
Slagboom, Draisma, Martin & Boomsma, 2012; Veenma etal., 2012). Most of the
DNA differences in somatic cells and in mitochondrial DNA among MZ twin pairs
are not related to heritability, because these DNA differences are acquired and not
in the inherited DNA (Miller, DeYoung & McGue, 2012). In the measured geno-
types, MZ twins are not very different (Miller etal., 2012). For example, one study
found that the genetic, epigenetic, or transcriptomic differences among MZ twins
did not explain disease discordance (Baranzini etal., 2010). For DZ twin pairs and
nontwin siblings from the same parents, studies have found the average of genetic
overlap (identity-by-descent sharing) is close to 50%, within the range of 4258%
(Visscher etal., 2007). In our own simulation study, we have examined the degree
of bias in parameter estimates when the assumption that MZ twins are genetically
identical fails. We have found when the assumption is violated, the nonshared envi-
ronmental influences are overestimated at the expense of additive genetic influ-
ences, while shared environmental influences remain the same. Although the
parameter estimates deviate from true parameter values, the bias is smaller than
10% points ((Liu, Molenaar & Neiderhiser, under review).
Finally, studies have tested the concern that the prenatal environment for MZ
twins is more stressful than that of DZ twins or singletons. The association between
the degree of prenatal sharing (such as sharing a chorion type) and similarities
between MZ twins among various traits, such as physical characteristics, cognitive
ability, temperaments, and problem behaviors have been examined in a few studies
(Jacobs etal., 2001; Riese, 1999; Van Beijsterveldt etal., 2015). The findings con-
sistently indicate limited influence of intrauterine prenatal environment on MZ twin
similarities. Similarly, a recent review of the effects of chorionicity on heritability
estimates in twin studies found that chorionicity effects led to both overestimates
and underestimates of heritability depending on the construct (Marceau etal., 2016).
One consistent finding reported in this review is that heritability estimates are
underestimated for birth weight and early growth when chorionicity is not taken
into account. Taken together these findings strongly support the validity of equal
environment assumption in the prenatal environments of twins.
Based on this review of the concerns and responses to these concerns, we conclude
that the assumptions of family-based behavioral genetic studies (mainly twin studies)
are robust and adequate. Although some assumptions may be violated, the effects of
these violations on heritability and environmental influence estimates are small and
have not been found to change the interpretation of the findings. Thus, these technical
criticisms do not invalidate the family-based behavioral genetic studies.
The Value ofFamily-Based Behavioral Genetic Studies
Until relatively recently, parenting was thought to have a direct environmental

impact on the development of children with no clear acknowledgment of the role of
the child. There have been exceptions to this unidirectional focus, with some advo-
cating a focus on bidirectional influences between children and parents (Bell, 1968;
Belsky, 1984; Russell & Russell, 1992), but it is not until the 1980s that the possible
role of childs or parents heritable influences has been considered as a possible
influence on parenting (e.g., Rowe, 1981, 1983; Scarr & McCartney, 1983). Recent
reviews of the literature examining heritable and environmental influences on chil-
dren and on their relationship with their parents have concluded that parenting of
children from infancy through adolescence is influenced by both child and parent
heritable influences, with important and substantial roles for both shared and non-
shared environmental influences (Chen & Deater-Deckard, 2015; Henry, Boivin &
Tarabulsy, 2015; Pike & Oliver, 2015; Klahr & Burt, 2014). In other words, through
the use of family-based behavioral genetic designs we have gained a more nuanced
view of the relationship between parents and children.
The majority of the studies reviewed earlier are twin studies, most often studies
of twin children. Other family-based behavioral genetic designs have examined the
role of heritable influences in parenting and child behavioral outcomes and have
drawn similar conclusions. For example, two studies using a parentoffspring adop-
tion design found that parenting behaviors during infancy and middle childhood
could be partly explained by heritable influences of the child (Elam etal., 2014;
Harold etal., 2013; Hajal etal., 2015). Similarly, a series of studies using the com-
bination designs of children-of-twins and extended children-of-twins have been
able to identify direct environmental parenting to child effects (e.g., Marceau etal.,
2015; Horwitz etal., 2015) as well as effects of child and parent heritable influences
on parenting behavior and adolescent outcome (e.g., Narusyte etal. 2008; Marceau,
Horwitz, etal., 2013; and see McAdams etal., 2014 for a review). This sizable and
growing body of literature helps to highlight the continued important role for
family-based behavioral genetic studies in helping to elucidate the mechanisms of
effect of parenting on child behavioral outcomes. Specifically, by clarifying when
parenting effects are directly environmental as compared to evoked by heritable

characteristics of the child, for example, it is possible to design strategies that may
be more effective for changing the behavioral outcomes of children.
he Implications ofFamily-Based Behavioral Genetic Studies

T
onIntervention Science
Only recently has research in intervention attempted to incorporate findings from

genetically informed studies. There are also a handful of intervention studies that
have collected DNA from participants in an effort to identify how interventions may
moderate genetic risk (e.g., Brody etal., 2013; Vandenberg etal., 2016). We believe
that family-based behavioral genetic designs have a critical role in helping to guide
intervention research in ways that are distinct from candidate gene-based and other
molecular genetic designs. Specifically, family-based behavioral genetic studies
help to distinguish direct environmental effects from those that are correlated with
parents genes and childs genes (geneenvironment correlation: rGE) which can
help to identify which intervention strategies may be most effective. For example, if
parental monitoring is found to have a direct environmental effect on rule-breaking
behaviors in adolescents, intervention strategies that focus on increasing effective
monitoring strategies would be preferred over strategies that focus on changing the
childs behavior. Similarly, if the parental hostility is evoked by genetically influ-
enced characteristics of the child (e.g., irritability via evocative rGE), strategies that
combine parent-focused (teaching parents to respond to irritability in their child
without hostility) and child-focused (teaching children to manage their own behav-
ior in interpersonal interactions) may be most effective. It is important to note a
strong caveat to these suggestions, however. Rarely are findings from family-based
behavioral genetic studiesor any other kind of naturalistic study designas clear
as the two examples provided earlier. Most often the findings suggest that there is a
combination of genetic and environmental influences on both parent and child
behavior sometimes with different patterns of findings for mothers and fathers (e.g.,
Hajal et al., 2015; Neiderhiser et al., 2004, 2007). Therefore, it is premature to
attempt to directly apply findings from family-based behavioral genetic research to
intervention, although this work can directly inform intervention research (see Leve
etal., in press for a detailed discussion of these issues).
The most critical role of family-based behavioral genetic studies for prevention
science is the ability to describe the mechanisms involved in geneenvironment
interplay and to highlight the importance of individual differences. The vast major-
ity of intervention research focuses on mean differences between control and inter-
vention groups in an attempt to demonstrate causal effects of an intervention on
target behaviors although more recent research has also incorporated individual dif-
ferences in response to intervention (Dishion et al., 2014; Mohr et al., 2013).
Similarly, family-based behavioral genetic studies can help to clarify the mecha-
nisms of effect, thus clarifying where an intervention may be most effective. For
example, a series of studies using a parentoffspring adoption design have found

that prenatal risks mediate genetic influences on child risk behaviors suggesting that
changing the prenatal environment may also decrease the effects of genetic risks on
child behavioral outcomes ((Marceau et al., 2013); Neiderhiser etal., 2016).
Conclusions andFuture Directions
The purpose of this chapter was to clarify that family-based behavioral genetic
designs continue to have an important role to play in understanding child behavioral
outcomes, especially in regard to the effects of parenting on children. There are
some who have questioned the continued need for the specialized and difficult data
collection in family-based behavioral genetic samples given the availability of strat-
egies for genotyping and long-standing concerns about the validity and interpret-
ability of findings from such studies. We have addressed these concerns and provided
additional support for the unique information that family-based behavioral genetic
designs can provide and implications for intervention science. It is through the con-
tinued use of these designs, taking a variety and combination of approaches, along
with the rapidly emerging findings from molecular genetic studies, that we will gain
a more complete understanding of the complex interrelationships and mechanisms
of genes and environments that influence parenting and children.
References
Baranzini, S. E., Mudge, J., van Velkinburgh, J. C., Khankhanian, P., Khrebtukova, I., Miller,
N.A., Kim, R.W. (2010). Genome, epigenome and RNA sequences of monozygotic twins
discordant for multiple sclerosis. Nature, 464(7293), 13511356. doi:10.1038/nature08990.
Barnes, J., & Boutwell, B.B. (2013). A demonstration of the generalizability of twin-based research
on antisocial behavior. Behavior Genetics, 43(2), 120131. doi:10.1007/s10519-012-9580-8.
Battaglia, M. (2012). Is behavioral genetics too-big-to-knowscience? Behavioral and Brain
Sciences, 35(05), 360360. doi:10.1017/S0140525X12001331.
Bell, R. Q. (1968). A reinterpretation of the direction of effects in studies of socialization.
Psychological Review, 75(2), 8195. doi:10.1037/h0025583.
Belsky, J.(1984). The determinants of parenting: A process model. Child Development, 55, 8396.
doi:10.2307/1129836.
Borkenau, P., Riemann, R., Angleitner, A., & Spinath, F.M. (2002). Similarity of childhood experi-
ences and personality resemblance in monozygotic and dizygotic twins: A test of the equal
environments assumption. Personality and Individual Differences, 33(2), 261269. doi:10.1016/
S0191-8869(01)00150-7.
Brody, G.H., Beach, S.R., Hill, K.G., Howe, G.W., Prado, G., & Fullerton, S.M. (2013). Using
genetically informed, randomized prevention trials to test etiological hypotheses about child
and adolescent drug use and psychopathology. American Journal of Public Health, 103(S1),
S19S24.
Burt, C.H., & Simons, R.L. (2014). Pulling back the curtain on heritability studies: Biosocial crimi-
nology in the postgenomic era. Criminology, 52(2), 223262. doi:10.1111/1745-9125.12036.
Charney, E. (2012). Behavior genetics and postgenomics. Behavioral and Brain Sciences, 35(05),
331358. doi:10.1017/S0140525X11002226.
Chen, N., & Deater-Deckard, K. (2015). Geneenvironment processes in adolescent family rela-
tionships. In B.N. Horwitz & J.M. Neiderhiser (Eds.), Gene-environment interplay in interper-
sonal relationships across the lifespan. Advances in behavior genetics (Vol. 3, pp.131150).
NewYork: Springer. doi:10.1007/978-1-4939-2923-8_6.
Collins, R., & Coltrane, S. (1985). Sociology of marriage and the family: Gender, love, and prop-
erty. Chicago, IL: Nelson-Hall.
Conley, D., Rauscher, E., Dawes, C., Magnusson, P.K., & Siegal, M.L. (2013). Heritability and
the equal environments assumption: Evidence from multiple samples of misclassified twins.
Behavior Genetics, 43(5), 415426. doi:10.1007/s10519-013-9602-1.
Cronk, N.J., Slutske, W.S., Madden, P.A., Bucholz, K.K., Reich, W., & Heath, A.C. (2002).
Emotional and behavioral problems among female twins: An evaluation of the equal environ-
ments assumption. Journal of the American Academy of Child and Adolescent Psychiatry,
41(7), 829837. doi:10.1097/00004583-200207000-00016.
Crusio, W.E. (2012). Heritability estimates in behavior genetics: Wasnt that station passed long
ago? Behavioral and Brain Sciences, 35(05), 361362. doi:10.1017/S0140525X12000970.
Deary, I.J., Spinath, F.M., & Bates, T.C. (2006). Genetics of intelligence. European Journal of
Human Genetics, 14(6), 690700. doi:10.1038/sj.ejhg.5201588.
Demo, D. H., & Cox, M. J. (2000). Families with young children: A review of research in the
1990s. Journal of Marriage and the Family, 62, 876895. doi:10.1111/j.1741-3737.2000.00876.x.
Dishion, T.J., Brennan, L.M., Shaw, D.S., McEachern, A.D., Wilson, M.N., & Jo, B. (2014).
Prevention of problem behavior through annual family check-ups in early childhood:
Intervention effects from home to early elementary school. Journal of Abnormal Child
Psychology, 42(3), 343354.
Domingue, B.W., Fletcher, J., Conley, D., & Boardman, J.D. (2014). Genetic and educational
assortative mating among US adults. Proceedings of the National Academy of Sciences,
111(22), 79968000. doi:10.1073/pnas.1321426111.
van Dongen, J., Slagboom, P.E., Draisma, H.H., Martin, N.G., & Boomsma, D.I. (2012). The
continuing value of twin studies in the omics era. Nature Reviews Genetics, 13(9), 640653.
doi:10.1038/nrg3243.
Duncan, L. E. (2014). Geneenvironment interactions (G E) in behavioral genetics. In S. H.
Rhee, & A.Ronal (Eds.), Behavior genetics of psychopathology (pp.253282). NewYork, NY:
Springer. doi:10.1007/978-1-4614-9509-3_9.
Elam, K.K., Harold, G.T., Neiderhiser, J.M., Reiss, D., Shaw, D.S., Natsuaki, M.N., Leve,
L.D. (2014). Adoptive parent hostility and children's peer behavior problems: Examining the
role of genetically informed child attributes on adoptive parent behavior. Developmental
Psychology, 50(5), 15431552. doi:10.1037/a0035470.
Fletcher, A.C., Steinberg, L., & Williams-Wheeler, M. (2004). Parental influences on adolescent
problem behavior: Revisiting Stattin and Kerr. Child Development, 75, 781796.
doi:10.1111/j.1467-8624.2004.00706.x.
Fraga, M.F., Ballestar, E., Paz, M.F., Ropero, S., Setien, F., Ballestar, M.L., Benitez, J.(2005).
Epigenetic differences arise during the lifetime of monozygotic twins. Proceedings of the
National Academy of Sciences of the United States of America, 102(30), 1060410609.
doi:10.1073/pnas.0500398102.
Fromm-Reichmann, F. (1948). Notes on the development of treatment of schizophrenics by psy-
choanalytic psychotherapy. Psychiatry, 11(3), 263273.
Gottesman, I.I., & Shields, J.(1966). Schizophrenia in twins: 16years consecutive admissions to
a psychiatric clinic. The British Journal of Psychiatry, 112(489), 809818. doi:10.1192/
bjp.112.489.809.
Gottlieb, G. (1998). Normally occurring environmental and behavioral influences on gene activity:
From central dogma to probabilistic epigenesis. Psychological Review, 105(4), 792.
doi:10.1037/0033-295X.105.4.792-802.
Grayson, D. (1989). Twins reared together: Minimizing shared environmental effects. Behavior
Genetics, 19(4), 593604. doi:10.1007/BF01066256.
Gunderson, E.P., Tsai, A.-L., Selby, J.V., Caan, B., Mayer-Davis, E.J., & Risch, N. (2006). Twins
of mistaken zygosity (TOMZ): Evidence for genetic contributions to dietary patterns and phys-
iologic traits. Twin Research and Human Genetics, 9(04), 540549. doi:10.1375/twin.9.4.540.
Hajal, N.J., Neiderhiser, J.M., Moore, G.A., Leve, L.D., Shaw, D.S., Harold, G., Reiss, D.
(2015). Angry responses to infant challenges: Parent, marital, and child genetic factors associ-
ated with harsh parenting. Child Development, 86(1), 8093. doi:10.1111/cdev.12345.
A. (2013). Biological and rearing mother influences on child ADHD symptoms: Revisiting the
Haworth, C., Wright, M., Luciano, M., Martin, N., De Geus, E., Van Beijsterveldt, C., Davis, O.
(2010). The heritability of general cognitive ability increases linearly from childhood to young
adulthood. Molecular Psychiatry, 15(11), 11121120. doi:10.1038/mp.2009.55.
Heijmans, B.T., Kremer, D., Tobi, E.W., Boomsma, D.I., & Slagboom, P.E. (2007). Heritable
rather than age-related environmental and stochastic factors dominate variation in DNA meth-
ylation of the human IGF2/H19 locus. Human Molecular Genetics, 16(5), 547554.
doi:10.1093/hmg/ddm010.
Henry, J., Boivin, M., Tarabulsy, G. (2015). Geneenvironment interplay in parenting young chil-
dren. In B.N. Horwitz & J.M. Neiderhiser (Eds.), Geneenvironment interplay in interper-
sonal relationships across the lifespan. Advances in behavior genetics (Vol. 3, pp. 1355).
NewYork: Springer. doi:10.1007/978-1-4939-2923-8_2.
Hettema, J. M., Neale, M. C., & Kendler, K. S. (1995). Physical similarity and the equal-
environment assumption in twin studies of psychiatric disorders. Behavior Genetics, 25(4),
327335. doi:10.1007/BF02197281.
Hill, W.G., Goddard, M.E., & Visscher, P.M. (2008). Data and theory point to mainly additive
genetic variance for complex traits. PLoS Genetics, 4(2), e1000008. doi:10.1371/journal.
pgen.1000008.
Ho, M.-W. (2013). No genes for intelligence in the fluid genome. Advances in Child Development
and Behavior, 45, 6792. doi:10.1016/B978-0-12-397946-9.00004-X.
Horwitz, A.V., Videon, T.M., Schmitz, M.F., & Davis, D. (2003). Rethinking twins and environ-
ments: Possible social sources for assumed genetic influences in twin research. Journal of
Health and Social Behavior, 44, 111129. doi:10.2307/1519802.
Horwitz, B.N., & Neiderhiser, J.M. (Eds.) (2015). Geneenvironment interplay in interpersonal
relationships across the lifespan. Advances in behavior genetics (Vol. 3), NewYork: Springer.
doi:10.1007/978-1-4939-2923-8.
Horwitz, B.N., Marceau, K., Narusyte, J., Ganiban, J., Spotts, E.L., Reiss, D., Neiderhiser,
J.M. (2015). Parental criticism is an environmental influence on adolescent somatic symptoms.
Journal of Family Psychology, 29(2), 283289. doi:10.1037/fam0000065.
Jacobs, N., Gestel, S., Derom, C., Thiery, E., Vernon, P., Derom, R., & Vlietinck, R. (2001).
Heritability estimates of intelligence in twins: Effect of chorion type. Behavior Genetics, 31(2),
209217. doi:10.1023/A:1010257512183.
Jensen, A.R. (1978). Genetic and behavioral effects of nonrandom mating. In R.T. Osborne, C.E.
Nobel, & N. Weyl (Eds.), Human variation: The biopsychology of age, race, and sex
(pp.51105). NewYork, NY: Academic Press.
Joseph, J. (2004). The gene illusion: Genetic research in psychiatry and psychology under the
microscope. Ross on Wye, UK: PCCS Books.
Kanner, L. (1949). Problems of nosology and psychodynamics of early infantile autism. American
Journal of Orthopsychiatry, 19(3), 416. doi:10.1111/j.1939-0025.1949.tb05441.x.
Kendler, K.S., & Gardner, C.O. (1998). Twin studies of adult psychiatric and substance depen-
dence disorders: Are they biased by differences in the environmental experiences of monozy-
gotic and dizygotic twins in childhood and adolescence? Psychological Medicine, 28(03),
625633. doi:10.1017/S0033291798006643.
Kendler, K.S., Neale, M.C., Kessler, R.C., Heath, A.C., & Eaves, L.J. (1993). A test of the equal-
environment assumption in twin studies of psychiatric illness. Behavior Genetics, 23(1),
2127. doi:10.1007/BF01067551.
Kendler, K.S., Neale, M.C., Kessler, R.C., Heath, A.C., & Eaves, L.J. (1994). Parental treatment
and the equal environment assumption in twin studies of psychiatric illness. Psychological
Medicine, 24(03), 579590. doi:10.1017/S0033291700027732.
Klahr, A.M., & Burt, S.A. (2014). Elucidating the etiology of individual differences in parenting:
A meta-analysis of behavioral genetic research. Psychological Bulletin, 140(2), 544586.
doi:10.1037/a0034205.
Klump, K.L., Holly, A., Iacono, W.G., McGue, M., & Willson, L.E. (2000). Physical similarity
and twin resemblance for eating attitudes and behaviors: A test of the equal environments
assumption. Behavior Genetics, 30(1), 5158. doi:10.1023/A:1002038610763.
Krueger, R.F., Moffitt, T.E., Caspi, A., Bleske, A., & Silva, P.A. (1998). Assortative mating for
antisocial behavior: Developmental and methodological implications. Behavior Genetics,
28(3), 173186. doi:10.1023/A:1021419013124.
Lerner, R.M. (2004). Genes and the promotion of positive human development: Hereditarian ver-
sus developmental systems perspectives. In C.Garcia Coll, E.Bearer, & R.M. Lerner (Eds.),
Nature and nurture: The complex interplay of genetic and environmental influences on human
behavior and development (pp.133). Mahwah, NJ: Erlbaum.
Leve, L.D., Neiderhiser, J.M., Harold, G.T., Natsuaki, M.N., Bohannan, B.J. M., Cresko, W.A.
(in press). Naturalistic experimental designs as tools for understanding the role of genes and the
environment in prevention research. Prevention Research.
Lickliter, R. (2012). From gene activity to behavior (and back again). Behavioral and Brain
Sciences, 35(5), 369370. doi:10.1017/S0140525X12001033.
Liu, C., Molenaar P.C.M., & Neiderhiser, J.M. The Impact of Variation in Twin Relatedness on
Estimates of Heritability and Environmental Influences. Manuscript submitted for
publication.
Marceau, K., Hajal, N., Leve, L.D., Reiss, D., Shaw, D.S., Ganiban, J.M., Neiderhiser, J.M.
(2013a). Measurement and associations of pregnancy risk factors with genetic influences, post-
natal environmental influences, and toddler behavior. International Journal of Behavioral
Development, 37, 366375. doi:10.1177/0165025413489378.
Marceau, K., Horwitz, B.N., Narusyte, J., Ganiban, J.M., Spotts, E.L., Reiss, D., & Neiderhiser,
J. M. (2013b). Geneenvironment correlation underlying the association between parental
negativity and adolescent externalizing problems. Child Development, 84(6), 20312046.
doi:10.1111/cdev.12094.
Marceau, K., Narusyte, J., Lichtenstein, P., Ganiban, J.M., Spotts, E.L., Reiss, D., & Neiderhiser,
J.M. (2015). Parental knowledge is an environmental influence on adolescent externalizing.
Journal of Child Psychology and Psychiatry, 56(2), 130137. doi:10.1111/jcpp.12288.
Marceau, K., McMaster, M., Smith, T., Daams, J., van Beijsterveldt, C., Boomsma, D., & Knopik,
V. (2016). The prenatal environment in twin studies: A review on chorionicity. Behavior
Genetics, 46, 286303.
Mare, R.D. (1991). Five decades of educational assortative mating. American Sociological Review,
56, 1532. doi:10.2307/2095670.
McAdams, T.A., Neiderhiser, J.M., Rijsdijk, F.V., Narusyte, J., Lichtenstein, P., & Eley, T.C.
(2014). Accounting for genetic and environmental confounds in associations between parent
and child characteristics: A systematic review of children-of-twins studies. Psychological
Bulletin, 140(4), 11381173. doi:10.1037/a0036416.
McClearn, G.E., Johansson, B., Berg, S., Pedersen, N.L., Ahern, F., Petrill, S.A., & Plomin, R.
(1997). Substantial genetic influence on cognitive abilities in twins 80 or more years old.
Science, 276(5318), 15601563. doi:10.1126/science.276.5318.1560.
McGue, M., Elkins, I., Walden, B., & Iacono, W.G. (2005). The essential role of behavioral genet-
ics in developmental psychology: Reply to Partridge (2005) and Greenberg (2005).
Developmental Psychology, 41, 993997.
Miller, M.B., DeYoung, C.G., & McGue, M. (2012). Assumptions in studies of heritability and
genotypephenotype association. Behavioral and Brain Sciences, 35(05), 372373.
doi:10.1017/S0140525X12001380.
Moffitt, T.E., & Beckley, A. (2015). Abandon twin research? Embrace epigenetic research? Premature
advice for criminologists. Criminology, 53(1), 121126. doi:10.1111/1745-9125.12061.
Mohr, D.C., Burns, M.N., Schueller, S.M., Clarke, G., & Klinkman, M. (2013). Behavioral inter-
vention technologies: Evidence review and recommendations for future research in mental
health. General Hospital Psychiatry, 35(4), 332338.
Monks, S., Leonardson, A., Zhu, H., Cundiff, P., Pietrusiak, P., Edwards, S., Schadt, E. (2004).
Genetic inheritance of gene expression in human cell lines. The American Journal of Human
Genetics, 75(6), 10941105. doi:10.1086/426461.
Narusyte, J., Neiderhiser, J. M., DOnofrio, B. M., Reiss, D., Spotts, E. L., Ganiban, J., &
Lichtenstein, P. (2008). Testing different types of genotypeenvironment correlation: An
extended children-of-twins model. Developmental Psychology, 44(6), 15911603. doi:10.1037/
a0013911.
Neiderhiser, J.M., Marceau, K., de Araujo-Greecher, M., Ganiban, J.M., Shaw, D.S., Reiss, D., &
Leve, L.D. (2016). Comparing medical records and self-report to measure pregnancy and deliv-
ery risk: Estimating the roles of genetic risk, perinatal risk, and rearing environment on early
childhood adjustment. Behavior Genetics, 46(3), 334352. doi:10.1007/s10519-016-9788-0.
Neiderhiser, J.M., Reiss, D., Pedersen, N., Lichtenstein, P., Spotts, E.L., Hansson, K.,
Elthammer, O. (2004). Genetic and environmental influences on mothering of adolescents: A
comparison of two samples. Developmental Psychology, 40(3), 335351. PMID15122961.
Neiderhiser, J.M., Reiss, D., Lichtenstein, P., Spotts, E.L., & Ganiban, J.(2007). Father-adolescent
relationships and the role of genotypeenvironment correlation. Journal of Family Psychology,
21(4), 560571. PMID18179328.
Ollikainen, M., Smith, K.R., Joo, E.J.-H., Ng, H.K., Andronikos, R., Novakovic, B., Saffery,
R. (2010). DNA methylation analysis of multiple tissues from newborn twins reveals both
genetic and intrauterine components to variation in the human neonatal epigenome. Human
Molecular Genetics, 19(21), 41764188. doi:10.1093/hmg/ddq336.
Partridge, T. (2005). Are genetically informed designs genetically informative? Comment on
McGue, Elkins, Walden, and Iacono (2005) and quantitative behavioral genetics. Developmental
Psychology, 41(6), 985988. doi:10.1037/0012-1649.41.6.985.
Patterson, G.R. (1982). Coercive family process (Vol. 3). Eugene, OR: Castalia.
Pike, A., & Oliver, B.R. (2015). Parenting in childhood. In B.N. Horwitz & J.M. Neiderhiser (Eds.),
Geneenvironment interplay in interpersonal relationships across the lifespan. Advances in
behavior genetics (Vol. 3, pp.5781). NewYork: Springer. doi:10.1007/978-1-4939-2923-8_3.
Plomin, R. (1986). Development, genetics, and psychology. Hillsdale, NJ: Lawrence Erlbaum
Associates.
Plomin, R., Willerman, L., & Loehlin, J. C. (1976). Resemblance in appearance and the equal
environments assumption in twin studies of personality traits. Behavior Genetics, 6(1), 4352.
doi:10.1007/BF01065677.
Plomin, R., DeFries, J.C., Knopik, V.S., & Neiderhiser, J.M. (2013). Behavioral genetics (6 ed.,
). NewYork, NY: Worth.
Plomin, R., DeFries, J.C., Knopik, V.S., & Neiderhiser, J.M. (2016). Top 10 replicated findings
from behavioral genetics. Perspectives on Psychological Science, 11(1), 323.
doi:10.1177/1745691615617439.
Polderman, T.J., Benyamin, B., de Leeuw, C.A., Sullivan, P.F., van Bochoven, A., Visscher, P.M.,
& Posthuma, D. (2015). Meta-analysis of the heritability of human traits based on fifty years of
twin studies. Nature Genetics, 47, 702709. doi:10.1038/ng.3285.
Reynolds, C.A., Barlow, T., & Pedersen, N.L. (2006). Alcohol, tobacco, and caffeine use: Spouse
similarity processes. Behavior Genetics, 36, 201215. doi:10.1007/s10519-005-9026-7.
Riese, M.L. (1999). Effects of chorion type on neonatal temperament differences in monozygotic
twin pairs. Behavior Genetics, 29(2), 8794. doi:10.1023/A:1021604321243.
Rose, S., Lewontin, R., & Kamin, L. (1984). Not in our genes: Biology, ideology, and human
nature. The Wilson Quarterly, 152.
Rowe, D.C. (1983). Biometrical genetic models of self-reported delinquent behavior: Twin study.
Behavior Genetics, 13, 473489. doi:10.1007/BF01065923.
Rowe, D.C. (1981). Environmental and genetic influences on dimensions of perceived parenting:
A twin study. Developmental Psychology, 17, 203208. doi:10.1037/0012-1649.17.2.203.
Russell, A., & Russell, G. (1992). Child effects in socialization research: Some conceptual and
data analysis issues. Social Development, 1(2), 163184. doi:10.1111/j.1467-9507.1992.
tb00122.x.
Rutter, M., Moffitt, T.E., & Caspi, A. (2006). Geneenvironment interplay and psychopathology:
Multiple varieties but real effects. Journal of Child Psychology and Psychiatry, 47(34), 226
261. doi:10.1111/j.1469-7610.2005.01557.x.
Scarr, S., & Carter-Saltzman, L. (1979). Twin method: Defense of a critical assumption. Behavior
Genetics, 9(6), 527542. doi:10.1007/BF01067349.
Scarr, S., & McCartney, K. (1983). How people make their own environments: A theory of
genotypeenvironment effects. Child Development, 424435.
Simons, R.L., Stewart, E., Gordon, L.C., & Conger Jr., R.D. (2002). Test of life-course explana-
tions for stability and change in antisocial behavior from adolescence to young adulthood.
Criminology, 40, 401. doi:10.1111/j.1745-9125.2002.tb00961.x.
Spuhler, J. N. (1968). Assortative mating with respect to physical characteristics. Eugenics
Quarterly, 15(2), 128140. doi:10.1080/19485565.1968.9987763.
on cognition across the life span: A meta-analysis of longitudinal twin and adoption studies.
Psychological Bulletin, 140(4), 949979. doi:10.1037/a0035893.
Turkheimer, E. (2011). Commentary: Variation and causation in the environment and genome.
International Journal of Epidemiology, 40(3), 598601. doi:10.1093/ije/dyq147.
Van Beijsterveldt, C.E. M., Overbeek, L.I. H., Rozendaal, L., McMaster, M.T. B., Glasner, T.J.,
Bartels, M., Boomsma, D.I. (2015). Chorionicity and heritability estimates from twin stud-
ies: The prenatal environment of twins and their resemblance across a large number of traits.
Behavior Genetics, 111. doi:10.1007/s10519-015-9745-3.
Vandenberg, D.J., Schlomer, G.L., Cleveland, H.H., Schink, A.E., Hair, K.L., Feinberg, M.E.,
Redmond, C. (2016). An adolescent drug use intervention blocks the effect of CHRNA5
genotype on smoking in high school. Nicotine & Tobacco Research, 18, 212220. doi:10.1093/
ntr/ntv095.
Vandenberg, S.G. (1972). Assortative mating, or who marries whom? Behavior Genetics, 2(23),
127157. doi:10.1007/BF01065686.
Veenma, D., Brosens, E., de Jong, E., van de Ven, C., Meeussen, C., Cohen-Overbeek, T.,
Tibboel, D. (2012). Copy number detection in discordant monozygotic twins of Congenital
Diaphragmatic Hernia (CDH) and Esophageal Atresia (EA) cohorts. European Journal of
Human Genetics, 20(3), 298304. doi:10.1038/ejhg.2011.194.
Visscher, P.M., Macgregor, S., Benyamin, B., Zhu, G., Gordon, S., Medland, S., Boomsma,
D.I. (2007). Genome partitioning of genetic variation for height from 11,214 sibling pairs. The
American Journal of Human Genetics, 81(5), 11041110. doi:10.1086/522934.
Wright, J.P., Barnes, J., Boutwell, B.B., Schwartz, J.A., Connolly, E.J., Nedelec, J.L., & Beaver,
K.M. (2015). Mathematical proof is not minutiae and irreducible complexity is not a theory: A
final response to Burt and Simons and a call to criminologists. Criminology, 53(1), 113120.
doi:10.1111/1745-9125.12059.
Chapter 6
GeneEnvironment Correlation asaSource
ofStability andDiversity inDevelopment
ChristopherR.Beam andEricTurkheimer
A Brief but Successful Baseball Career
The second authors son Manning was a pretty good ball player when he was little.
The second author himself had been a terrible athlete in his youth but a lifelong
baseball fan, and as soon as Manning could pick up a plastic bat and ball they were
out together in the front yard. Manning loved it, and by an apparent miracle of
behavior genetics he was pretty good. When he started t-ball at age five or so he was
always one of the better players, even discounting his fathers deeply biased judg-
ment. And the better he got, the more rewarding it was for father and son to hang out
in the front yard, and later on the baseball field itself. Once Little League started,
money was invested in the local gym that specialized in baseball, and individualized
coaching made him even better. His father was the stereotypical (i.e., over-) involved
parent of youth athletics: at every game, to be sure, but also at most practices, and
still out in the front yard every afternoon. When Manning was 10 years old he
pitched for an All-Star team that made it to the state semifinals, and despite the
teams eventual loss he was (we have by now abandoned any pretense of objectivity)
the best pitcher at the tournament, arguably the best 10-year-old pitcher in the
Commonwealth of Virginia.
C.R. Beam
University of Southern California,
3620 South McClintock Ave. Seeley G.Mudd, Room 523, Los Angeles, CA 90089, USA
e-mail: beamc@usc.edu
E. Turkheimer (*)
Department of Psychology, University of Virginia,
PO Box 400400 Gilmer, 313, Charlottesville, VA 22904, USA
e-mail: ent3c@eservices.virginia.edu

DOI10.1007/978-3-319-49227-8_6
112 C.R. Beam and E. Turkheimer
But then, slowly but surely, things started to change. As Manning graduated from
youth baseball at 14, he matriculated in a high school program that had been in
disarray for many years: badly maintained physical facilities, indifferent coaching,
and a long record of losing. As Manning and his cohort went through adolescence,
several of the boys he had played with for years started to develop into gifted ath-
letes: big, sleek, and fast. Many of those boys, based on their parents residence,
went to different high schools with superior baseball programs. Manning wound up
mid-sized (he is today fully grown at 5 11, too short for a serious pitcher) and
somewhat slow afoot. His easy-going personality, which had served him well in
pressured situations on the mound, did not necessarily predispose him to rigorous
working out. He remained a pretty good player, one of the two or three best on a
lousy team. But uncoached and unchallenged, his skills leveled off; it became clear
that no matter how hard he practiced, he was never going to be as good as some
other boys, gifted by nature with powerful shoulders, quick hands, and swift feet.
And as his skills leveled off, so did his interest. This was adolescence, after all,
when many children who were once obsessed with sports find other thingsmostly
not shared with Dadto occupy them. Eventually his father had to recognize that it
would be unfair to push him into more coaching and greater effort, because in fact
he was not going to have the skills to compete at a serious college level. So, slowly
but surely, in an intricate and occasionally fraught dance between teenager and par-
ent, father and son let it go. Manning now happily plays nonvarsity club baseball in
college, and he may or may not continue with that.
If Monkeys Could Play Baseball
How can one characterize processes of this kind, in which correlated genetic
endowments and environmental exposures jointly produce a phenotype, which then
influences expression of genetic traits and nonrandom exposure to future environ-
ments? It is a truism that it is meaningless to ask how much of an individuals suc-
cess or eventual asymptote is attributable to genes or environments. The methods
we have for quantifying such things apply to populations, not individuals. OK, but
what about a population of such individuals? It isnt as though Mannings experi-
ence was unusual, or limited to baseball; that is more or less how all skills develop,
improve, and eventually level off. And it isnt just skills; most negative behaviors
come about in the same way. People dont just wake up one day as heavy drinkers.
It begins slowly, with initial exposures to alcohol that change behavior and lead to
nonrandom environmental exposures in the future. Like sports ability, drinking habits
usually level off somewhere; occasionally, catastrophically, they do not.
In behavior genetics, the phenomenon we are discussing is called geneenvironment
correlation, or rGE.People with different genotypes are not exposed to environments
at random. Future environmental exposure depends on the current and past charac-
teristics of the person. Some of the problems induced by rGE are statistical: typical
methods for partitioning variance in quantitative genetics depend on independence
6 GeneEnvironment Correlation asaSource ofStability andDiversity inDevelopment 113
of genes and environments. The deeper questions, however, are conceptual and
developmental. In any prediction problem, once multiple predictors are correlated,
there is no definitive way to distinguish their causal effects. Ultimately, that is why
the experimental method exists: randomization to experimental condition is a way
of ensuring that ostensible causes of an outcome are independent of each other.
In lower animals, we can randomly assign genetically identical animals to different
environmental conditions or raise individuals who differ genetically in identical
(or as close to identical as a cage permits) circumstances. If monkeys could play
baseball we could breed them to be runners or hitters or throwers, assign them at
random to different kinds of upbringing and training programs, or perform surgeries
on them if it helped us learn how to make them better ballplayers.
Such experiments are not possible in humans, so we are left with inferring com-
plex developmental processes from natural variation and covariation between genetic
and environmental causes of behavior. This is a daunting task. This chapter explores
what siblings, and especially twins, can teach us about developmental processes with
correlated genes and environments. In particular, we will develop the idea that accu-
mulating rGEthe tendency of talented young baseball players to get better than
average coaching as long as they continue to play well and enjoy itproduces
trajectories, developmental pathways with stability and momentum. In individuals
with inborn reasons to differ, for example, within-pair DZ twins, accumulating rGE
produces differential trajectories that magnify differences between members of a
pair over time. The development of differences within twin and sibling pairs is a
quasi-experimental laboratory for the study of rGE.
Plomin, Loehlin andDeFries (1977)
The classic paper on the subject in developmental behavior genetics is Plomin,

Loehlin, and DeFries (1977), which has informed the discussion ever since. Gene
environment (rGE) correlation statistic refers to the relationship between heritable
characteristics of a behavior and nonrandom exposure to certain environmental fac-
tors associated with that same behavior. Geneenvironment correlation is produced
in three ways: Passive rGE occurs when parents provide both their genes and envi-
ronments to their children (intelligent parents provide smart genes and enriched
environments to their children). Evocative (or reactive) rGE occurs when people
elicit nonrandom responses from others in their environments (intelligent children
receive more attention from their teachers). Finally, active rGE occurs when people
nonrandomly seek out certain environments over others (intelligent children socialize
with other intelligent children).
Plomin, DeFries, and Loehlin (1977) is written from a very different perspective
from many of the papers we will be reviewing here. They specifically deny any
relationship between the quantitative genetic quantities that interest them and
interactionism in the developmental sense. They are, at bottom, concerned with
the partitioning of variance, and the ways that interaction and correlation might
complicate classical additive models. Heritable and environmental estimates of

individual difference variables, they note, may be biased by the presence of gene
environment correlation and interaction. With respect to adoption twin studies,
Plomin, DeFries, and Loehlin write, It seems likely that the genotype-environment
correlations in the case of separated biological relatives would ordinarily be positive,
tending to inflate estimates of [variance attributed to heritable factors] from adoption
designs (p. 319). By extension to nonadoption twin designs, geneenvironment
correlation generated by matching within-family environments to twins shared or
unshared genetic factors should produce the same positive inflation.
Heritability Increases withAge
The contrast proposed by Plomin etal. (1977) between science concerned only with
variance partitioning and developmental science investigating causal processes
underlying development cannot be maintained in the absolute. There would be no
reason to care about analyses of variation in behavior if they did not bear some
relevance to the development of actual organisms. The missing link between quan-
titative genetic analysis of individual differences and interactionist accounts of
developmentwhat Lewontin (2006) referred to as the analysis of variance and the
analysis of causesis time. Time is both a potential variable in the quantitative
models of the classical behavioral geneticist and the necessary substrate for investi-
gations of actual development. It is interesting to note that the tests for rGE pro-
posed in Plomin, Loehlin and DeFries were all cross-sectional, while the hypotheses
about development put forward by the interactionistsunsurprisingly in this case
are all longitudinal. Static cross-sectional analyses of families can analyze individ-
ual differences in the absence of theorizing about developmental causation;
longitudinal analyses cannot.
After the fundamental findings of univariate behavior genetics (Turkheimer, 2000),
perhaps the most widely cited and best replicated finding in the field is that the heri-
tability of behavior increases with age. The observation that heritability increases
over age has been observed, replicated, and meta-analyzed many times, and in
this chapter we will be concerned primarily with its historical and theoretical roots.
The paradigmatic example of the phenomenon is cognitive ability, but it seems to
apply widely to psychopathology as well (Bergen, Gardner, & Kendler, 2007). It is
generally thought not to apply for differences in personality (Turkheimer, Pettersson,
& Horn, 2014), although the steepest period of growth in heritability is in preado-
lescent childhood, and few longitudinal personality studies have been conducted in
that age range. For our purposes, we focus on phenotypes in which longitudinal
twin data are plentiful: height, weight, and cognitive ability.
We begin with height and weight. Siv Fischbein conducted important, and too
rarely cited, work on changes in the similarity of twin pairs over age in Sweden in
the 1970s. In 1964, Fischbein started one of the earliest developmental twin studies,
the SLU project. More than 300 pairs of MZ and DZ twins were followed between
HEIGHT
R
1.00
MZ
MZ
.80
DZ
.60
DZ
.40
.20
BOYS
GIRLS
10 11 12 13 14 15 16 17 18 AGE
R WEIGHT
1.00
MZ MZ
.80
.60
DZ
.40
.20 DZ
BOYS
GIRLS
10 11 12 13 14 15 16 17 18 AGE
Fig. 6.1 MZ and DZ Twin correlations for height and weight for males and females, by age. From
Fischbein (1978)
ages 10 and 18. Physical growth measurements were obtained every 6 months
throughout the study. Fischbein (1978) computed pairwise intraclass correlations
(ICCs) for height and weight in male and female MZ and DZ pairs. The results,
reproduced in Fig. 6.1, were striking: for height, all twin pairs started out highly
similar, MZ twins somewhat more so, and stayed that way. Heritability remained
constant across the age range. For weight, however, the MZ twins remained highly
similar, while the similarity of DZ twins dropped. The effect was especially notice-
able in the female pairs: by the age of 14 the DZ twin correlation was less than half
the MZ correlation, violating the assumptions of the classical twin model, and lead-
ing to the common conclusion that family environmental influences on the weight
of female twins were zero.
Fischbein (1978) interpreted the divergence of DZ correlations for weight with
increasing age as a heritability by environment interaction. Genetic differences
within DZ pairs caused them to respond differently to environmental stimulation,
driving their phenotypes apart over time. She based her divergence hypothesis on
Fuller and Thompsons (1960) assertion that individuals, when free to choose their
own environments, would take advantage of their environments differently and
develop different modes of adaptation over time. Fischbein (1978) proposed that
permissive (enriched) environments are required to facilitate divergence between DZ
twins within the same family for the reason that restrictive (impoverished) environ-
ments negatively reinforce genetic differences by limiting the ability to organize
ones own behavior. In other words, restrictive environments decrease heritability.
The first systematic study of changes in MZ and DZ correlations for cognitive
ability was conducted by Ronald Wilson in the Louisville Twin Study (LTS);
Bouchard (2013) has referred to age-related increases in the heritability of IQ as the
Wilson effect. The LTS was a longitudinal study of twin children and their fami-
lies. Children were assessed up to 15 times between the ages of 3 months and 15
years. Cognitive ability tests were given at each age, starting with Bayley Scales of
Infant Development, followed by early childhood IQ tests, the WISC, and finally the
WAIS-R.The twin and family correlations are illustrated in Fig. 6.2, reproduced
from Wilson (1983). Wilson concluded that the MZ correlation increased during
childhood while the DZ correlation decreased, producing statistical increases in
heritability estimates of cognitive ability with age.
There is a complication in the finding that heritability of cognitive ability
increases linearly over the life span, which we have illustrated in the LTS data
(Turkheimer & Beam, 2012) The line in Fig. 6.2 labeled age to age tracks the
within-person correlation between consecutive time points; it rises as a function of
.90
.84
MZ
.78
.72
DZ
CORRELATIONS
.66
.60
Age-to-age
.54
.48
.42 Sib-twin
.36
.30 Midparent-offspring
.24
0 3 6 12 18 24 30 36 4 5 6 7 8 9 15
MONTHS YEARS
AGES
Fig. 6.2 MZ and DZ twin, parentchild, sibling and stability correlations, by age. From Wilson (1983)
age. Turkheimer and Beam (2012) showed that this trend was in part because of
increasing stability of intelligence in time, but also because of increasing-reliability
of intelligence tests, especially in infancy (c.f. Humphreys & Davey, 1988). When
the twin correlations are corrected for changes in reliability, it becomes clear that
the data fit the same pattern as the Fischbein weight data: stable MZ correlations
and sharply declining DZ correlations.
Another important early study of longitudinal changes in twin similarity was the
meta-analysis of twin studies by McCartney, Harris, and Bernieri (1990). This was
a groundbreaking paper in several important ways. It was, first of all, important
simply as a meta-analysis of a complex and multidimensional research question, as
opposed to the straightforward analyses of treatment outcome that were typical at
the time. Its attempt to summarize twin studies over a wide range of phenotypes
prefigured the recent comprehensive meta-analysis of the twin literature conducted
by Polderman etal. (2015). In fact McCartney etal. (1990) exceeded Polderman
etal. (2015) in some ways by focusing productively on age as a determinant of twin
similarity. Another crucial insight was that the development of twin pairs, beyond
its classical application to the estimation of static variance components, is a natural
laboratory for the complex and uncontrolled processes that underlie the develop-
ment of human beings outside the restrictions of the experimental setting.
The McCartney, Bernieri and Harris meta-analysis showed that both MZ and DZ
twin pairs become more different as they age, but beyond that the results were dif-
ficult to interpret. Their meta-analytic methods, although current at the time, were
primitive by contemporary standards. Moreover, age is a difficult moderator to
include in summaries of twin studies, because twin samples are often internally
heterogeneous as regards age. In addition, results seemed to vary according to phe-
notype: for intelligence and other ability outcomes, both MZ and DZ twins became
more dissimilar over time, DZ twins relatively faster. For personality and tempera-
ment variables age trends were harder to discern, and if anything MZ twins differ-
entiated more recently. Most important, perhaps, were the theoretical considerations
that concluded the paper. They realized that the most common explanation for twin
and sibling differentiationsibling differentiationin which twin and sibling pairs
competed with each other for niches within the family, were unlikely to explain
their findings, as such a theory would suggest that identical twins differentiated
especially randomly. Instead, they proposed a very important possibility that once
again prefigured much of what followed: it appears that non-systematic envi-
ronmental effects might play a more dominant role than systematic environmental
effects. The implications for future research are challenging. (p.233).
Since these early studies, age-related increases in heritability estimates of cogni-
tive ability have been found in the U.S. (McGue, Bouchard, Iacono, & Lykken,
1993; Plomin & Spinath, 2004), Holland (Bouchard, 2009), and Sweden (Pedersen,
Plomin, Nesselroade, & McClearn, 1992). The steepest rise in heritability of cogni-
tive ability, as shown in a pooled analysis of U.S., U.K, Australian, and Dutch twin
samples (Haworth etal., 2010), appears to be from childhood to young adulthood
(increasing from approximately .40 to .70 across the studied range).
The Question ofHow
In any case, we are more interested in the mechanism of the effect than in its mere
occurrence: Why do heritability coefficients increase with age? As these heritabili-
ties are usually based on twin studies, if heritability is increasing it means that either
identical twin correlations are going up, fraternal twin correlations are going down,
or both. Addressing the problem on the level of twin correlations rather than herita-
bility coefficients keeps us closer to the actual development of individual people and
may help us sort out the causal processes involved.
The dynamics of heritability over the life span has been the crucible of the most
important theoretical discussions of developmental behavior genetics over the last
50 years. The reason, we will suggest, is that complex developmental processes
embody a paradox: dynamic nonrandom selection of environments according to
preexisting phenotypic differences causes divergence of genetically related indi-
viduals, at a rate inversely proportional to degree of genetic relatedness. So, for
example, developmental complexity causes DZ twins to diverge more rapidly than
MZ twins, driving up heritability even as complex dynamic processes render the
specification of individual genetic effects virtually impossible. The most complex
human individual differences, the ones that humans experience as psychological
and under willful control, are therefore often also the most heritable.
All human phenotypes are the result of developmental processes. There are no
genes for anything, not even height. Genes are merely inputs into developmental
processes, combining with environment, biology, randomness, and time to produce
trajectories leading to outcomes that are correlated with, but not explained by, the
original genes (Turkheimer, Haley, Waldron, DOnofrio, & Gottesman, 2003). The
correlation between genotype and outcome that remains detectable across the devel-
opmental process is the basis for the heritability of traits and is the explanation of
the universal heritability of everything from height to marital status. The difference
between traits like height that seem fundamentally genetic, in the sense that their
determination seems beyond environmental intervention or conscious control, and
more psychologically complex traits like personality that seem to encompass a
combination of genetic predisposition, environmental influence, and individual self-
determination, is not that one is more genetic than the other, but rather the time
frame and level of analysis at which the developmental processes take place.
Whatever the processes may be that transform genetic inputs into differences in
heightand they are by no means simple or well understoodthey certainly occur
very early in development, probably prenatally, and at a cellular and organismic
level that is beneath the purview of human consciousness. The perception that
developmental processes intervene between genes and developmental outcomes,
and that such processes complicate any simple understanding of the heritability of
behavior, has informed theoretical analysis of behavior genetics for more than half
a century. We review that literature later.
In Anne Anastasis (1958) classic Psychological Review paper, she argued that
separation of traits (both physical and psychological) into independent and additive
genetic and environmental components is a fools errand for the reason that genes
and environments are so tightly intertwined. Any and all genes are linked to identifiable
phenotypic outcomes by some sort of developmental process. Traits differ not in the
magnitude of their relationship to genes, but rather in the complexity of the gene-to-
phenotype pathway, what Anastasi called a continuum of indirectness. The longer
the indirect path between a gene and a phenotypic product, the more opportunities
there are for the results to diversify. Among the numerous examples she provided
for how genetic inputs are inert without accompanying environmental antecedents
and consequents, she makes the point that no trait is entirely a product of genetic
factors. She states, heredity sets the norm or limits within which environmental
differences determine the eventual outcome. In the case of some traits, such as
blood groups or eye color, this norm is much narrower than in the case of other
traits (p.200). Her question of How? heredity and environmental factors work
together to produce behavioral diversification anticipated much of the theoretical
and methodological work studying development in twin samples between the 1970s
and the present day. From her seven methodological suggestions for answering her
question of How?, we eventually propose a statistical model that satisfies her
seventh suggestion, in which Anastasi suggests modifying traditional twin methods to
follow the differential development within pairs twins raised in the same family to
better understand, The operation of differential environmental pressures, the devel-
opment of specialized roles, and other environmental influences [that] could thus be
more clearly identified and correlated with intellectual and personality changes in
the growing twins.
Wilson (1983) outlined a developmental genetic model to explain twin and
sibling synchronies in cognitive ability from 3 months to 18 years of age. Based on
Waddingtons (1942) developmental canalization hypothesisdevelopment of
species-typical behavior is robust to deviations in environment and genotypethe
pattern of convergence and divergence in MZ and DZ twin correlations supported
the genetically determined viewpoint that shared genotype ultimately superseded
effects of the environment. Based on his findings in the Louisville Twin Study, he
concluded There is a strong developmental thrust in the growth of intelligence that
continues through adolescence and is guided by an intrinsic template or ground
plan. The template is rooted in genetic processes that act throughout childhood and
adolescence (p.311). Wilson argued that developmental genetic processes prede-
termined the within-family ordering of twins (and by extension all sibling types)
based on their degree of consanguinity. Wilson credited environmental factors in
the development of childrens cognitive and psychological development, but his
argument that abilities are subject to maturational processes that are gene-con-
trolled (p.312) could be understood in several ways. Genetic processes may deter-
mine the trajectories of development and the range of differences between siblings,
as Wilson, in fact, suggests in his own discussion of his findings. Alternatively, we
will suggest that genetic factors are necessary (but insufficient) inputs into develop-
mental processes that require combinatorial environments. Under our alternative
view, genes would represent something like the parts of solenoid starters, and envi-
ronments are like every other element of a combustion engine: If the battery is dead,
or the spark plugs are not gapped properly, or the alternator is on the fritz, the engine
will not function no matter how good the starter. Simply put, gene action depends
on environments just as much as the converse.
In a paper titled, How People Make Their Own Environments: A Theory of
GenotypeEnvironment Effects, Scarr and McCartney (1983) developed a theory
based on the idea that free-ranging organisms select their environments based on
differences in their genotypes. One of the main motivations of their theory was to
explain the Wilson effect: The second question left unanswered by previous
research concerned the declining similarities of dizygotic twins and adopted siblings
from infancy to adolescence Neither environmental nor genetic theories can
effectively address these data. How can it be that the longer you live with someone,
the less like them you become? (p.431432). Their answer, very important for the
model we will put forward here, is that genetic differences within fraternal twin
pairs lead them to select different environments, which in turn lead them to become
ever more different phenotypically.
Scarr and McCartneys model was more explicitly gene driven than either
Anastasis or Fischbeins. Often, developmental models are invoked to soften the
potential determinism in genetic accounts of behavior, by filtering the effects of
genes through interactive developmental processes. In Scarr and McCartneys
model, however, the genes have unquestioned causal primacy. As they put it, In
summary, the theory of genotype-environment correlations proposed here describes
the usual course of human development in terms of three kinds of genotype-
environment correlations that posit cooperative efforts of the nature-nurture team,
directed by the genetic quarterback. Both genes and environments are constituents
in the developmental system, but they have different roles. Genes direct the course
of human experience, but experiential opportunities are also necessary for develop-
ment to occur. Individual differences can arise from restrictions in environmental
opportunities to experience what the genotype would find compatible. With a rich
array of opportunities, however, most differences among people arise from geneti-
cally determined differences in the experiences to which they are attracted and
which they evoke from their environments.
The causal primacy Scarr and McCartney afforded to geneticsthey were
un-phased by the loaded phrase genetically determined can be questioned on several
grounds. They take a remarkably agentic, almost anthropomorphic view of the
genome, which is capable of assessing environments and making decisions about
those it would find compatible. As we have pointed out before (Turkheimer &
Waldron, 2000), the genome is not an invisible hand capable of reaching out through
the organism and making determinations of behavioral selection: choices are made
by phenotypes, not genotypes. We have thus recharacterized Scarr and McCartneys
GE model as PE, phenotype to environment. Doing so only increases the devel-
opmental complexity of the geneenvironment system, as future selection into envi-
ronments depends on the complex state of the organism, as determined jointly by
genotype, environment, and prior developmental processes.
A second issue involves the ultimate attribution of all geneenvironment devel-
opmental processes to the genome. Even if one accepts Scarr and McCartneys
genetic quarterback metaphor, does it really make sense to attribute a process that is
jointly dependent on genes and environment solely to genes? Continuing with the
sports metaphor, suppose children with a small genetic advantage in quarterback
skills are systematically exposed to special coaching, which proves highly effective
and produces a large increase in their quarterbacking skills. Under Scarr and
McCartneys model, the manifestly environmental effects of coaching would be
attributed exclusively to genotype, which was presumably the factor determining
exposure to the relevant environment. Such an analysis might make sense if
environments were never under intentional control, but of course they often are.
The whole debate repeats an earlier discussion initiated by Roberts (1967):
The genotype may influence the phenotype either by means of biochemical or other pro-
cesses, labeled for convenience as development, or by means of influencing the animals
choice of environment. But this second pathway, just as much as the first, is a genetic one;
formally it matters not one whit whether the effects of the genes are mediated through the
external environment or directly through, say, the ribosomes. (p.218)
It was, in fact, Plomin, Loehlin, and DeFries (1977), in their seminal paper on
rGE, who formulated the equally well-known reply:
First, although formally it may not matter one whit in which way the effects of the genes are
mediated, in practice it often matters quite a few whits, especially if one should happen to be
interested in intervening in the process. Changing behavior by changing parental attitude is a
decidedly different proposition from tinkering with the ribosomes, even though a similar
behavioral change might conceivably be brought about by either means. (Plomin etal., 1977)
Meta-analysis is one way to understand complex nonexperimental developmental

problems; simulation is another. Turkheimer (Turkheimer, 2004; Turkheimer &
Gottesman, 1996) studied the complexities of environmental influences on behav-
ioral development by simulating the interactive relations among genetic and envi-
ronmental components of behavior. This paradox of environmental influences,
briefly summarized, is that we know that environments matter a great deal for devel-
opment but inquiry into how specific environmental factors cause development has
mainly been fruitless. Using simulation studies, Turkheimer and Gottesman showed
that the specific effects of individual environmental events can be detected for a
single fixed genotype, but when genotype is allowed to vary (as it does, of course,
in an actual sample of people) the specific effects of environment are impossible to
detect. The reason for this, in the simulations, was that genotype was relatively
fixed, compared to an environment that was responsive to changes in the phenotype.
(This was before the current fashion for epigenetic thinking, which has complicated
the easy assumption that genotype is fixed at birth.)
Similar to Fischbeins developmental model (1978; Fischbein, 1986; Fischbein,
Guttman, Nathan, & Esrachi, 1990), Bronfenbrenner and Cecis (1994) bioeco-
logical model emphasized how environmental differences determine whether and
which genetic factors become activated over the course of development. Using
the term proximal processes, they proposed that, human development takes place
through processes of progressively more complex reciprocal interaction between an
active, evolving biopsychological human organism and the persons, objects, and
symbols in its immediate environment (p. 572). Proximal processes actualize

innate genetic potential for positive development (e.g., weight, cognitive ability,
psychological health), leading to greater heritability of traits and abilities when
people are exposed to good (e.g., permissive) environments compared to poor envi-
ronments. Heritability coefficients in general, they conclude, are not exclusive of
environmental effects. Instead, they suggest, proximal processes determine which
individual capacities are realized and to what extent (p.583). In contrast to Scarrs
more genocentric views of human development, the bioecological model is quite
consistent with Gottliebs position that traits and behavior are probabilistic out-
comes of a host of recurrent interactions or coactions situated temporally and physi-
cally within complex developmental manifolds or systems (p.504in Lickliter &
Harshaw, 2010).
One effect of complex developmental processes is that they blur the causal
effects of individual genetic and environmental units. In a response to Plomin and
Daniels classic demonstration that the so-called nonshared environment was usu-
ally the largest source of variance in twin and family studies, Turkheimer and
Waldron (2000) conducted a meta-analysis of the associations between measured
within-family environmental variables (e.g., differential parenting for siblings) and
childhood outcomes. Despite the magnitude of the nonshared environmental vari-
ance component, the effect sizes for the measured environmental variables were
vanishingly small. The reason, they speculated, is related to what Plomin and
Daniels referred to as the gloomy prospect: the possibility that genetic and envi-
ronmental units were so caught up in interactive and nonlinear developmental pro-
cesses that their individual effects could not be meaningfully observed. Moreover,
Turkheimer and Waldron proposed that the developmental engine of these processes
was not really GE, as Scarr and McCartney had proposed, but rather PE:
The idea of P E interactions does not make sense in strictly cross-sectional models
because it would involve an interaction between a dependent (P) and an independent (E)
variable, but in developmental models it makes perfect sense to postulate that the effect of
an environmental event depends on the phenotype of the organism at the time the event
occurs; indeed, this model appears much more plausible than the idea that environmental
effects are somehow mediated directly by the genotype. If intelligent children evoke more
complex linguistic interactions with their caregivers, it is observable phenotypic aspects of
their behavior, not their genotype, that is having an effect on surrounding adults. This phe-
notype is in turn the cumulative result of developmental interactions between the childs
genotype and previous environmental events. (Turkheimer & Waldron, 2000, p.9192.)
Tucker-Drob andBriley
One prominent line of research into the developmental mechanisms that might
underlie age-related increases in heritability has been conducted by Elliot Turker-
Drob and Daniel Briley. (Briley & Tucker-Drob, 2013; Tucker-Drob & Briley,
2014). On the basis of the innovation and amplification theory of development
(Plomin, 1986), according to which increases in heritability with age may be
attributed to novel (or innovative) genetic effects or the amplification of stable

genetic effects from some earlier age, Tucker-Drob and Briley conclude that innova-
tive genetic effects on personality and cognitive ability at earlier ages give way to
amplification of genetic effects over time as twins access to novel environments
decreases.
A developmentally oriented Gene X Environment interaction perspective would predict
that the activation or expression of genes relevant for cognition is dependent on the environ-
ment in which the child is situated but that, as the child grows older, cognition becomes
increasingly shielded from this action. Thus, the stability of genetic effects would be low at
early ages when new gene expression is occurring due to novel environmental experiences,
and genetic stability would increase with age as the influence of novel environmental inputs
decreases (p.971).
There is no doubt that Tucker-Drob and Brileys analyses are the most compre-
hensive and detailed yet conducted of increases of heritability with age. Nevertheless,
it remains more of a description of the phenomenon than an explanation of it. As
children age, DZ twins become less similar relative to MZ twins, so heritability
increases while the effect of the shared environment decreases; at the same time, the
within-person stability of phenotypic intelligence increases. Tucker and Brileys
models demonstrate that these phenomena occur in the genetic portion of the vari-
ance as well as in the phenotype, but the implication that the process occurs because
of new genetic variation coming online during early childhood, which is then
replaced by stable genetic effects later, is the result of imposing a strictly additive
model on a developmental process that doesnt require it. We will argue that the
same phenomena summarized by Tucker-Drob and Briley can be accounted for by
a model in which phenotypic matching of individuals to environments accelerates
differentiation processes in genetically distinct individuals, such as DZ pairs.
henotype toEnvironment Transmission andGene

P
Environment Correlation
The phenomena that are the subject of developmental behavior geneticsincreasing

heritability, phenotypic stability of traits, stability of sibling (within-pair) differ-
ences, evidence of gene action that depends on the quality of environmental enrich-
mentare dispersed across multiple disciplines over nearly 60 years. To this end,
we hope to have characterized accurately the complexity of studying the mutual
push and tug of genetic and environmental influences in the production of physical,
psychological, and cognitive outcomes. While no one but the most committed
hereditarians and environmentalists would contend that genetic and environmental
causes of behavior operate independently of one another, the field of developmental
behavior genetics has been handicapped with statistical assumptions that limit
description of actual development. The final section of this chapter outlines a longi-
tudinal statistical approach intended to characterize how the reciprocal relation
between people and their environments over development can produce the changes
in heritability and stability that characterize developmental behavior genetics.
Using the basic statistical and quantitative genetic constraints of the classical
twin model, (identical twins share all of their genes whereas fraternal twins share
half of their genes on average) and familial and unique environments, the percent-
age of variance in behavior attributed to genetic and environmental causes can be
quantified and tested statistically. An important problem with the conventional twin
model, however, is that the assumption that genetic and environmental effects oper-
ate independently is not realistic. Indeed, in a developmental context the indepen-
dence assumption is not only unrealistic but wrong, and worse yet is wrong in a way
that can lead people to the conclusion that genes are the ultimate determinant of the
development of traits and abilities, rather than a crucial thread in a complex woven
outcome.
The crucial characteristic of causal models that reject GE independence is that
they recognize that genes (or, more accurately, phenotypes that are influenced by
genes) and environments mutually cause each other. Dickens and Flynn (2001)
referred to such models as reciprocal effects models and used them to explain how
environmental factors may have brought about the Flynn effect, in which intelli-
gence appears to be mostly genetic cross-sectionally, but nevertheless has increased
sharply over generational time. The Dickens and Flynn (2001) model described how
reciprocal causation between individuals and their environments at the level of
larger social pressures (like the effects of television or computers) could generate an
increasingly better match between a populations genetic material and the environ-
ments to which they were exposed (e.g., the advent of computers in society elicited
the expression of technological capabilities from a population at a faster rate than a
society still using adding machines). We have extrapolated the Dickens and Flynn
(2001) model to describe how twins and siblings diverge over the course of their
lives (Beam & Turkheimer, 2013). Within families, siblings with slight technologi-
cal inclinations or more agreeable dispositions over their sibling are more likely to
be reinforced when they are met with environments with computers or classrooms
filled with school children.
The three panels in Fig. 6.3 illustrate how the reciprocal exchange between peo-
ple and environments can propel siblings raised in the same family onto different
developmental trajectories. The developmental model we have proposed, referred to
hereafter as a PE model, can explain many of the phenomena of interest to devel-
opmental theorists and behavior geneticists. The model presented in Panel A of Fig.
6.3 is a reproduction of the model we originally presented. The model is explained
in detail elsewhere (Beam & Turkheimer; Beam etal., 2016), so we summarize our
approach here to orient the reader. Measured behavior, like cognitive ability, is par-
titioned into between-family effects (i.e., genetic and environmental effects that
make twins and siblings similar to one another) and within-family effects (i.e.,
genetic and environmental effects that make twins and siblings different from one
another). Although between-family genetic and environmental effects do contribute
to phenotypic stabilityrepresented by the transmission of the Ab and Eb over the entire
modelthey are not the focus of PE models.
Fig. 6.3 Panel of PE models. PE model represented as a fixed effects model as described in Beam and Turkheimer (2013) (Panel A). Clinical example of
PE process that underlies the development of depressed affect (Panel B). Close-up of the within-family portion of the fixed effects PE model (Panel C)
6 GeneEnvironment Correlation asaSource ofStability andDiversity inDevelopment
125
Within-family processes are the wires holding the violin together, the paths for
which are enlarged in Panel C of Fig. 6.3. First, the within-family genetic and
environmental transmission of effects (aar and ear) also contributes to phenotypic
stability, as these parameters simply summarize observed longitudinal correlations
of cognitive ability in the same way the between-family effects do. Secondand
more importantlythe paths projecting from phenotype to the subsequent within-fam-
ily environmental score (Ew) represent the transmission of siblings within-family
(unique) phenotype to their next set of environmental exposures (bPE). The PE
pathways, when included, help explain increasing heritability over time as well as
sibling differentiation.
In prior simulations, we have shown that when PE developmental processes
are misspecified as conventional genetic simplex models (Panel A in Fig. 6.3 with-
out the bPE pathways included) or as correlated factors models (Bartels, Rietveld,
Van Baal, & Boomsma, 2002), they produce increases in heritability estimates even
though genetic effects are constrained to be constant over time. The PE model
also illustrates how small initial differences in genetic and environmental effects
between siblings snowball into stable observed differences in behavior. PE mod-
els induce increasing within-family rGE over time. In our PE model, the accumu-
lation of rGE is a statistical representation of the consequences of real developmental
processes. As siblings establish themselves to be adept at academics, baseball,
musicianship or tendencies for mischief, deviance, or antisocial behavior, they grav-
itate to environments that facilitate their developing skill; these new environmental
factors (friends, teachers, bad influences) in turn pull for more pronounced behavior
from them. As the match between within-family differences in siblings and their
environments strengthens, within-family rGE increases. These effects, however, are
subtle and require large sample sizes to detect (Beam etal., 2016; Beam, Turkheimer,
Dickens, & Davis, 2015; de Kort et al., 2014; Dolan, de Kort, van Beijsterveldt,
Bartels, & Boomsma, 2014).
Applications toChildhood Intervention
The developmental model we proposed earlier has at least two implications for
childhood intervention research. First, PE models may help elucidate the devel-
opmental period in which the match between siblings and their environments out-
side of the family has greatest impact (early versus middle childhood, middle
childhood versus early adolescence, and so forth). The developmental period in
which siblings achievement scores or depressive mood symptoms begin to diverge
may provide cautionary information to parents that the siblings require differential
treatment to thrive as they develop. Additionally, the academically weaker or moodier
of the siblings may benefit from clinical evaluation with the intention of identify-
ing environmental factors befitting of the siblings innate strengths. Paradoxically,
the commonly observed increase in heritability of behavioral differences during
middle to late adolescence is an indicator that this period is when behavior is at its
most environmentally malleable. As twin siblings rely on their phenotypes to select

modifying environments, PE processes push fraternal twins apart more rapidly
than identical twins, driving up heritability even as the environment is having its
greatest effects.
Second, PE models mean that there are always at least two different develop-
mental trajectoriesone for advantaged children and another for disadvantaged
children. From the perspective of modeling developmental processes like achieve-
ment and depressed mood, there will no doubt be heterogeneity in developmental
trajectories overlooked in models that fit a single developmental process (linear
growth) while taking into account sample variance around the mean initial position
and mean slope. For this reason, childhood intervention research should focus on fit-
ting several developmental models (e.g., nonlinear as well as linear growth curve
models, simplex models as well as growth models) and to understand that a one-size-
fits-all model is likely to perform poorly over diverse developmental trajectories.
We recommend two longitudinal research designs that will extend understanding of
how correlation and interaction between genetic and environmental differences among
individuals developing in similar environments can set them onto different developmen-
tal trajectories. As polygenic risk scores (PRS) using measured DNA become increas-
ingly feasible as a means of quantifying genetic effects on behavior, we propose
genotyping children who share a common environment (classroom, for example) and
following their cognitive and academic development. Although PRS studies are cer-
tainly not going to replace twin studies (correlations between PRS are rarely higher than
.2), the small genetic effects found in PGS studies are precisely what a PE model
requires to trigger reciprocal exchange between peoples phenotype and their environ-
ments. They also have the advantage of allowing genetically informed research in data-
sets that do not include twins. As genotypic data become increasingly cost effective, we
expect studies of this variety to emerge sooner rather than later.
The second design we would like to see explored is a cohort longitudinal twin
study, the purpose of which would be to address when in childhood and adolescence
sibling divergence really begins to take shape. Using, for example, two cohorts of
twinsone aged 10 and another aged 15 at study entryfollowed longitudinally
would make it possible to identify the developmental time span when siblings (1)
selection of their own environments and the reciprocal reinforcement of those envi-
ronments have long-lasting development effects and (2) the amount of time it takes
for the reciprocal exchange between siblings and their environments to solidify into
stable phenotypic differences in traits of interest (cognitive ability, depressed mood,
risky behavior) prior to emerging adulthood.
Back toBaseball
Manning has two sisters who had their own athletic careers; we will close, however,
by imagining that he had a twin brother. If that twin had been identical, it would, of
course, have been male. Barring some unusual circumstance, he would also have been 5
11, probably also a little pudgy preadolescence, stretching out thereafter. He prob-
ably would not have been a fast runner. In all likelihood he would have shared
Mannings relaxed temperament. All of which goes to say that in those early days
out in the yard both twins would have been there, they both would have done well,
and Dad would have been rewarded for investing baseball time and coaching with
both of them. Probably they both would have done well in Little League, maybe
both would have pitched, worked side to side with their expensive coaches. And,
when their bodies started to change and assume their adult form, as Wilson showed,
they probably would have done so in synchrony. Their skills would have leveled off
together, and with their skills their interests.
Suppose, on the other hand, that Manning had a same-sex DZ twin. They would,
in all likelihood, still have been similar physically and psychologically, though not
as similar as the imagined MZ pair. Maybe the cotwin, more like his father, would
have a tendency to be anxious and perform poorly under pressure. Early youth
pitching, alone on the mound in a tight spot, would not have gone well, as it had not
for Dad. In the absence of in-game success, coaching would be less fun, and the
cotwin might have started to avoid it, or to pursue some other interest. The pheno-
typic difference in baseball interest would be magnified as soon one sibling would
be spending all available time on the baseball field, the other none at all. The phe-
notypic difference would grow even more, until, at age 10, the imagined sibling
called it quits.
Should we conclude from this imaginary anecdote that baseball ability is
somehow inherently genetic, that coaching is ineffective? Not at all, because the
effectiveness of coaching, and more generally of environmental exposure, is an
essential part of the story. But coaching is not administered at random; rather indi-
viduals are exposed to it in response to preexisting phenotypic differences, and the
effectiveness of coaching then cascades into new phenotypic differences that influ-
ence the next environment that comes along. This bottomlessly complex dance of
genetic and environmental effects is what has come to be called, geneenviron-
ment interplay. Interplay is a buzzword, sometimes a euphemism for developmen-
tal processes that are too complex to be managed scientifically, but it encompasses
an important paradox of developmental psychology (Turkheimer, 2006). Complex
processes that are full of interplayprocesses that involve free human selection of
courses of action and human responsibility for the outcomes of those actionsare
at the core of what we think of as behavior, behavior that is complex and free. But
the courses of diversification set off by interplay processes differentiate pairs of
individuals at a rate inversely proportional to their degree of genetic relatedness. In
particular, in a complex developmental world identical twins are very special,
shielded by their identity from the winds of interplay that disperse the rest of us,
even siblings.
The fact that developmental complexity differentiates fraternal twins faster
than identical twins, and identical twins more slowly than anyone else, leads to a
paradox. Additive models of genes and environment confound interplay, in the tra-
dition of Roberts Not one whit, with genetic variance, and conclude that our most
profoundly human developmental processes are inherently genetic. Shocked by
such a counter-intuitive outcome, opponents of behavior genetics conclude that,
despite all the evidence, complex human behavioral differences are somehow immune
from the otherwise universal effects of genes. Both extremes are incorrect: genes are
a crucial part of behavioral development, but their effects are only manifest at the
conclusion of the processes of which they are a necessary but insufficient
component.
References
Anastasi, A. (1958). Heredity, environment, and the question how?. Psychological Review,
65(4), 197208.
Bartels, M., Rietveld, M.J. H., Van Baal, G.C. M., & Boomsma, D.I. (2002). Genetic and envi-
ronmental influences on the development of intelligence. Behavior Genetics, 32(4), 237249.
Beam, C.R., & Turkheimer, E. (2013). Phenotype-environment correlations in longitudinal twin
models. Development & Psychopathology, 25(1), 716.
Beam, C.R., Emery, R.E., Reynolds, C.A., Gatz, M., Turkheimer, E., & Pedersen, N.L. (2016).
Widowhood and the stability of late life depressive symptomatology in the Swedish Adoption
Twin Study of Aging. Behavior Genetics, 46(1), 100113.
Beam, C.R., Turkheimer, E., Dickens, W.T., & Davis, D.W. (2015). Twin differentiation of
cognitive ability through phenotype to environment transmission: The Louisville Twin Study.
Behavior Genetics, 45(6), 622634.
Bergen, S. E., Gardner, C. O., & Kendler, K. S. (2007). Age-related changes in heritability of
behavioral phenotypes over adolescence and young adulthood: A meta-analysis. Twin Research
and Human Genetics, 10(03), 423433.
Bouchard, T. J. (2009). Genetic influence on human intelligence (Spearmans g): How much?
Annals of Human Biology, 36(5), 527544.
Bouchard, T. J. (2013). The Wilson effect: The increase in heritability of IQ with age. Twin
Research and Human Genetics, 16(05), 923930.
Briley, D.A., & Tucker-Drob, E.M. (2013). Explaining the increasing heritability of cognitive
ability across development: A meta-analysis of longitudinal twin and adoption studies.
Bronfenbrenner, U., & Ceci, S. J. (1994). Nature-nurture reconceptualized in developmental
perspective: A bioecological model. Psychological Review, 101(4), 568586.
de Kort, J.M., Dolan, C.V., Kan, K.-J., van Beijsterveldt, C.E. M., Bartels, M., & Boomsma, D.I.
(2014). Can GE-covariance originating in phenotype to environment transmission account for
the Flynn Effect? Journal of Intelligence, 2(3), 82105.
Dickens, W.T., & Flynn, J.R. (2001). Heritability estimates versus large environmental effects:
The IQ paradox resolved. Psychological Review, 108(2), 346369.
Dolan, C.V., de Kort, J.M., van Beijsterveldt, T.C. E.M., Bartels, M., & Boomsma, D.I. (2014).
GE covariance through phenotype to environment transmission: An assessment in longitudinal
twin data and application to childhood anxiety. Behavior Genetics, 44(3), 240253.
Fischbein, S. (1978). Heredity-environment interaction in the development of twins. International
Journal of Behavioral Development, 1(4), 313322.
Fischbein, S. (1986). Person-environment interaction in education setting. Report No. 1 from the
Department of Educational Research, Stockholm Institute of Education.
Fischbein, S., Guttman, R., Nathan, M., & Esrachi, A. (1990). Permissiveness-restrictiveness for
twins and controls in two educational settings: The Swedish compulsory school and the Israeli
kibbutz. Acta Geneticae Medicae et Gemellologiae, 39, 245257.
Fuller, J.L., & Thompson, W.R. (1960). Behavior genetics. NewYork: Wiley.
Haworth, C.M. A., Wright, M.J., Luciano, M., Martin, N.G., de Geus, E.J. C., van Beijsterveldt,
C.E. M., Plomin, R. (2010). The heritability of general cognitive ability increases linearly
from childhood to young adulthood. Molecular Psychiatry, 15(11), 11121120.
Humphreys, L.G., & Davey, T.C. (1988). Continuity in intellectual growth from 12 months to 9
years. Intelligence, 12(2), 183197.
Lewontin, R.C. (2006). The analysis of variance and the analysis of causes. International Journal
of Epidemiology, 35(3), 520525.
Lickliter, R., & Harshaw, C. (2010). Canalization and malleability reconsidered: the developmen-
tal basis of phenotypic stability and variability. In K. Hood, C. Halpern, G. Greenberg, & R.
Lerner (Eds.), Handbook of developmental science, behavior, and genetics (p. 491). Malden:
Wiley.
McCartney, K., Harris, M.J., & Bernieri, F. (1990). Growing up and growing apart: A developmental
meta-analysis of twin studies. Psychological Bulletin, 107(2), 226237.
McGue, M., Bouchard, T.J., Iacono, W.G., & Lykken, D.T. (1993). Behavioral genetics of cognitive
ability: A life-span perspective (pp.5976).
Pedersen, N.L., Plomin, R., Nesselroade, J.R., & McClearn, G.E. (1992). A quantitative genetic
analysis of cognitive abilities during the second half of the life span. Psychological Science,
3(6), 346352.
Plomin, R. (1986). Multivariate analysis and developmental behavioral genetics: Developmental
change as well as continuity. Behavior Genetics, 16(1), 2543.
Plomin, R., DeFries, J. C., & Loehlin, J. C. (1977). Genotype-environment interaction and correla-
tion in the analysis of human behavior. Psychological Bulletin, 84(2), 309.
Plomin, R., & Spinath, F. M. (2004). Intelligence: Genetics, genes, and genomics. Journal of
personality and social psychology, 86(1), 112.
Polderman, T.J., Benyamin, B., de Leeuw, C.A., Sullivan, P.F., van Bochoven, A., Visscher, P.M.,
& Posthuma, D. (2015). Meta-analysis of the heritability of human traits based on fifty years of
twin studies. Nature Genetics, 47(7), 702709.
Roberts, R.C. (1967). Some concepts and methods in quantitative genetics. In J.Hirsch (Ed.),
Behavior-genetic analysis (pp.214257). NewYork: McGraw-Hill.
Scarr, S., & McCartney, K. (1983). How people make their own environments: A theory of
genotype-environment effects. Child Development, 54, 424435.
on cognition across the life span: a meta-analysis of longitudinal twin and adoption studies.
Psychological Bulletin, 140(4), 949979.
Turkheimer, E. (2000). Three laws of behavior genetics and what they mean. Current Directions in
Turkheimer, E. (2004). Spinach and ice cream: Why social science is so difficult. In L.DiLalla
(Ed.), Behavior genetics principles: Perspectives in development, personality, and psychopa-
thology (pp.161189). Washington, DC: American Psychological Association.
Turkheimer, E. (2006). Inter: Action and play [Review of the book Genes and behavior: Nature-
nurture interplay explained]. PsycCRITIQUES, 51(43).
Turkheimer, E., & Beam, C.R. (2012). Lindon Eaves: Master of developmental models. Festschrift
in honor of Lindon Eaves. Edinburgh, Scotland.
Turkheimer, E., & Gottesman, I.I. (1996). Simulating the dynamics of genes and environment in
development. Development & Psychopathology, 8, 667677.
Turkheimer, E., & Waldron, M. (2000). Nonshared environment: A theoretical, methodological,
and quantitative review. Psychological Bulletin, 126, 78108.
Turkheimer, E., Haley, A., Waldron, M., DOnofrio, B., & Gottesman, I.I. (2003). Socioeconomic
status modifies heritability of IQ in young children. Psychological Science, 14(6), 623628.
Turkheimer, E., Pettersson, E., & Horn, E.E. (2014). A phenotypic null hypothesis for the genetics
of personality. Annual Review of Psychology, 65, 515540.
Waddington, C.H. (1942). Canalization of development and the inheritance of acquired characters.
Nature, 150(3811), 563565.
Wilson, R.S. (1983). The Louisville Twin Study: Developmental synchronies in behavior. Child
Chapter 7
Sampling intheExamination ofGene-
Environmental Interactions Within
aNeurodevelopmental Disorder Framework
YoungShinKim
Studying the etiological substrates of neurodevelopmental disorders (NDDs) have

proven to be vexing for a variety of reasons, including: phenotypic heterogeneity,
limited tools for directly assessing brain function, complex genetics and interac-
tions between biological/genetic and environmental factors. Work in this area is of
monumental importance due to the high prevalence and high level of clinical and
social impact of these conditions. With high prevalence, disorders like Autism
Spectrum Disorder (ASD2.6%) (Kim etal., 2011), Intellectual Disability (ID
3%) (APA, 2013), and Attention Deficit Disorder (ADHD7%) (APA, 2013), these
early onset conditions last a lifetime and consume enormous personal and commu-
nity resources. And, despite decades of research, at this point, there are no preven-
tive interventions and treatments are limited. Thus, there is a social, clinical, and
scientific imperative to examine the very best options for moving forward in under-
standing, preventing, and treating NDDs as well as many other neuropsychiatric
disorders that share in etiologic complexity and clinical need.
Recent discoveries with rare de novo mutations have helped move the field
forward by providing significant insight into biological mechanisms underlying the
neurodevelopmental pathophysiology of disorders ranging from ASD and ID to
schizophrenia. This work has started to pave the way for a better understanding of
the etiological heterogeneity and the important roles of transactions between genes
and environmental factors including gene-environmental correlations (rGE) and
gene-environment interactions (GxE) (Ahn etal., 2013; Bijlsma etal., 2009; Chaste
& Leboyer, 2012; Kim & State, 2014; Malhotra & Sebat, 2012; McCarthy etal.,
2009; Mefford etal., 2009; Sanders etal., 2011; Zufferey etal., 2012). GxE has
Y.S. Kim, M.D., Ph.D. (*)

Langley Porter Psychiatric Institute, University of California at San Francisco,
Box 0984-CAS, Room LP-377 401 Parnassus Avenue, San Francisco, CA 94143-0984, USA
e-mail: youngshin.kim@ucsf.edu

DOI10.1007/978-3-319-49227-8_7
132 Y.S. Kim
recently become a particular important focus of etiological research. This is because

ignoring these GxE may obscure independent genetic/environmental effects that
can lead to false-negative and inconsistent findings. Equally importantly, under-
standing GxE may help to identify particular genetic subpopulations that are at
elevated risk for the effects of environmental factors. Taken together, understanding
the impact of GxE can lead to the development of targeted interventions for these
disorders and may even modify the course of currently available treatments (Dempfle
etal., 2008; Tsuang, Bar, Stone, & Faraone, 2004). Recent studies in pharmacoge-
netics offers an example of how GxE can have a critical impact on the safe and
effective application of treatment: the recent discovery of a genotype affecting the
metabolism of the commonly used anticoagulant, warfarin. Warfarin has been used
for decades but it was only recently discover that the maintenance dose of warfarin
is significantly related to a specific genotype with the variant carriers requiring sub-
stantially less medication than those without the variant (variant carriers: 31.25mg/
week; wild-type: 37.5 mg/week). This leads to greater safety/fewer side effects
because the dosing of the warfarin can be tailored to the individual needs of the
patient (Albert etal., 2015). Is the same possible for the treatment of NDDs?
It is critical to note that GxE is often difficult to detect and apply. In particular,
methodological challenges arise when a GxE framework is utilized. At the current
stage of research in this area, the most critical of these methodological issues appear
to be related to sample determination and makeup. This is a common problem in all
research but genetic and GxE studies are particularly sensitive to methodologic
strategies of this sort. Unfortunately, sampling appropriateness is often overlooked
in designs for NDDs study. This is most problematic as proper sampling is critical
for establishing the validity and generalizability of study findings. Again, sampling
is critical in all fields of research but, in this area, results are particularly sensitive to
the sampling methods chosen by investigators.
Challenges in sampling for studies of GxE, among other things, all-too-often
include (Moldin, 1997; Rutter, Silberg, OConnor, & Simonoff, 1999; Salanti,
Sanderson, & Higgins, 2005):
1 . Reliance on clinical samples for making population-level inferences
2. Insufficient attention to population stratification and representativeness
3. Uncertainty about the sample size and type needed for the examination of GxE
4. Lack of precision in measures of processes and outcomes of interest
5. Emergence of increasing complexity in attempt to track GxE in NDDS
Until recently, the handful GxE investigations conducted in NDDs has been lim-
ited to observational studies, in particular, case-control studies. This is largely due
to ethical concerns related to the manipulation of environmental factors in human
research. These studies have been further hampered by inadequate sampling, includ-
ing such factors as small sample size, exclusive use of clinical sample, heterogene-
ity, and substructure of samples, as well as inaccurate assessment of environmental
exposures and false-positive findings stemming from multiple comparisons. This
sort of work has led to inaccurate and even misleading conclusions. Below, several
examples are offered in the field of ASD GxE research.
7 Sampling intheExamination ofGene-Environmental Interactions 133
Well-designed intervention research such as randomized controlled trials (RCT)

can manipulate hypothesized environmental factors by altering or adjusting envi-
ronmental influences or individual sensitivity to the environment in experimental
conditions. When combined with adequate sampling, RCTs can offer unique oppor-
tunities to examine the validity of hypothesized causal relationships, including GxE
in NDDs.
We will now examine the importance of sampling and its impact when examin-
ing GxE in NDDs. This will take place in three steps by reviewing:
1. Methodological issues in previous GxE studies of NDDS (that mostly utilized
case-control study design) with a focus on sampling
2. RCT strategies to explore the roles of GxE in the etiology of NDDS
3. Sampling strategies for future etiological research on GxE in neurodevelopmen-
tal framework
ethodological Issues inPrevious GxE Studies ofNDDs,

M
withFocus onSampling
Case-control studies have been utilized very often to examine GxE in NDDs. This
is largely due to convenience as it is much easier, and cheaper (and, hence more
feasible), to ascertain study subjects with rare disorders. For valid sampling in case-
control GxE study, one must validate the assumption that study participation does
not differ by genetic risks conditional on environmental exposure and disease/phe-
notype status. Violation of this assumption can lead to fatal flaws in the study due to
selection bias that places study validity in jeopardy. Gene by environment correla-
tion (rGE) and population stratification are among the most common examples of
flaws that lead to violations of this basic research assumption.
rGE occurs when genetic factors influence an individuals exposure to particular
environments, making those environmental factors, themselves, to appear as though
they are heritable. rGE reflect genetic differences in exposure to particular environ-
ments; as a result, genetic risks are not equally distributed across these environ-
ments. rGE may be heritable, even if consequences of environmental exposure are
not (Jaffee & Price, 2007; Kendler & Eaves, 1986). rGE can be mediated by heri-
table personality and behavioral characteristics. For example, risk for negative life
events may or may not be heritable. Events that are likely to be beyond ones con-
trol, include such things as exposure to a natural disaster; these are probably not
heritable. On the other hand, some negative life events that may be quite dependent
on an individuals behaviors such as getting into accidents or having contact with
legal authorities; factors directly leading to these events may very well be heritable
(Meek, Lemery-Chalfant, Jahromi, & Valiente, 2013). Another example is the rela-
tionship between aldehyde dehydrogenase (ALHD2) polymorphisms and alcohol
exposure. A functional polymorphism in the mitochondrial gene for ALDH2 metab-
olizes an ethanol byproduct, acetaldehyde, into acetate. Homozygotes for the mutant
ALDH2*2 allele have negligible ALDH2 activity; as a result, they have high levels
134 Y.S. Kim
of acetaldehyde accumulation that leads to the experience of an unpleasant flushing

reaction after alcohol intake. Heterozygotes have reduced ALDH2 activity and
experience less severe flushing. The ALDH2*2 allele is common in East Asian pop-
ulations, in whom it has a well-established, protective effect associated with an
approximate tenfold reduction in risk of alcoholism. The protective effect is thought
to be a direct consequence of the flushing reaction and associated nausea, drowsi-
ness, and headache that negatively reinforces exposure to alcohol and thus discour-
ages drinking. Therefore, the ALDH2 polymorphism is associated with alcohol
consumption and other alcohol-related phenotypes (e.g., fetal alcohol syndrome
(FAS)) because they have functional influences on ethanol metabolism, the interme-
diate products of which are potentially toxic (Jaffee & Price, 2012). This heritable
factor has a direct impact on environmental exposure. Failure to detect an event
such as this, even in the case of a very small rGE can lead to significant Type I errors
in GxE case-control studies (Jaffee & Price, 2007).
Population stratification occurs when allele frequencies are known to vary
widely between different populations, irrespective of phenotype status. When
cases and controls have different allele frequencies attributable to the diverse back-
grounds of the population in the sample that are unrelated to the phenotype, a study
is confounded and invalidated due to population stratification (Clayton et al.,
2005). The most frequently cited example of population stratification is from a
study on the association between an HLA haplotype and diabetes in a population
ascertained on a Pima Indian reservation (Knowler, Williams, Pettitt, & Steinberg,
1988). This study showed a classic case of confounding attributable to the admix-
ture of white European and Pima Indian ancestry on the association of the haplo-
type Gm3;5,13,14 with non-insulin-dependent diabetes mellitus. The association
disappeared when analysis was restricted to full-heritage Pima-Papago Indians,
demonstrating that Gm3;5,13,14 is a marker for Caucasian admixture rather than a
disease-risk allele. As such, unrecognized and/or unadjusted population stratifica-
tion is likely to increase Type I error (Clayton etal., 2005).
Another challenge for GxE research is the determination of what is an adequate
sample size. GxE studies are very sensitive to sample size because they are power
intensive. Even when testing for just a single GxE that is also specified a priori, the
exponential growth in the number of comparisons requires very large samples
(Boffetta etal., 2012). Inadequate sample size can result in a Type II error, leading
to erroneous false-negative findings due to lack of statistical power to detect mod-
est effect sizes.
Additionally, multiple comparisons in statistical tests are of particular concern in
GxE studies. This is due to:
(a) The number of genetic risk factors (e.g., different candidate risk alleles)
(b) The number of environmental risk factors (e.g., each chemical component in air
pollution)
(c) The number of phenotype measured (e.g., diagnosis of ASD, time of first words,
severity of ASD symptoms)
(d) The combinations of a, b, and c used in the analyses.
Taken individually or together, this can lead to false-positive (Type I) errors.

Case-control GxE studies of ASD tend to be especially vulnerable to and high-
light these methodological challenges. As an example, one group published three
GxE papers using the same study population, using a case-control study design. In
the first study of 429 children with ASD and 278 typical children, maternal MTHFR
677TT (rs1801133), CBS rs234715 GT+TT, and child COMT 472AA (rs4680)
genotypes conferred greater ASD risk when the mother did not take vitamins peri-
conceptionally (Schmidt etal., 2011). There were three problems with this report
which compromised the validity of the findings:
(a) The observed GxE lost significance after correction for multiple comparisons.
(b) Population stratification was not controlled when investigators reported differ-
ences in racial composition between cases and controls: indeed, investigators
did not even mention about population stratification in their report.
(c) Periconceptional vitamin intake data were collected retrospectively, without
biomarker and/or biomonitoring leading to increasing vulnerability to differen-
tial recall bias and measurement error between cases and controls.
In the second study of 429 children with ASD, 130 with developmental disorders
and 278 with typical development, the strongest protective effects of maternal folate
intake during first month of pregnancy were reported in mothers and children with
the MTHFR 677C>T variant genotype (Schmidt etal., 2012). Once again, multiple
comparisons were not addressed. Residual population stratification resulting from
crude controlling for racial categories rather than genomic controls as well as retro-
spective exposure data collection are sources of additional potential Type I error.
The third study reported interactions between high pollution and NO2 and the
MET CC genotype (rs1858830) in 251 cases and 156 controls. Air pollution was
determined using public air quality data for the area where individuals reported resi-
dence at birth (Volk etal., 2014). In addition to uncorrected multiple comparisons
and population stratification, pollution exposure was measured at a group level
while risk was stated at an individual level. The leads to serious concerns of ecologi-
cal fallacy. Ecological fallacy is a well-known concept in public health that refers an
erroneous inference because an association observed between variables on an
aggregate group level does not necessarily represent or reflect the association that
exists at an individual level. A causal relationship that exists on a group level or
among groups may not exist among the group individuals (Association IIE, 2014).
Sadly, it is an all-too-common but fundamental error that undermines the validity of
a studyas it does in this case. As a result, this study also has an overall significant
risk for Type I error.
To add to the problems in interpreting this body of work, all three studies used a
candidate gene approach for genetic analyses. This strategy has been largely aban-
doned because it is known to be prone to false-positive findings primarily due to the
low prior odds of association often combined with population stratification, small
sample size, multiple comparisons, and phenotypic heterogeneity (especially, in
neuropsychiatric disorders) (Duncan & Keller, 2011).
136 Y.S. Kim
The point here is not to criticize a single investigator or set of studies but, rather,
to demonstrate the incredible threats to validity for case-control studies of GxE.And,
then to illustrate that this vulnerability is readily compounded by other relatively
straightforward design and analytic decisions. It is not surprising that these and
similar studies have failed to be independently replicated.
CT Strategies toExplore theRoles ofGxE intheEtiology

R
ofNDDs
The many attempts to use observational, case-control studies to identify the role of
GxE in NDDs have proven to be inconclusive, at best. In fact, it appears that the
body of work and the failures of replication seem to indicate compromises to study
validity as a result of various biases, including sampling. These sorts of bias often
lead to spurious results and statistical significance that generally does not imply
causality. In the end, despite its challenges, random allocation of environmental risk
factors is necessary to clarify causal relationships between exposure and outcomes.
Such a research strategy is ethically unacceptable for studies in humans.
However, RCT designs can be an alternative strategy. In contrast to observational
studies, the study conditions in RCTs are controlled directly by the investigators,
including the specifically defined environmental factors. When properly designed,
they minimizes the possibility of bias and increases the likelihood of determining
causality. Furthermore, interventions using the GxE approach can focus on differen-
tial responses to interventions (i.e., ameliorative, preventive, and protective efforts)
to test hypothesized GxE causation; in the end, this will more likely lead to more
effective and better-targeted interventions. Since the existence of GxE suggests that
a particular genetic subpopulation is at elevated risk for the various impacts of envi-
ronmental factors, interventions are likely to be most effective in that
subpopulation.
When considering RCTs for use in examining GxE, several sampling conditions
must be met. It is only these circumstances that the results from these RCTs will
meet the standards of evidence, validity, reliability, and certainty to arrive at conclu-
sions about the role GxE in pathophysiology. The key factors in this design are:
1. The study population must be genetically homogenous to minimize population
stratification.
2. The study population should be population-based, representative samples that
minimize selection biases.
3. The sample size must also be calculated a priori to determine if it is adequate
sample to detect the estimated GxE effect sizes. Sampling must then be adjusted
to meet the expectations set by the power analysis.
4. Vigorous statistical analytic strategies must be planned a priori in order to mini-
mize both Type I and II errors.
An example of an RCT in the study of GxE is a study of the modification effects

of the FTO genotype on dietary interventions using various macronutrients and their
effects on insulin resistance. Investigators genotyped FTO variants rs1558902 and
rs9939609 and measured insulin resistance in fasting plasma samples at baseline, at
6 months and at 2 years. The sample included 743 overweight or obese adults (aged
3070 years, 60% women) from a randomized weight-loss dietary interventional
trial. The investigators then examined the interactions between FTO variants and
intakes of dietary fat and protein in relation to changes in body weight and insulin
resistance. Significant interactions were identified between rs1558902 and dietary fat
on changes in homeostasis model assessment of insulin resistance (HOMA-IR) and
insulin levels (P=0.003 and 0.004, respectively). Each risk allele (A) of rs1558902
showed a trend to be related to a 0.05-unit less reduction in both log(insulin) and
log(HOMA-IR) among the participants assigned to low-fat diets. However, this was
not significantly related to reductions in log(insulin) and log(HOMA-IR) for those
individuals assigned to high-fat diets during the 2-year period of intervention. This
suggests that carriers of the risk alleles at rs1558902 benefit differentially with
respect to improving insulin sensitivity by consuming high-fat, weight-loss diets
rather than low-fat diets (Zheng etal., 2015). While this study demonstrates the fea-
sibility of such trials in humans, at present, there are no published reports from simi-
lar types of RCTs to examine GxE in NDDs, including ASD.
ampling Strategies forFuture Etiological Studies

S
withinaNeurodevelopmental Psychopathology Framework
Despite its slow and relatively unimpressive start, there is a promising future for
studies to successfully advance our understanding of the roles of GxE in the etiol-
ogy of NDDs and other developmental psychopathology. There will be opportuni-
ties to use many studies, using both observational and RCT methods; however, by
this point, it should be clear that the careful design, research planning, execution,
and analyses of these studies will require extraordinary and the highest levels of
scientific rigor, especially in the areas of sampling, environmental measurement,
and analytic strategies.
For a variety of reasons, to date, NDDs research has focused almost entirely on
clinical populations. Part of the logic behind this strategy has been the assumption
that most individuals with significantly impaired disorders appear at clinics for care
thus allowing for the conclusion that clinical samples are representative of the dis-
order in the general population, an unfortunate erroneous conclusion. Additionally,
cost, efficiency, time, and other factors of convenience have led investigators to
ascertain study subjects from their clinics. Unfortunately, such convenience has
come at a very dear price because sampling errors are inherent to the exclusive use
of clinical samples. This problem is gone largely unrecognized by investigators in
the study of NDDSs and editors who publish this work.
138 Y.S. Kim
Certainly, clinical samples can serve as a useful initial step in the process of
identifying a clinical syndrome or phenotype. However, by their very nature, clini-
cal samples are not likely to represent the entire spectrum or dimensionality of any
particular form of developmental psychopathology. Furthermore, clinical samples
are known to be biased with respect to case severity, comorbidity, and factors asso-
ciated with health care system access (Berkson, 1946). And, this is often combined
with selection bias introduced from nonparticipation to a particular study. Taken
together, the level of bias often exceeds the estimation of the main or GxE effects
thus substantially diminishing the likelihood of a valid study outcome (Cheng &
Lee, 2012). As a result, it must be concluded that while clinical samples may have
some initial utility in the process of determining GxE effects on etiology in NDDs,
at best, they are the first step and must be assumed to provide a biased view of
NDDs which can only be corrected with other study designs which utilize different
methods of sampling.
In order to optimize study design and minimize the bias introduced from clinical
samples, epidemiologically ascertained, population-based samples that represent
the broad distribution of target phenotype and environmental risk factors are crucial
for the study to identify the roles of GxEs that increase risks for NDDs (Manolio,
2009; Rutter, Moffitt, & Caspi, 2006; Sabatti etal., 2009). While this approach may
be inconvenient, expensive, and not always possible, it should be the methodologi-
cal goal for such studies, at the present time.
The use of rigorous epidemiological methods is essential to obtain the necessary
samples for the etiologic role of GxE in NDDs. For example, our research team has
completed a total population prevalence survey of ASD in Korea and found that
ASD is much more prevalent than previously thought: 2.6% (Kim etal., 2011). It is
very surprising (to us and to reviewers) that 2/3s of the children ascertained (1.89%
prevalence) with ASD were previously unidentified in the community and would
have never been included in the study if we had relied solely on clinical sampling.
Even more informative in our study was the observation that the phenotypic charac-
teristics of those individuals with ASD who were ascertained in the community
(nonclinical settings) were distinct from clinically ascertained study participants.
Those identified from the community had a mean IQ of 98 (versus 75in the clinical
sample) and 2.5:1 male predominance (versus 5:1in the clinical sample). These and
other differences between the two populations represent striking differences in the
nature of the ASD phenotype. And, these findings have now been replicated in two
recently published ASD studies in US and Swedish populations (ADDMNS, 2012;
Sandin etal., 2014). Our study suggests that the full spectrum of the ASD phenotype
is consistent with current research that demonstrates a continuous distribution of the
dimensional phenotype of autism traits (Constantino & Todd, 2005). Because
previous research did not include this large and important portion of the ASD popu-
lation/phenotype, opportunities in research on the role of GxE in ASD could have
been completely missed.
It is also critical that future studies take a strongly developmental viewpoint in
trying to identifying plausible GxEs. Not only do the consequences of exposures to
environmental risk factors vary over time, but also gene expression varies over the
course of development. Exposure to environmental factors during a developmental

stage when the gene is not expressed may not be meaningful in pathogenetic mecha-
nisms of NDDSs. For example, exposures to environmental factors, including the
chemical components or nutrients in adulthood will not affect final height of an
individual while exposure to exactly the same nutrients in childhood may play a
crucial role in determining height and other characteristic via their GxE with growth
factor gene(s) (Liu, Maity, Lin, Wright, & Christiani, 2012). Another example is in
utero exposure to diethylstilbestrol confers a dramatically increased the risk for
vaginal cancer in female offspring of mothers taking the drug who themselves did
have increased the risk for the same cancer (Herbst, Ulfelder, & Poskanzer, 1971).
These and other examples make it clear that exposure to environmental risk factors
often have different doseresponse curves depending on the developmental stage at
which exposure occur. Indeed, at the very least, GxE should be conceptualized as a
three-way interaction in which the time of the exposure is the third factor (Liu,
Maity, Lin, Wright, & Christiani, 2012).
GxE studies are very power intensive. This often translates into the need of very
large samples. In reality, meaningfully large samples are generally beyond the
capacity of most individual investigators or centers. An ideal way to increase sample
sizes is by establishing large, datasharing consortia. This appears to be gaining trac-
tion in genetic studies and show promise in GxE studies. Meta-analyses may be
another way to overcome sample size challenges (Kim & State, 2014).
When large samples are not readily available and there are not multiple studies
for meta analysis, it may be possible to study subjects who are at the extreme end of
a particular environmental exposure (selective sampling based on measurable envi-
ronmental exposure levels). Such a study of extreme exposure can serve as a quasi-
experimental design to examine different treatments, for example, extremely low
vs. high exposure levels of environmental risks in observational studies or interven-
tions in RCT (Rellstab, Gugerli, Eckert, Hancock, & Holderegger, 2015). While this
selective sampling strategy effectively increases study power, its benefit is limited
to increasing the power to detect GxE, but not main effects, and it may inadvertently
introduce population stratification bias. Additionally, for the successful application
of this selective sampling approach, the phenotype as well as environmental factors
under study should be dimensional and measurable directly in the affected individu-
als with valid and reliable measures (Boks etal., 2007).
NDDs phenotype heterogeneity poses challenges in etiological research, includ-
ing GxE studies. Despite the current categorical approaches, most human pheno-
types (both typical and pathological), including NDDs, are dimensional and
continuously distributed in the population. The broad heterogeneity of phenotypes
that result from current categorical classifications has created substantial chal-
lenges in the discovery of pathophysiological mechanisms. These challenges espe-
cially include problems in determining the comparability of subjects in particular
clinical studies; this often leads to disparate study findings and failures to replicate
what may actually be meaningful initial findings.
Categorical diagnostic systems have their own advantages. In fact, when rigor-
ously applied, categorical diagnostic criteria for NDDs may be advantageous in
140 Y.S. Kim
some etiological studies by enriching the extremes of a particular phenotype.

Dimensional phenotyping is less vulnerable to phenotype heterogeneity and can
complement traditional diagnostic approaches (Abrahams & Geschwind, 2008).
In the end, categorical and dimensional system can be used for phenotyping, as
long as all other methodological challenges are carefully addressed in a well-
designed study.
There are many models for the study of GxE in NDDs but the most developed
may be in the study of ASD. Despite the many studies, research identifying how
environmental factors impact ASD pathogenesis has often been characterized by
relatively low levels of precision and reliability in the measurement of environmental
exposures, leading to both Type I and II errors. Environmental measures in ASD GxE
studies have been traditionally based on questionnaires or broad or community-level
ecological measures rather than necessary measurement of exposure at the level of
the individual. In the previously cited air pollution GxE study, interpreting individual
exposure levels based on group-level exposures to air pollution is difficult, if not
impossible. This is because there are so many specific factors that contribute to expo-
sure duration and individual biological factors. These include time individuals actu-
ally spend in the identified area, dosing (inhalation/absorption of pollutants), activity
levels, ages of those affected, timing of exposures (by age or trimester), and preexist-
ing health conditions, to mention but a few. Despite assertions of some investigators,
at this point, there is no valid information about how maternal air pollutant exposure
is correlated to fetal exposure in humans (Kim & Leventhal, 2015).
Finally, in order to conduct valid GxE studies, it is essential to have a plausible
model to test pathways from exposure to pathogenesis (or protective factors).
Understanding the biological pathways from gene variants to behaviors is critical if
researchers are going to properly identify study variable and then interpret data on
GxE.Each of these components of the pathogenic process must be understood in
relation to the neural systems they perturb. The advent of new tools to measure dis-
ruptions in neural systems will be helpful in this process and may actually provide
at least proximal biomarkers that will contribute to study validity and reliability.
Biomarkers, such as anatomic and functional neuroimaging and EEGs along with
forthcoming technologies, such as geographic information systems and biological
monitoring/sensing, will enable more precise measurement of neural behavior,
human behavior, and environmental exposures at the level of the individual. This
linked to the increasing speed and decreasing cost for measurement of molecular
genetics and epigenetics will improve the quality of studies of NDDs (Kim & State,
2014). Direct assessment of individual level exposure via biomarkers of exposure
will also provide insights into chemical, social, and physical factors that are impor-
tant to specific individuals (Liu etal., 2012). Utilization of such biomarkers with
optimal specificity, sensitivity, and reliability will be likely to help bridging the gap
between neural systems and NDDS outcomes. Equally importantly, these sort of
studies have enormous potential in facilitating useful animal models that can, in
turn, point scientists back to exploratory models in humans. Taken together, it
increases the likelihood that researchers will identify significant associations
between genes and the behaviors that bring about environmental experiences.
Conclusions
There is no doubt that there has been a rapid advance in our capacity to understand
many elements of NDDs. However, in the ever-growing fascination with incredible
advances in technology, there has been a relative insensitivity to the very basics of
scientific methodology. In particular, this has led to many missed opportunities due to
problematic sampling and measurement. This is unfortunate as there never was a time
when careful attention to scientific methods has been so important as the importance
of the complex interactions between biology/genes, environment, and their time
sequences in pointing the way to NDDs. Understanding the roles of GxE in the patho-
genesis of NDDs requires much sophistication but that sophistication includes careful
attention to the basics. Larger sample sizes will be necessary but not sufficient to
address complex questions about the role of GxE in NDDs. Study designs must
include careful attention to sampling with a preference for ascertainment that is epide-
miological and population based to allow for the capture of the entire spectrum of both
a given phenotype and the environmental risks. Furthermore, it is not necessary to
dispute whether phenotypic characterization must be categorical or dimensional. Each
can serve a purpose and, in some cases, both can be applied. More importantly, equally
careful attention must be applied to direct environmental measurement at the level of
the individual that is conducted with accurate, specific, and sensitive measures/bio-
markers. All of this work must be conducted in large scale, largely collaborative stud-
ies which will likely require thousands if not tens of the thousands of research
participants. As is happening in other areas of science, datasharing and large-scale
collaborations will set the stage for progress in understanding GxE in NDDs. And,
finally, despite the challenges, there are important roles for RCTs in uncovering
GxE.While such studies may be very difficult to design and implement, their poten-
tial for carefully and safely manipulating specific biological systems may rapidly
move the field forward. With very few mechanisms understood at this time, even one
replicable mechanistic GxE study will be a giant step forward in the field.
Excellence in sampling and measurement are neither exciting nor elegant in this era of
colorful neuroimages or genetic heat maps. However, without careful attention to these
areas of experimental design, we will be left with a future of small studies that cannot be
replicated. One is tempted to speculate about what the future will hold but, in fact, we
should really focus on what we can do in the present as we can make major advances in
understanding the role of GxE in NDDs now. We need not wait for the development of a
new technology, but rather carefully apply what we know in the present.
References
Abrahams, B.S., & Geschwind, D.H. (2008). Advances in autism genetics: On the threshold of a
new neurobiology. Nature Reviews, 9, 341355.
ADDMNS, C.D. C. (2012). Prevalence of autism spectrum disordersAutism and Developmental
Disabilities Monitoring Network, 14 sites, United States, 2008. MMWR Surveillance
Summaries, 61, 119.
142 Y.S. Kim
Ahn, K., Gotay, N., Andersen, T. M., Anvari, A. A., Gochman, P., Lee, Y., Rapoport, J. L.
(2013). High rate of disease-related copy number variations in childhood onset schizophrenia.
Molecular Psychiatry., 19(5), 568572.
Albert, D., Belsky, D.W., Crowley, D.M., Bates, J.E., Pettit, G.S., Lansford, J.E., Dodge,
K.A. (2015). Developmental mediation of genetic variation in response to the Fast Track pre-
vention program. Development and Psychopathology, 27, 8195.
APA (2013). DSM-5. Washington, DC: American Psychiatric Association.
Association IIE. (2014). In M.Porta (Ed.), A dictionary of epidemiology (6 ed.). NewYork: Oxford
University Press.
Berkson, J.(1946). Limitations of the application of the fourfold table analysis to hospital data.
Biometrics, 2, 4753.
Bijlsma, E.K., Gijsbers, A.C., Schuurs-Hoeijmakers, J.H., van Haeringen, A., van de Putte DE,
F., Anderlid, B.M., Ruivenkamp, C.A. (2009). Extending the phenotype of recurrent rear-
rangements of 16p11.2: Deletions in mentally retarded patients without autism and in normal
individuals. European Journal of Medical Genetics, 52, 7787.
Boffetta, P., Winn, D., Ioannidis, J.P., Thomas, D., Little, J., Smith, G., Khoury, M. (2012).
Recommendations and proposed guidelines for assessing the cumulative evidence on joint
effects of genes and environments on cancer occurrence in humans. International Journal of
Epidemiology, 41, 686704.
Boks, M.P., Schipper, M., Schubart, C.D., Sommer, I.E., Kahn, R.S., & Ophoff, R.A. (2007).
Investigating gene environment interaction in complex diseases: increasing power by selective
sampling for environmental exposure. International Journal of Epidemiology, 36, 13631369.
Chaste, P., & Leboyer, M. (2012). Autism risk factors: Genes, environment, and gene-environment
interactions. Dialogues in Clinical Neuroscience, 14, 281292.
Cheng, K.F., & Lee, J.Y. (2012). Assessing the joint effect of population stratification and sample
selection in studies of gene-gene (environment) interactions. BMC Genetics, 13, 5.
Clayton, D.G., Walker, N.M., Smyth, D.J., Pask, R., Cooper, J.D., Maier, L.M., Todd, J.A.
(2005). Population structure, differential bias and genomic control in a large-scale, case-control
association study. Nature Genetics, 37, 12431246.
Constantino, J.N., & Todd, R.D. (2005). Intergenerational transmission of subthreshold autistic
traits in the general population. Biological Psychiatry, 57, 655660.
Dempfle, A., Scherag, A., Hein, R., Beckmann, L., Chang-Claude, J., & Schafer, H. (2008). Gene-
environment interactions for complex traits: Definitions, methodological requirements and
challenges. European Journal of Human Genetics, 16, 11641172.
environment interaction research in psychiatry. The American Journal of Psychiatry, 168,
10411049.
Herbst, A.L., Ulfelder, H., & Poskanzer, D.C. (1971). Adenocarcinoma of the vagina. Association
of maternal stilbestrol therapy with tumor appearance in young women. The New England
Journal of Medicine., 284, 878881.
Jaffee, S.R., & Price, T.S. (2007). Gene-environment correlations: A review of the evidence and
implications for prevention of mental illness. Molecular Psychiatry, 12, 432442.
Jaffee, S. R., & Price, T. S. (2012). The implications of genotype-environment correlation for
establishing causal processes in psychopathology. Development and Psychopathology, 24,
12531264.
Kendler, K.S., & Eaves, L.J. (1986). Models for the joint effect of genotype and environment on
liability to psychiatric illness. The American Journal of Psychiatry., 143, 279289.
Kim, Y. S., & Leventhal, B. L. (2015). Genetic epidemiology and insights into interactive
genetic and environmental effects in autism spectrum disorders. Biological Psychiatry, 77,
6674.
Kim, Y.S., & State, M.W. (2014). Recent challenges to the psychiatric diagnostic nosology: A
focus on the genetics and genomics of neurodevelopmental disorders. International Journal of
Epidemiology, 43, 465475.
Kim, Y.S., Leventhal, B.L., Koh, Y.J., Fombonne, E., Laska, E., Lim, E.C., Grinker, R.R.
(2011). Prevalence of autism spectrum disorders in a total population sample. The American
Journal of Psychiatry., 168, 904912.
Knowler, W.C., Williams, R.C., Pettitt, D.J., & Steinberg, A.G. (1988). Gm3;5,13,14 and type 2
diabetes mellitus: An association in American Indians with genetic admixture. American
Journal of Human Genetics, 43, 520526.
Liu, C.Y., Maity, A., Lin, X., Wright, R.O., & Christiani, D.C. (2012). Design and analysis issues
in gene and environment studies. Environmental Health, 11, 93.
Malhotra, D., & Sebat, J. (2012). CNVs: Harbingers of a rare variant revolution in psychiatric
genetics. Cell, 148, 12231241.
Manolio, T.A. (2009). Cohort studies and the genetics of complex disease. Nature Genetics, 41, 56.
McCarthy, S. E., Makarov, V., Kirov, G., Addington, A. M., McClellan, J., Yoon, S., Sebat,
J. (2009). Microduplications of 16p11.2 are associated with schizophrenia. Nat Genet, 41,
12231227.
Meek, S. E., Lemery-Chalfant, K., Jahromi, L. B., & Valiente, C. (2013). A review of gene-
environment correlations and their implications for autism: A conceptual model. Psychological
Review, 120, 497521.
Mefford, H.C., Cooper, G.M., Zerr, T., Smith, J.D., Baker, C., Shafer, N., Eichler, E.E. (2009).
A method for rapid, targeted CNV genotyping identifies rare variants associated with neuro-
cognitive disease. Genome Research, 19, 15791585.
Moldin, S.O. (1997). The maddening hunt for madness genes. Nature Genetics, 17, 127129.
Rellstab, C., Gugerli, F., Eckert, A. J., Hancock, A. M., & Holderegger, R. (2015). A practical
guide to environmental association analysis in landscape genomics. Molecular Ecology, 24,
43484370.
Rutter, M., Silberg, J., OConnor, T., & Simonoff, E. (1999). Genetics and child psychiatry:
I.Advances in quantitative and molecular genetics. Journal of Child Psychology and Psychiatry,
and Allied Disciplines, 40, 318.
Rutter, M., Moffitt, T.E., & Caspi, A. (2006). Gene-environment interplay and psychopathology:
Multiple varieties but real effects. Journal of Child Psychology and Psychiatry, and Allied
Disciplines, 47, 226261.
Sabatti, C., Service, S.K., Hartikainen, A.L., Pouta, A., Ripatti, S., Brodsky, J., Peltonen, L.
(2009). Genome-wide association analysis of metabolic traits in a birth cohort from a founder
population. Nature Genetics, 41, 3546.
Salanti, G., Sanderson, S., & Higgins, J.P. (2005). Obstacles and opportunities in meta-analysis of
genetic association studies. Genetics in Medicine, 7, 1320.
Sanders, S.J., Ercan-Sencicek, A.G., Hus, V., Luo, R., Murtha, M.T., Moreno-De-Luca, D.,
State, M.W. (2011). Multiple recurrent de novo CNVs, including duplications of the 7q11.23
Williams syndrome region, are strongly associated with autism. Neuron, 70, 863885.
Sandin, S., Lichtenstein, P., Kuja-Halkola, R., Larsson, H., Hultman, C.M., & Reichenberg, A.
(2014). The familial risk of autism. Journal of the American Medical Association, 311,
17701777.
Schmidt, R., Tancredi, D.J., Ozonoff, S., Hansen, R.L., Hartiala, J., Allayee, H., Hertz-
Picciotto, I. (2012). Maternal periconceptional folic acid intake and risk of autism specturm
disorders and developmental delay in the CHARGE (CHilhood Autism Risk from Genetics
and Environment) case-control study. The American Journal of Clinical Nutrition, 96,
8089.
Schmidt, R.J., Hansen, R.L., Hartiala, J., Allayee, H., Schmidt, L.C., Tancredi, D.J., Hertz-
Picciotto, I. (2011). Prenatal vitamins, one-carbon metabolism gene variants, and risk for
autism. Epidemiology (Cambridge, MA), 22, 476485.
Tsuang, M.T., Bar, J.L., Stone, W.S., & Faraone, S.V. (2004). Gene-environment interactions in
mental disorders. World Psychiatry., 3, 7383.
Volk, H., Kerin, T., Lurmann, F., Hertz-Picciotto, I., McConnell, R., & Campbell, D. (2014).
Autism spectrum disorder interaction of air pollution with the MET receptor tyrosine kinase
gene. Epidemiology (Cambridge, MA), 25, 4447.
144 Y.S. Kim
Zheng, Y., Huang, T., Zhang, X., Rood, J., Bray, G.A., Sacks, F.M., & Qi, L. (2015). Dietary fat
modifies the effects of FTO genotype on changes in insulin sensitivity. The Journal of Nutrition,
145, 977982.
Zufferey, F., Sherr, E. H., Beckmann, N. D., Hanson, E., Maillard, A. M., Hippolyte, L.,
Jacquemont, S. (2012). A 600kb deletion syndrome at 16p11.2 leads to energy imbalance and
neuropsychiatric disorders. JMed Genet, 49, 660668.
Part II
Application to Developmental
Psychopathology
Chapter 8
Do Gene-by-Environment Interactions Offer
Potential Intervention Strategies inAnxiety
Disorders?
NealRyan
In many ways, anxiety disorders offer a fertile ground for testing whether greater
understanding of gene-by-environment (GxE) interactions may lead to new inter-
vention strategies. The neural circuitry underlying fear learning and fear extinction
has been well mapped in rodents, non-human primates, and humans. fMRI para-
digms are now available for the involved circuitry and studies of both fear learning
and fear extinction can be ethical and relatively straightforward to employ through-
out development. In addition to effective pharmacological treatments, we have at
least two different treatment approaches thought to work through different neural
mechanisms, exposure, and attention bias modification, which permit a much wider
range of testable hypotheses. There are interesting changes in fear learning and fear
extinction with development which provide us with a richer opportunity for treat-
ment development and thus potentially more opportunity to leverage GxE effects in
treatment selection (Casey, Glatt, & Lee, 2015).
Understanding genetic and GxE effects could potentially lead to improved inter-
vention in several different ways. First, finding a gene could help us understand
mechanism which might lead to potential novel pharmacological or other therapeu-
tic approaches (Levy etal., 2009). Second, a gene or informative GxE interaction
could potentially prove to be so strongly associated with differential treatment effect
that this alone would be informative regarding optimal treatment approaches. Much
more likely, however, is that a combination of genetic and GxE information together
as well as demographic, syndromal, imaging, EEG, neuropsychiatric test, and other
information may allow us to meaningfully predict differential treatment response
(Kim, Sharma, & Ryan, 2015; Wallace, Frank, & Kraemer, 2013).
N. Ryan, MD (*)
University of Pittsburgh, Pittsburgh, PA, USA
e-mail: ryannd@upmc.edu

DOI10.1007/978-3-319-49227-8_8
148 N. Ryan
There appear to be significant developmental changes in brain circuitry associ-

ated with fear learning and fear extinction (Casey etal., 2015). Therefore, while we
do not yet have data on genetic or GxE effects in anxiety disorders as a function of
developmental stage, it is unlikely, given our current understanding, that these are
invariant through the life span.
However, there are many challenges. To date, evidence about potential gene-by-
environment (GxE) interactions related to anxiety disorders has been gathered pri-
marily by studies using candidate gene approaches; examining common single
nucleotide polymorphisms (SNPs) in preselected candidate genes. Generally, such
approaches have not been successful in complex psychiatric disorders (Merikangas
& Risch, 2003).
Before going further, it is worth considering the different ways that genes and
environment could influence onset, course, and/or treatment of a disorder. As nicely
elucidated by Kraemer (2012), there obviously could be direct genetic effects and/
or direct environmental effects on a disorder. Genes could moderate the effects of
environment, and environment could mediate the effects of genes. Or the environ-
ment could be a proxy for genetic effects and not otherwise related to the disorder.
Given few findings to date, we are interested in all ways where genes or genes plus
environment may contribute to psychiatric disorders and may lead us to better
treatments.
Since the publication of Ioannidiss paper, Why most published research find-
ings are false in 2005 (Ioannidis, 2005) biomedical research has been sensitized to
the many ways that hypothesis tests can meet statistical criteria for significance and
be replicated and still not be correct. All of the problems resulting in findings that
ultimately dont replicate are critical in evaluating candidate gene and candidate
GxE (cGxE) studies. In aggregate, they raise the meaningful possibility that none of
the current findings are true and make it likely that, at most, few of them are.
The following issues make candidate gene studies particularly challenging:
Enormous numbers of reasonable hypotheses exist given the size of the human
genome and the hundreds of reasonable environmental measures that might
influence course, so each hypothesis tested has a low or very low probability of
being true. Many hypotheses will be tested and only those which are deemed
statistically significant will be reported (Duncan & Keller, 2011). Therefore,
most significant P values are likely to be Type I statistical errors rather than
statistical confirmation of a true hypothesis.
There is a strong publication bias for significant findings. Investigators exam-
ine many different potential gene and environmental effects and virtually all of
those that are reported are the ones that are statistically significant.
Few papers report the myriad ways that a particular dataset may have been exam-
ined before reporting the analysis with greatest significance (sometimes called
P-hacking) (Nuzzo, 2014).
There is a strong bias toward publication of positive replication studies over neg-
ative replication studies. In many cases, an examination of an existing dataset for
potential replication which does not replicate will not even be submitted for
publication.
8 Do Gene-by-Environment Interactions Offer Potential Intervention Strategies 149
The power to detect GxE interactions is lower than that to detect genetic effects
alone (Garcia-Closas & Lubin, 1999).
The inclusion of indirect GxE replications, replications differing in one or
more critical aspects, e.g., specific polymorphism or specific environmental
moderator tested, are problematic and may be associated with extremely high
Type I error rates (Duncan & Keller, 2011; Sullivan, 2007).
Incorrect control for potential confounders may artifactually suggest GxE inter-
action where none exists (Keller, 2014).
At present, there are no universally accepted candidate gene GxE effects in psy-
chiatry (Duncan & Keller, 2011; Psychiatric GWAS Consortium Coordinating
Committee etal., 2009).
Large (of the order of 25,000 subjects or larger) genome-wide association studies
(GWAS) of anxiety disorders may ultimately provide part of the solution to this
problem. GWAS studies can also test sets of functionally related genes involved in
a specific biological pathway which offers more potential power than the examina-
tion of individual genes. Available GWAS studies of anxiety traits strongly suggest
that they are highly polygenic (thousands of genes) and that each of the individual
genes has a very modest effect (Duncan, Pollastri, & Smoller, 2014).
Serotonin Transporter Polymorphisms (SERT)
The serotonin transporter is an energy-dependent system moving serotonin from the

synaptic cleft into the presynaptic neuron. The promoter region of the gene has a
region with 14 (short) or 16 (long) tandem repeats though there are allelic varia-
tions with up to 22 repeats. The short repeat version is associated with less gene
transcription.
The GxE interaction of serotonin transporter (SERT) promoter polymorphisms with
environmental adversity in depression is the single most studied candidate gene GxE
effect in psychiatry since the groundbreaking study by Caspi etal. a decade ago (Caspi
etal., 2003; Uher, 2014). We briefly consider it because it provides a clear and poignant
example of the issues discussed above. Despite a number of published replications, it
is at best unclear whether or not the aggregate of evidence to date supports the Caspi
findings. Examining only direct GxE replications, which use the same statistical model
and outcome variables, the findings dont replicate (Munafo, Durrant, Lewis, & Flint,
2009; Risch etal., 2009). One meta-analysis that included many indirect replication
studies (e.g., studies using fundamentally different measures related to stress or funda-
mentally different outcome measures) did conclude that SERT moderates the relation-
ship between stress and depression (Karg, Burmeister, Shedden, & Sen, 2011).
Meta-analyses which include indirect replications are problematic because of the much
greater opportunity for P-hacking, unpublished failures to replicate, and negative
publication bias against failures to replicate (Duncan & Keller, 2011). More recently, it
has been suggested that by restricting the analysis to childhood maltreatment, rather
than stressful events in general, and restricting the outcome to chronic depression in
150 N. Ryan
adulthood, the evidence for the GxE effect becomes much stronger (Uher, 2014). But
again, finding a better-fitting model once all the data is known maximizes opportunities
to over fit the data and can only provide hypotheses for future testing.
A small number of studies looking at serotonin transporter promoter region variants
have found GxE effects with family adversity and stressful life events and development
of anxiety. These studies have found greater anxiety in those with S/S phenotype and
sexual abuse (Cicchetti, Rogosch, & Sturge-Apple, 2007) and, conversely, greater anx-
iety with L/L genotype and stress in another study (Laucht etal., 2009).
Several studies have looked at functional serotonin transporter variants and anxi-
ety sensitivity, a measure of cognitive vulnerability to anxiety. Unfortunately, two
relatively large studies had exactly opposing findings with one finding S/S individu-
als with maltreatment had greater anxiety sensitivity (Stein, Schork, & Gelernter,
2008) while the other found that L/L individuals with greater history of trauma had
greater anxiety sensitivity (Klauke etal., 2011).
Serotonin transporter genotype is not reliably associated with CBT response in
anxiety. This is indicated by two studies that show better response in individuals
with SS genotype, one study showing worse response, and six studies showing no
effects of SERT genotype on CBT response (Hudson etal., 2013).
fMRI studies of SERT and fear conditioning have found that S/S adults have
increased activity in multiple brain areas including amygdala, insula, thalamus, and
occipital cortex (Klucken etal., 2015).
Mouse animal models of fear conditioning in mice after inescapable shock find
impaired fear learning in serotonin transporter knockout mice (both / and /+)
compared to normal mice (+/+) (Muller etal., 2011). In a comparison of serotonin
knockout mice raised with: (1) early adversity (stressed mothers); (2) normal early
life (standard nesting conditions), and (3) enriched early life (communal nesting),
there were main effects of both genotype (/ mice had highest state and trait anxi-
ety) and adversity (mice raised with early adversity showed lower trait anxiety) but
no interaction between genotype and early adversity (Kloke etal., 2013).
Genotype may influence treatment response whether or not they are related to
development of disorder. Fox and colleagues found that La/La SERT genotype healthy
adults developed less biases toward both positive and negative stimuli in an attention
bias modification paradigm than did those with S/S, S/Lg, and Lg/Lg (reduced trans-
porter activity) genotypic individuals (Fox, Zougkou, Ridgewell, & Garner, 2011).
Lonsdorf and colleagues found that SERT S/S and S/L had greater acquisition of fear
potentiated startle than adults with L/L genotype (Lonsdorf etal., 2009).
In summary, evidence for SERT GxE effects in anxiety disorders is modest.
BDNF
Brain derived neurotrophic factor (BDNF) is a neurotrophic growth factor found in

the brain, including hippocampus and cortex, and in the periphery. It binds to TrkB
and LNGFR.It is involved in signaling, synaptogenesis (in the hippocampus), and
synaptic stability. It plays an important role in dendritic remodeling in the hippo-

campus and basolateral amygdala, areas important in fear systems. Acute stress is
associated with a 10-day rise in BDNF expression in the BLA (McEwen et al.,
2015). BDNF is necessary for formation of emotional memories including fear
memories and BDNF enhances fear extinction (Andero & Ressler, 2012). Several
studies have found that the Vall66Met BDNF polymorphism in individuals with
early life stress is associated with higher depression and anxiety along with changes
in size of hippocampus and amygdala (Gatt etal., 2009; La Greca, Lai, Joormann,
Auslander, & Short, 2013).
COMT
Catechol-O-methyltransferase (COMT) degrades (inactivates) dopamine, epineph-

rine, and norepinephrine. COMT degradation of dopamine is thought to be impor-
tant in the prefrontal cortex, which has relatively lower levels of dopamine
transporter (Matsumoto etal., 2003).
One study suggests that the COMT val158met polymorphism met/met genotype
(which has decreased enzyme activity) is associated with decreased fear extinction
(Lonsdorf etal., 2009). Another found a possible GxE effect with val158met poly-
morphism met/met genotype and high early life adversity scores associated with
higher anxiety sensitivity (Baumann etal., 2013).
Putting It All Together
Given the very strong evidence against the existence of genes of medium or large
effect size in anxiety disorders how much does this constrain possible GxE effect
sizes? This depends on whether the interaction term results in a crossover effect. No
GxE studies in psychiatry have yet demonstrated crossover (Duncan etal., 2014).
For example, in the Caspi study the S/S genotype was associated with greater
depression in those with a high burden of stressful life events but was not associated
with less depression on those with low burden of stressful events (Caspi et al.,
2003). Ellis and others have put forward a differential susceptibility hypothesis
they hypothesize that some individuals are genetically more susceptible to both
favorable and unfavorable environments while others are relatively impervious to
environment (Ellis & Boyce, 2011; Ellis, Boyce, Belsky, Bakermans-Kranenburg,
& van Ijzendoorn, 2011). While this effect has not yet been clearly proven to occur
in psychiatric disorders, it would permit large GxE effects even in the absence of
any overall genetic effect!
Therefore, it may be that among the studies briefly reviewed above one or several
have identified a true GxE interaction related to anxiety disorders. Or perhaps not.
Given the known issues with p-hacking, lack of reporting of all tests and compari-
152 N. Ryan
sons, publication bias, low prior likelihood of most GxE hypotheses being true, and
the other issues discussed above, it is impossible to say.
GWAS studies in psychiatric disorders (excluding Alzheimers) have found
odd ratios less than 1.25 for the strongest common risk variants and most of the
risk variants found have been in unexpected places, not in neurotransmitter-related
genes and most in intergenetic and intronic regions of the genome (Duncan etal.,
2014). As discussed, it is theoretically possible that there are significant GxE
effects involving genes with no direct gene effect or that there are GxE effects of
greater magnitude than any direct genetic effects but these are both thought to be
implausible (Duncan etal., 2014). Risch has argued that cGxE effects should be
investigated only after identification of a genetic main effect (Risch etal., 2009).
So candidate GxE studies available today in anxiety disorders (and psychiatric
disorders in general) are almost certainly significantly underpowered.
There is hope on the horizon. First, as GWAS studies of anxiety disorder are
completed and published the genes found in these studies will be excellent candi-
date genes for cGxE studies. As we have seen with schizophrenia and other psychi-
atric GWAS studies, many most of these genes are likely to be surprising and in
areas that have received little attention (Schizophrenia Working Group of the
Psychiatric Genomics, C., 2014). Second, genome-wide GxE studies (GWIS) are
possible as even larger aggregate datasets become available.
There is a second reason for optimism. We can aggregate potential genetic and
environmental risk factors together to arrive at a robust aggregate risk score in the
absence evidence of statistically significant contribution of any of the individual
factors. Straightforward linear approaches have been used to combine genetic fac-
tors into an overall risk score (Belsky etal., 2013) and to combine demographic
and clinical factors into a score predicting treatment response to two different treat-
ments (Wallace etal., 2013). Other approaches, e.g., random forest machine learn-
ing, can be used to combine genetic and environmental factors including potential
interaction terms to predict risk or treatment response even when there are more
variables than subjects (Kim et al., 2015; McKinney, Reif, Ritchie, & Moore,
2006). This may allow us to better target the treatment to the individual even with-
out knowing which individual genetic and environmental factors are important in
that individual.
Viewing genetic and environmental effects related to anxiety disorders as large
number of relatively small effects with potential complicated interactions it isnt
surprising that the new algorithmic approaches may give us good prediction without
giving us understanding of deep causes. In fact, it is the best we can do if all the
effects are smallin that case, there is no identifiable main cause for an individual
developing anxiety disorder, just risk that aggregates from many separate factors
and their potentially complex interactions.
While the discovery of fundamentally new pharmacological treatments for psy-
chiatric disorders has been distressingly hard, there is also the chance that newly
identified genetic correlates of disorder will elucidate fundamentally new paths to
disorder with potential treatment insights.
References
Andero, R., & Ressler, K. J. (2012). Fear extinction and BDNF: Translating animal models of
PTSD to the clinic. Genes, Brain, and Behavior, 11(5), 503512.
Baumann, C., Klauke, B., Weber, H., Domschke, K., Zwanzger, P., Pauli, P., Reif, A. (2013).
The interaction of early life experiences with COMT val158met affects anxiety sensitivity.
Genes, Brain, and Behavior, 12(8), 821829.
Belsky, D. W., Moffitt, T. E., Sugden, K., Williams, B., Houts, R., McCarthy, J., & Caspi, A.
(2013). Development and evaluation of a genetic risk score for obesity. Biodemography and
Social Biology, 59(1), 85100.
Casey, B.J., Glatt, C.E., & Lee, F.S. (2015). Treating the developing versus developed brain:
Translating preclinical mouse and human studies. Neuron, 86(6), 13581368.
Caspi, A., Sugden, K., Moffitt, T.E., Taylor, A., Craig, I.W., Harrington, H., Poulton, R. (2003).
Influence of life stress on depression: moderation by a polymorphism in the 5-HTT gene.
Science, 301(5631), 386389.
Cicchetti, D., Rogosch, F.A., & Sturge-Apple, M.L. (2007). Interactions of child maltreatment
and serotonin transporter and monoamine oxidase A polymorphisms: Depressive symptom-
atology among adolescents from low socioeconomic status backgrounds. Development and
Psychopathology, 19(4), 11611180.
environment interaction research in psychiatry. The American Journal of Psychiatry, 168(10),
10411049.
Duncan, L.E., Pollastri, A.R., & Smoller, J.W. (2014). Mind the gap: Why many geneticists and
psychological scientists have discrepant views about gene-environment interaction (GxE)
research. The American Psychologist, 69(3), 249268.
Ellis, B. J., & Boyce, W. T. (2011). Differential susceptibility to the environment: toward an
understanding of sensitivity to developmental experiences and context. Development and
Psychopathology, 23(1), 15.
Ellis, B. J., Boyce, W. T., Belsky, J., Bakermans-Kranenburg, M. J., & van Ijzendoorn, M. H.
(2011). Differential susceptibility to the environment: An evolutionaryNeurodevelopmental
theory. Development and Psychopathology, 23(1), 728.
Fox, E., Zougkou, K., Ridgewell, A., & Garner, K. (2011). The serotonin transporter gene alters
sensitivity to attention bias modification: Evidence for a plasticity gene. Biological Psychiatry,
70(11), 10491054.
Garcia-Closas, M., & Lubin, J.H. (1999). Power and sample size calculations in case-control stud-
ies of gene-environment interactions: Comments on different approaches. American Journal of
Epidemiology, 149(8), 689692.
Gatt, J.M., Nemeroff, C.B., Dobson-Stone, C., Paul, R.H., Bryant, R.A., Schofield, P.R.,
Williams, L.M. (2009). Interactions between BDNF Val66Met polymorphism and early life
stress predict brain and arousal pathways to syndromal depression and anxiety. Molecular
Psychiatry, 14(7), 681695.
Hudson, J.L., Lester, K.J., Lewis, C.M., Tropeano, M., Creswell, C., Collier, D.A., Eley, T.C.
(2013). Predicting outcomes following cognitive behaviour therapy in child anxiety disorders:
the influence of genetic, demographic and clinical information. Journal of Child Psychology
and Psychiatry, 54(10), 10861094.
Ioannidis, J.P. (2005). Why most published research findings are false. PLoS Medicine, 2(8), e124.
Karg, K., Burmeister, M., Shedden, K., & Sen, S. (2011). The serotonin transporter promoter vari-
ant (5-HTTLPR), stress, and depression meta-analysis revisited: Evidence of genetic modera-
tion. Archives of General Psychiatry, 68(5), 444454.
Keller, M. C. (2014). Gene x environment interaction studies have not properly controlled for
potential confounders: The problem and the (simple) solution. Biological Psychiatry, 75(1),
1824.
154 N. Ryan
Kim, J.W., Sharma, V., & Ryan, N.D. (2015). Predicting methylphenidate response in ADHD
using machine learning approaches. The International Journal of Neuropsychopharmacology,
18(11), pyv052.
Klauke, B., Deckert, J., Reif, A., Pauli, P., Zwanzger, P., Baumann, C., Domschke, K. (2011).
Serotonin transporter gene and childhood traumaA G x E effect on anxiety sensitivity.
Depression and Anxiety, 28(12), 10481057.
Kloke, V., Heiming, R. S., Blting, S., Kaiser, S., Lewejohann, L., Lesch, K. P., & Sachser, N.
(2013). Unexpected effects of early-life adversity and social enrichment on the anxiety profile of
mice varying in serotonin transporter genotype. Behavioural Brain Research, 247, 248258.
Klucken, T., Schweckendiek, J., Blecker, C., Walter, B., Kuepper, Y., Hennig, J., & Stark, R.
(2015). The association between the 5-HTTLPR and neural correlates of fear conditioning and
connectivity. Social Cognitive and Affective Neuroscience, 10(5), 700707.
Kraemer, H.C. (2012). Determining gene moderation of environmental risk factors for a mental
disorder: A perfect storm of methodological problems. International Journal of Methods in
Psychiatric Research, 21(3), 185194.
La Greca, A.M., Lai, B.S., Joormann, J., Auslander, B.B., & Short, M.A. (2013). Childrens risk
and resilience following a natural disaster: Genetic vulnerability, posttraumatic stress, and
depression. Journal of Affective Disorders, 151(3), 860867.
Laucht, M., Treutlein, J., Blomeyer, D., Buchmann, A.F., Schmid, B., Becker, K., Banaschewski,
T. (2009). Interaction between the 5-HTTLPR serotonin transporter polymorphism and envi-
ronmental adversity for mood and anxiety psychopathology: Evidence from a high-risk com-
munity sample of young adults. The International Journal of Neuropsychopharmacology,
12(6), 737747.
Levy, D., etal. (2009). Genome-wide association study of blood pressure and hypertension. Nature
Genetics, 41(6), 677687.
Lonsdorf, T. B., Weike, A. I., Nikamo, P., Schalling, M., Hamm, A. O., & Ohman, A. (2009).
Genetic gating of human fear learning and extinction: Possible implications for gene-
environment interaction in anxiety disorder. Psychological Science, 20(2), 198206.
Matsumoto, M., Weickert, C. S., Akil, M., Lipska, B. K., Hyde, T. M., Herman, M. M.,
Weinberger, D.R. (2003). Catechol O-methyltransferase mRNA expression in human and rat
brain: evidence for a role in cortical neuronal function. Neuroscience, 116(1), 127137.
McEwen, B.S., Bowles, N.P., Gray, J.D., Hill, M.N., Hunter, R.G., Karatsoreos, I.N., & Nasca,
C. (2015). Mechanisms of stress in the brain. Nature Neuroscience, 18(10), 13531363.
McKinney, B.A., Reif, D.M., Ritchie, M.D., & Moore, J.H. (2006). Machine learning for detect-
ing gene-gene interactions: a review. Applied Bioinformatics, 5(2), 7788.
Merikangas, K.R., & Risch, N. (2003). Will the genomics revolution revolutionize psychiatry?
The American Journal of Psychiatry, 160(4), 625635.
Muller, J.M., Morelli, E., Ansorge, M., & Gingrich, J. A. (2011). Serotonin transporter deficient
mice are vulnerable to escape deficits following inescapable shocks. Genes, Brain, and
Behavior, 10(2), 166175.
Munafo, M.R., Durrant, C., Lewis, G., & Flint, J.(2009). Gene X environment interactions at the
serotonin transporter locus. Biological Psychiatry, 65(3), 211219.
Nuzzo, R. (2014). Scientific method: Statistical errors. Nature, 506(7487), 150152.
Psychiatric GWAS Consortium Coordinating Committee, Cichon, S., Craddock, N., Daly, M.,
Faraone, S. V., Gejman, P. V., Sullivan, P. F. (2009). Genomewide association studies:
History, rationale, and prospects for psychiatric disorders. The American Journal of Psychiatry,
166(5), 540556.
Risch, N., Herrell, R., Lehner, T., Liang, K.Y., Eaves, L., Hoh, J., Merikangas, K.R. (2009).
of depression: a meta-analysis. JAMA, 301(23), 24622471.
Schizophrenia Working Group of the Psychiatric Genomics, C (2014). Biological insights from
108 schizophrenia-associated genetic loci. Nature, 511(7510), 421427.
Stein, M.B., Schork, N.J., & Gelernter, J.(2008). Gene-by-environment (serotonin transporter
and childhood maltreatment) interaction for anxiety sensitivity, an intermediate phenotype for
anxiety disorders. Neuropsychopharmacology, 33(2), 312319.
Sullivan, P.F. (2007). Spurious genetic associations. Biological Psychiatry, 61(10), 11211126.
Uher, R. (2014). Gene-environment interactions in common mental disorders: An update and strat-
egy for a genome-wide search. Social Psychiatry and Psychiatric Epidemiology, 49(1), 314.
Wallace, M.L., Frank, E., & Kraemer, H.C. (2013). A novel approach for developing and inter-
preting treatment moderator profiles in randomized clinical trials. JAMA Psychiatry, 70(11),
12411247.
Chapter 9
Challenge andPotential forResearch onGene-
Environment Interactions inAutism Spectrum
Disorder
CarlyL.A.Wender andJeremyVeenstra-VanderWeele
Introduction
Specific disease models may be useful to explore gene-environment interactions in

developmental psychopathology. Autism spectrum disorder (ASD) may be a good
target for such exploration even though treatment studies are lacking in strength and
specific association with potential etiologic substrates of ASD continue to emerge
and there are likely to be very many. However, attempts to examine gene-environ-
ment interactions and their potential impact have been an area of intense interest in
ASD research.
Autism spectrum disorder (ASD) is a neurodevelopmental disorder defined by
two categories of core symptoms: impairment in social functioning/communication
and restricted/repetitive behaviors (American Psychiatric Association, 2013). Within
these broad categories, specific symptoms are highly variable across affected indi-
viduals. Social communication symptoms range from a complete lack of social moti-
vation to intrusive behavior that disregards social cues (Leekam, Libby, Wing, Gould,
& Taylor, 2002). Repetitive behaviors can include simple motor m annerisms, intense
C.L.A. Wender
Department of Psychiatry, Columbia University Medical Center, New York, NY, USA
J. Veenstra-VanderWeele, M.D. (*)
Department of Psychiatry and Sackler Institute for Developmental Psychobiology, Columbia
University Medical Center, New York, NY, USA
Mortimer D.Sackler Associate Professor of Psychiatry, New York State Psychiatric Institute,
1051 Riverside Drive, Mail Unit 78, New York, NY 10032, USA
Center for Autism and the Developing Brain, New York Presbyterian Hospital,
New York, NY, USA
e-mail: veenstr@nyspi.columbia.edu

DOI10.1007/978-3-319-49227-8_9
158 C.L.A. Wender and J. Veenstra-VanderWeele
and restricted interests, or complex rituals (American Psychiatric Association, 2013;

Richler, Bishop, Kleinke, & Lord, 2007; Richler, Huerta, Bishop, & Lord, 2010).
Despite sharing a single diagnosis, no two people manifest symptoms of ASD in the
exact same way.
Co-occurring symptoms and disorders also contribute to the diversity within the
ASD population. Figure 9.1 shows common comorbidities and biomarkers associ-
ated with ASD. Common co-occurring disorders fall into three broad domains:
behavioral, medical, and cognitive. Behavioral disorders prominently include
hyperactivity, anxiety, and aggression, but can also include mood or psychotic dis-
orders. A number of medical disorders are more frequent in the ASD population,
including functional constipation, food allergies, and seizure disorders. A substan-
tial fraction of children with ASD have intellectual disability (broadly defined as
IQ < 70) and speech or language disorders, which are now separately diagnosed
Fig. 9.1. ASD symptoms, comorbidities, and biomarkers. The core symptoms are represented in
the middle. According to the DSM-5 criteria, to be diagnosed with ASD, you must display all three
social interaction/communication symptoms and two out of the four restricted/repetitive behavior
symptoms. Peripheral to the core symptoms are comorbidities and biomarkers that often accom-
pany ASD.These comorbidities span across cognitive, behavioral, and medical domains. These
symptoms are not diagnostic requirements, but they have been found to occur more frequently in
individuals with ASD than in the general population. Additional biomarkers and genetic findings
shown in the periphery reflect the significant heterogeneity of ASD
9 Challenge and Potential for Research on Gene-Environment Interactions in 159
from ASD in the DSM-5 (American Psychiatric Association, 2013). These various
co-occurring disorders can be used to identify subsets of individuals who may ben-
efit from a common treatment approach. Subgroups of individuals could also be
defined on the basis of biomarkers such as increased head growth, abnormal EEG,
elevated blood serotonin levels, and maternal autoantibodies (Veenstra-VanderWeele
& Blakely, 2012). At this point, subgroups based upon biomarkers have only been
studied in the research setting, and measurement of biomarkers is not yet clinically
indicated.
Most available treatments for ASD only address portions of the core or co-occurring
symptoms. Behavioral interventions are the most utilized and effective form of treat-
ment, with children showing improved communication or cognitive performance
(Howard, Sparkman, Cohen, Green, & Stanislaw, 2005; Sallows & Graupner, 2005).
Most children, however, are left with significant impairments despite early intensive
behavioral intervention (Warren etal., 2011). Medications can be helpful in treating
co-occurring symptoms, but they do not address core symptoms (McPheeters etal.,
2011). The extreme heterogeneity observed in ASD hinders the discovery of effective
treatments. As opposed to common risk factors that might contribute in the majority
of individuals, most robust ASD risk factors, such as fragile X syndrome, tuberous
sclerosis complex, or prenatal exposure to valproic acid, are only present in less than
12% of the total ASD population (Christensen etal., 2013; Ehninger etal., 2012;
Hessl etal., 2001). Current research has largely focused on individual genetic and
environmental risk factors, but there are certain to also be interactions between these
risk factors. These unknown gene-gene, gene-environment, or environment-
environment risk factor interactions can be thought of as an impediment to treatment
development. In the long run, however, these interactions may present critical oppor-
tunities for new treatments since treatments that target interactions could have dra-
matic effects on ASD risk or outcomes.
This chapter outlines the evolution of ASD risk factor research. Much of the
initial focus in this area was on genetic risk factors, but key findings have also high-
lighted the importance of environmental influence. To date, the more fruitful
research has focused on rare but robust factors with clear evidence for risk in a small
subset of individuals with ASD.Common risk factors have been studied as well, but
with less informative results to date, both on the genetic and environmental fronts
(Chaste & Leboyer, 2012). A very small number of studies, reviewed below, have
begun to explore the interaction between individual risk factors.
Risk Factors Expected toContribute toASD
Early twin studies estimated that ASD was primarily genetically determined as indicated
by a large gap in concordance between monozygotic (MZ) and dizygotic (DZ) twins.
Initially, Folstein and Rutter found a 36% concordance rate for autism and an 82%
concordance for a broader range of cognitive and language impairment in MZ twins
and a 0% and 10% concordance rate in MZ and DZ twins, respectively (Folstein &
Rutter, 1977). Almost 20 years later, Bailey and colleagues reported a 60% concor-
dance rate for autism and 92% for autism spectrum symptoms in MZ twins versus a
010% concordance rate in DZ twins (Bailey etal., 1995). While initial high herita-
bility estimates provided strong support for the importance of genetic risk factors,
they also suggested that at least 10% of ASD comes from a different set of factors
and did not rule out broad environmental risk factors shared across the population.
One solitary cause of ASD therefore seemed unlikely. While they were carefully
conducted, these early twin studies were challenged by small sample sizes and
applied different diagnostic criteria than are in use today.
Recent twin studies have found somewhat lower heritability estimates. Hallmayer
etal. (2011), in particular, found higher DZ concordance rates than any other study.
They reported concordance rates for male DZ pairs with strict autism at 21% and
ASD at 31%, with even higher rates observed in female DZ twin pairs. Male MZ
twin pairs did show higher concordance rates (58% for autism and 77% for ASD),
but the high concordance rate among DZ twins suggests a more significant shared
environmental influence than estimated in other studies. Overall, this study calcu-
lated 37% genetic and 55% shared environmental risk for strict autism, and 38%
genetic and 58% shared environmental risk for ASD (Hallmayer etal., 2011). This
work has not been replicated and these heritability estimate remain well below that
of any other twin study. Nonetheless, these findings point to the importance of
studying shared environmental factors in ASD.
A recent integrated twin and family study points to the potential importance of
non-shared environment, which encompasses anything excluded from genetic or
shared environmental factors, including interaction effects. Sandin etal. (2014) used
a large-scale investigation of over two million children in Sweden to suggest 50%
heritability and 50% non-shared environmental influence in ASD.Concordance was
assessed for a multitude of sibling relationships: 54% for MZ twins, 25% for DZ
twins, 25% for full siblings, 11% for maternal half siblings, and 7% for paternal half
siblings (Sandin etal., 2014). In 2012, Czyz and colleagues sought to identify poten-
tial sources for the relatively high discordance of MZ twins observed in many stud-
ies, like that of Sandin and colleagues. This study investigated a myriad of possible
factors including phenotypic variability, in utero environment, de novo mutations,
stochasticity, and epigenetics. In support of Sandins findings, this study suggested
that non-shared environmental factors play a significantly higher role in ASD than
shared environmental factors (Czyz, Morahan, Ebers, & Ramagopalan, 2012).
Although non-shared environmental risk seems to play an important role in ASD,
much less is understood about what specific factors might underlie these findings
(Plomin, 2011). Clearly, unequal blood distribution in the uterus or the use of for-
ceps for one but not the other twin would result in non-shared environmental risk.
Less clear is the influence of stochastic events in brain development on the variabil-
ity of initial phenotype (Stromswold, 2006). What is the impact of an axon turning
1m earlier or later on one neuron firing slightly less in sync with its neighbor?
Early phenotypic variability may become amplified in interactions with the
environment (Asbury, Almeida, Hibel, Harlaar, & Plomin, 2008). As one simple
example, one MZ twin may initially show slightly more attention to social interac-
tions than the other, but the resulting increase in social experience may start a posi-
tive feedback loop of social reinforcement that results in a steady increase in social
engagement over time.
Genetic Risk Factors
An accelerating array of research has identified rare genetic risk factors in

ASD. Several comprehensive reviews compile recent genetic findings in ASD
(Devlin & Scherer, 2012; Huguet, Ey, & Bourgeron, 2013; Jeste & Geschwind,
2014; Ronemus, Iossifov, Levy, & Wigler, 2014; Satterthwaite & Baker, 2014).
Most genetic risk in ASD is likely due to common variants, but convincing evidence
for individual common variants is difficult to find because they likely have very
small effect sizes and may only have a larger impact when acting in concert. In
contrast, rare variants have much larger effect sizes but occur in a very small subset
of affected individuals. One hope is that rare variants will converge upon common
systems or pathways that may be perturbed in a larger group of patients. At this
time, evidence for this convergence is limited.
Reflecting the high number of co-occurring medical conditions (Fig. 9.1), a care-
ful medical evaluation will sometimes identify a genetic syndrome in a child who is
being evaluated for ASD.Over time, these rare syndromes may become treatable, at
least in part, motivating careful genetic testing. As an example, a few syndromes
due to single gene disruptions converge on disinhibition of the mammalian target of
rapamycin (mTor) pathway, including PTEN hamartoma syndrome, tuberous scle-
rosis (TSC), and neurofibromatosis (Lipton & Sahin, 2014). Collectively, these syn-
dromes may account for about 1% of children with ASD.In theory, rapamycin and
related medications may lead to improvements in cognitive or ASD-related out-
comes in these disorders (Ehninger, 2013); although this is only now being tested in
randomized, controlled trials (Sahin, 2012). The convergence of a few syndromes
on a common pathway, however, hints at the possibility that a larger group of
patients could similarly have disinhibition of mTor due to other causes.
Fragile X syndrome (FXS) is the most common, known genetic syndrome in
ASD, found in approximately 12% of affected children. Conversely, approxi-
mately 3060% of individuals affected with fragile X meet the criteria for ASD
(Hall, Lightbody, Hirt, Rezvani, & Reiss, 2010). FXS, therefore, is a strong genetic
risk factor of ASD, rather than a deterministic cause of ASD.The strongest and
most common characteristic of FXS is intellectual disability. In addition to ASD
symptoms, hyperarousal, social anxiety and withdrawal, and attention problems are
commonly seen as well. The variability of ASD symptoms in FXS indicates that
other genetic or environmental factors contribute to ASD risk within affected indi-
viduals, potentially suggesting that social and repetitive behaviors may be particu-
larly amenable to treatment within this syndrome. FXS is caused by an expanding
trinucleotide repeat in the 5 untranslated region of the FMR1 gene, with repeat
lengths of greater than 200 leading to gene hypermethylation and silencing (Ashley
etal., 1993; Kremer etal., 1991; ODonnell & Warren, 2002; Verkerk etal., 1991).
Two decades of research have led to multiple hypotheses connecting loss of FMRP,
the corresponding protein, to abnormalities in dendritic spine plasticity and result-
ing changes in learning and behavior. Most data suggest excessive, dysregulated
signaling downstream of a number of G-protein-coupled receptors, including the
metabotropic glutamate mGlu5 receptor (Bear, Huber, & Warren, 2004; Dolen &
Bear, 2008). Potential treatments are currently being tested based upon this and
other hypotheses (Berry-Kravis, 2014).
Over the past decade, several copy number variants (CNVs) have been identified
as risk factors in ASD.CNVs are inherited or de novo genetic alterations caused by
insertions or deletions of pieces of DNA ranging from 1kb to several Mb in length.
These variants are usually flanked by common, repeating boundaries. Working in
parallel, Levy etal. (2011) and Sanders etal. (2011) found a significantly higher
rate of de novo CNVs in ASD probands (e.g., 75 events in 889 children) than in their
unaffected siblings (e.g., 19 events in 895 children). These variants were also more
likely to disrupt genes in children affected with ASD than in their unaffected sib-
lings. Almost four times as many affected children had gene-disrupting de novo
copy number events compared to their siblings. Levy etal. (2011) identified 12 loci,
in particular, with recurrent de novo events in ASD.The presence of these loci is
suggestive of specific genetic targets implicated in ASD.These loci could become
possible targets for treatment, but they typically include multiple genes. Further
investigation will be necessary to understand the impact on downstream networks or
signaling pathways.
Recent advances in sequencing technology have allowed the identification of de
novo single nucleotide variants (SNVs) in ASD.Again, working in parallel, Sanders
etal. (2012), and ORoak etal. (2012) found that likely gene-disrupting (LGD) de
novo SNVs are approximately twice as common in ASD as in unaffected siblings,
which was also true for single base insertion/deletion mutations. Based on these
studies, the total number of LGD mutations (insertion/deletion plus point muta-
tions) contributes to ASD in roughly 10% of affected children, but most disrupted
genes are only identified in a single individual. Based upon the rarity of de novo
LGD events, genes that are disrupted in more than two or three individuals with
ASD very likely contribute to the phenotype. The list of such genes continues to
grow, prominently including genes encoding synaptic proteins and chromatin-mod-
ifying proteins (Iossifov etal., 2014; ORoak etal., 2014; Poultney etal., 2014).
Based upon current models, as many as 1000 genes are likely to contribute to ASD
via these rare, de novo events (Sanders et al., 2012). While this information is
clearly important in understanding ASD risk, it is not yet clear how to move from
these rare disruptions in a few individuals to potential treatments that would help
more than a handful of children.
Common genetic variants likely contribute to the majority of ASD cases, although
these variants remain largely unknown. By examining data from common variation
association studies, Klei etal. (2012) estimated heritability in two sets of families:
Simons Simplex Collection (SSC) and Autism Genome Project (AGP). Of note, none
of the SSC parents met ASD criteria; whereas a small portion of the AGP parents
were known to have ASD, while others were not assessed for ASD. The narrow-
sense heritability estimate was highest for AGP multiplex families (66%), followed
by AGP simplex families (50%), followed by SSC families (40%), which are simplex
by definition. The cascade of estimates suggests that heritability is higher when more
individuals are affected within a family. Using analyses of individual chromosomes,
Klei and colleagues suggest that multiple variants of small effect likely contribute to
risk in an additive fashion, rather than a few variants of larger effect across the
genome (Klei etal., 2012). Other recent studies similarly support these findings, sug-
gesting that the contribution of common variants is collectively much larger than the
contribution of rare variants (Anney etal., 2012; Gaugler etal., 2014). A much larger
sample size than has been analyzed to date, likely on the order of tens of thousands,
will be required to identify these common variants individually (Schizophrenia
Working Group of the Psychiatric Genomics, 2014).
Environmental Risk Factors
Rare prenatal events and exposures suggest a significant role for environmental risk
factors in ASD.The clearest examples are prenatal exposures to valproic acid, tha-
lidomide, and the rubella virus. Valproic acid (VPA) is an antiepileptic drug that is
also commonly used for bipolar disorder and persistent migraines. Christensen etal.
(2013) found significant association between VPA and ASD in Danish children. The
risk of having ASD after prenatal VPA exposure was 4.42%, which yields an
adjusted hazard ratio of 2.8 compared to the general population. More narrowly
defined autistic disorder yielded a higher hazard ratio of 4.2. To control, in part, for
the medical illnesses that led to VPA use, they verified that the risk of ASD was
higher in children whose mothers used VPA during pregnancy as opposed to moth-
ers who stopped using the drug at least 30days prior to conception. Furthermore,
there was a higher risk found in mothers with a history of epilepsy who used valpro-
ate during pregnancy than in mothers with a history of epilepsy who used a different
drug or no drug at all (Christensen etal., 2013).
Thalidomide was initially touted as a potent sedative and antiemetic but was
taken out of use in most countries after several harmful side effects were discovered.
Stromland, Nordin, Miller, Akerstrom, and Gillberg (1994) examined 100 Swedish
citizens who suffered from thalidomide embryopathy (prenatal exposure to thalido-
mide). Of these participants, four were confirmed to have autism, and three had
severe autism. This prevalence (45%) was substantially higher than the preva-
lence of autism in the general population, which at the time was 0.08% (Stromland
etal., 1994). Since thalidomide is no longer in common use, larger sample sizes are
not available to better understand its impact on ASD risk.
In 1971, Stella Chess conducted a study on 243 children with congenital rubella,
and found 18 who were affected with autism spectrum disorder, based upon her
criteria at the time. This rate was markedly higher than the rate of autism reported
in surveys at the time (0.72.1 per 10,000) (Chess, 1971). In a study conducted
6years later, she continued to identify a high prevalence of autism, but also a num-
ber of cases that no longer met her criteria for ASD, either due to successful behav-
ioral treatment or symptoms that were attributed to more general cognitive
impairment. Of note, congenital rubella syndrome also causes visual and auditory
sensory loss, as well as intellectual disability, but these symptoms did not appear to
account for the ASD symptoms in Chess analysis (Chess, 1977).
More common environmental risk factors have also been implicated in ASD, but
with a smaller effect size. Both advanced paternal and maternal age are robust risk fac-
tors for ASD, but these most likely reflect, in part, an increase in de novo copy number
variants (paternal and maternal) and accumulated de novo mutations (paternal) (Dong
etal., 2014; Kolevzon, Gross, & Reichenberg, 2007; Lampi etal., 2013; Mamidala
etal., 2013). Very low birth weight is likely the most robust environmental risk factor,
along with extreme preterm birth, which are difficult to separate from one another
(Mann, McDermott, Bao, Hardin, & Gregg, 2010; Nosarti etal., 2012). A wide variety
of factors relating to adverse birth events or neonatal distress have been implicated as a
group, including birth injury, meconium aspiration, and ABO or Rh incompatibility
(Gardener, Spiegelman, & Buka, 2011). A number of prenatal risk factors have also
been implicated, including maternal diabetes, lack of folate supplementation, and short
interpregnancy interval (Al-Farsi etal., 2013; Cheslack-Postava etal., 2014; Schmidt
etal., 2012; Xu, Jing, Bowers, Liu, & Bao, 2014).
Researchers have suggested a multitude of other environmental factors that could
be associated with a heightened risk of ASD, including various forms of environ-
mental pollution (Landrigan, 2010; London, 2000; Windham etal., 2011). At this
point, replication studies have not yet provided clear support for specific chemical
or other environmental exposures.
Gene-Environment Interaction Effects
Gene-environment interactions are a new avenue for understanding the complexities

of ASD risk. An interaction effect exists if the effect of both factors occurring simul-
taneously is greater than the sum of individual risks posed by either factor. As shown
in Fig. 9.2, a common gene variant X and a prenatal environmental factor Y may both
have weak associations with ASD on their own, which could entice researchers to
investigate these factors together. The interaction effect is the increase in risk when
the two factors occur together. Interaction effects are very likely to contribute to ASD
risk, but thus far, they have been minimally studied.
A few approaches can logically be taken to study gene-environment interactions
based upon the current state of knowledge about risk factors, as reviewed above.
These approaches, along with examples of each, will be described below; although
none of them have yet yielded clear support for a robust gene-environment interac-
tion effect in ASD.The first approach is to conduct an association study using a
wide variety of environmental risk factors in a specific subset of affected individuals
with a rare genetic syndrome linked to ASD.The converse approach is to conduct a
Fig. 9.2. Gene-Environment Interaction Effects. X represents a common gene variant and Y rep-
resents a prenatal environmental risk factor. Each of these factors heightens the risk of ASD on
their own, as shown in red (X) and blue (Y). When X and Y occur simultaneously, they create an
interaction effect (green). The observed effect from XY co-occurring is the combined effect of X
plus Y plus the interaction effect
genome-wide association study (GWAS) on a specific subset of affected individuals

who share a robust environmental risk factor. A third approach is to investigate more
common risk factors that have smaller effect sizes when studied alone.
I nfluence ofEnvironment inRare Genetic Syndromes That

Predispose toASD
As noted above, fragile X syndrome (FXS) is a robust risk factor for ASD, but some
additional risk factors must contribute since a large fraction of people with FXS do
not have ASD.Rare disorders known to heighten the risk of ASD may therefore
provide researchers with a unique opportunity to look at different environmental
factors that may play a decisive role when a robust genetic risk factor is present
across a population. Hessl etal. (2001) sought to investigate several environmental
factors that may interact with loss of expression of the fragile X syndrome gene
FMR1. Since the FMR1 expanding trinucleotide repeat polymorphism that contrib-
utes to FXS acts in an X-linked recessive fashion, they studied boys and girls sepa-
rately using in-home evaluations of families with at least one child affected with
FXS and at least one unaffected (Hessl etal., 2001).
Several different environmental factors were associated with autistic behavior in
the presence of FXS diagnosis, but only in males. Two variables in the home environ-
ment were most associated with boys behavioral issues: the effectiveness of educa-
tional and therapeutic services provided to the child, and maternal psychological
problems. The effectiveness of educational and therapeutic services was measured
by the correlation between parents ranking of their childs cognitive and behavioral
skills before services were provided and 6 months after (The Special Curriculum
Opportunity Rating ScaleSCORS) (Hessl et al., 2001). Parental psychological
symptoms were measured by a 90-item self-report of psychological symptoms (The
Symptom Checklist-90 Revised) (Derogatis & Savitz, 1999). Conversely, levels of
the fragile X protein, FMRP, and IQ score were most associated with behavioral
problems in girls. This dichotomy likely reflects the variable level of FMRP expres-
sion in girls with FXS and the consistent absence or near-absence of FMPR expres-
sion in boys with FXS (Hessl et al., 2001). Therefore, these results suggest that
FMRP expression is a more robust risk factor than the environmental factors assessed
in this study, whose effects are only apparent when FMRP expression is largely elim-
inated. Importantly, this study could not assess whether environmental factors caused
behavioral problems or vice versa; although it is tempting to believe that better
behavioral services could ameliorate some ASD symptoms. Further, this study did
not quantify a true interaction between FMRP deficiencies and environmental fac-
tors, but instead showed an added effect of the environment on top of FXS diagnosis.
Future studies could broaden the assessment of environmental risk factors and spe-
cifically assess the interaction between robust genetic risk factors and common envi-
ronmental risk factors, but this would require a much larger sample size, which could
be a challenge in a relatively rare condition like FXS.
Ehninger etal. (2012) conducted a study that empirically quantified the impact
of interaction between gestational immune activation and Tsc2 haploinsufficiency
on social behavior in mice (Ehninger etal., 2012). Corresponding mutations in the
human lead to the rare syndrome of tuberous sclerosis complex, with a high co-
occurrence of ASD (4050%) (Bolton & Griffiths, 1997; Gillberg, Gillberg, &
Ahlsen, 1994; Hunt & Dennis, 1987; Smalley, Tanguay, Smith, & Gutierrez, 1992).
Additionally some, but not all, studies have suggested that prenatal infections
heighten the risk for ASD; although considerable uncertainty remains given vari-
ability in the infection type and time window for exposure during pregnancy
(Atladottir et al., 2010; Brown, 2012; Fatemi, Pearce, Brooks, & Sidwell, 2005;
Libbey, Sweeten, McMahon, & Fujinami, 2005; Patterson, 2011).
To test the impact of gene-environment interaction, Ehninger and colleagues
administered Poly(I:C), a synthetic double-stranded RNA that stimulates innate
immune response, to pregnant mice carrying litters of wildtype (Tsc2 +/+) and hap-
loinsufficient (Tsc2+/) embryos. Significantly fewer pups were born per litter in the
Tsc2+/+Poly(I:C) group than in the Tsc2+/+saline group or in the wildtype groups
(with or without Poly(I:C)). Among the surviving mice, the Tsc2+/+Poly(I:C) group
was the only group of mice that did not spend significantly more time exploring a
social stimulus (caged mouse) in contrast with an inanimate object. There were no
significant differences between the groups in the other two behavioral experiments.
There was also no significant interaction seen between Tsc2 haploinsufficiency and
advanced paternal age, another known ASD risk factor. These results suggest that the
interaction between the Tsc2 genotype and gestational immune activation was the
primary cause for abnormal behavior witnessed in the affected mice (Ehninger etal.,
2012). Unlike the study in humans with FXS, the inclusion of a wildtype mouse
control group here allows assessment of the environmental risk factor alone, thereby
allowing a true test of an interaction effect.
In the second part of this study, Ehninger etal. (2012) collected data on the birth-
dates of children with ASD only, TSC only, and both ASD and TSC.Children were
categorized into four groups based upon the stage of pregnancy their mothers were in
at the peak of flu season: first trimester, second trimester, third trimester, or not preg-
nant at that time. There were a significantly higher number of mothers in their third
trimester at the peak of flu season for the kids with both TSC and ASD.The numbers
were evenly distributed across the four groups for the kids with only ASD or TSC.This
analysis supports the plausibility of an interaction between Tsc2 haploinsufficiency
and gestational immune activation; although this indirect analysis did not examine
infection or immune activation and later ASD diagnosis (Ehninger etal., 2012).
I nfluence ofGenetics onOutcomes Related toStrong

Hessl etal. (2001) studied the influence of several environmental factors when a
very rare and potent genetic factor was present. The converse study could logically
be conducted to investigate genetic risk in the presence of rare but potent environ-
mental factors. To date, there have been no published studies that took this approach.
As noted above, there are very few environmental factors that have been as strongly
associated with ASD as genetic factors like FMR1. Factors with the most support
are extreme preterm birth with very low birth weight, valproic acid exposure, and
thalidomide exposure. Theoretically, researchers could conduct a study that searches
for genes that increase risk of ASD in the populations of children with these risk
factors. While the technology for genome-wide testing is available, finding an ade-
quate sample size for this kind of study could prove challenging.
onvergent Risk DuetoGenetic andEnvironmental Risk

C
Factors
A third way to measure gene-environment interaction in ASD is to target more com-

mon risk factors that may have a smaller effect size when considered alone. Without
identifying specific interactions, Tordjman etal. (2014) have highlighted epigenetic
effects as potentially underlying such gene-environment interactions. Epigenetics,
or above genetics, is broadly defined as alterations in gene expression and result-
ing phenotypes that do not derive from changes in the primary DNA sequence.
Recent findings suggest that the epigenome, represented by DNA methylation and
chromatin modification, can be heritable, but can also change over the lifetime. Just
as de novo mutations can arise in the primary DNA sequence, stochastic epigenetic
changes, or epimutations, can arise in the parental germ line or in the developing
fetus. Some epigenetic modifications reflect the local environment of a cell and are
necessary for differentiation of cells, including neurons and glia, but other modifica-
tions may reflect the broader environment of the organism. Epigenetic changes
likely explain some differences between monozygotic twins that share their com-
plete DNA sequence (Tordjman etal., 2014).
In 2014, Volk and colleagues studied the interaction between the c-Met receptor
tyrosine kinase gene (MET) and air pollution with respect to ASD.They had reported
previous evidence of increased ASD risk among children who were likely exposed
in utero to high levels of air pollution from local traffic. Past research also showed
an increased risk for ASD in children with the C allele of the MET gene promoter
variant rs1858830. Volk and colleagues conducted interviews of parents of 25year
old children with and without ASD (251 and 156 cases, respectively). Likely air
pollution exposure was determined by where the mother lived during her pregnancy.
Independently, there was no evidence to suggest increased risk of ASD for children
with the MET CC genotype as opposed to those with the CG or GG genotypes.
There were weak associations between ASD and exposure to the top quartile of
traffic-related air pollution, particulate matter less than 10 m in diameter, and
regional nitrogen dioxide exposure. When both factors were considered together,
there was evidence to suggest a higher risk of ASD for subjects with the MET
rs1858830 CC genotype and high likely in utero air pollutant exposure (OR=2.9,
P=0.09) and nitrogen dioxide (OR=3.6, P=0.03) (Volk etal., 2014).
Schmidt etal. (2011) investigated the possible interaction between prenatal vita-
mins and one-carbon metabolism gene variants. Previous studies have suggested that
valproic acid contributes to autism risk by disrupting folate metabolism, which is
critical to neurodevelopment (Adams etal., 2007; Ornoy, 2009). Folate is a B vitamin
involved in DNA methylation. Further, folate supplementation appears to be protec-
tive against ASD (Roth etal., 2011; Schmidt etal., 2012; Suren etal., 2013). Previous
studies have also suggested associations between autism and functional gene variants
within B vitamin-dependent folate, methionine, and transmethylation pathways.
These genes include 5,10 methylenetetrahydrofolate reductase (MTHFR), 5-methyl-
tetrahydrofolate-homocysteine methyltransferase (MTRR), t ranscobalamin II
(TCN2), and catechol-O-methyltransferase (COMT) (Adams et al., 2007; Boris,
Goldblatt, Galanko, & James, 2004; James etal., 2006; James etal., 2010). Schmidt
and colleagues conducted phone interviews of mothers of children affected with
ASD about their vitamin intake up to 3 months prior to conception and throughout
their pregnancy. The families were also genotyped for genes of interest within the
folate-dependent one-carbon metabolism: MTHFR 667C>T, COMT 472G>A, MTRR
A66G, BHMT G716A, FOLR2 rs2298444, CBS rs234715, and TCN2 C776G.Based
on these retrospective interviews, mothers of children with ASD were significantly
less likely to report having taken prenatal vitamins 3 months before conception and
during the first month of their pregnancy, with all of the usual caveats about possible
recall bias. From the aforementioned list of gene variants, maternal genotypes found
to have an interaction with not taking vitamins were MTHFR 677 TT (OR=4.5,
P=0.04), and CBS rs234715 GT+TT (OR=2.6, P=0.05). Children with COMT
472 AA also had a significantly greater risk of ASD if their mothers did not take
supplements (OR=7.2, P=0.05) (Schmidt etal., 2011).
The challenges for studies such as these include the nature and size of the sam-
ples, ascertainment bias, population stratification, and the vast array of environmen-
tal factors and genetic variants that could be analyzed. Appropriate statistical
correction is therefore very challenging. Genome-wide correction requires P<105,
at minimum, which would have to be further amplified by dozens or hundreds of
environmental factors. Replication in a separate sample is therefore the easiest way
to assess whether the observed interaction is due to chanceand very few of these
studies have been replicated.
he Future ofGene-Environment Interaction Research

T
inASD
Research on gene-environment interactions and their associations with ASD could

lead to better preventative measures or potential treatments. The complexities of
ASD make constructing such studies very difficult. One approach would be to use a
robust but rare risk factor on one side of the equation, acknowledging that each of
these factors comes with some challenges. Rare genetic risk factors, such as fragile
X syndrome or tuberous sclerosis complex, are clearly defined and have available
populations; however, the design of these experiments is likely to identify an addi-
tive effect of environmental factors on top of a robust genetic risk factor. Although
a few rare environmental risk factors are robustly associated with ASD, genome-
wide association studies (GWAS) that focus on these factors are challenging due to
the large sample size needed to achieve adequate power. Candidate gene studies
could be combined with a rare environmental risk factor, provided that the effect of
this factor is identified and understood on a well-defined network or signaling path-
way. Another approach would be to use RNA expression or gene methylation profil-
ing to identify genes or gene networks that differ by ASD diagnosis in people who
share a risk factor, like fragile X or extreme preterm birth.
Another approach is to use larger populations for which there are a lot of data on
a common, but less robust, risk factor. There is not yet enough knowledge in the
field to utilize aggregate genetic risk as a starting point, but that knowledge will
rapidly emerge once sample sizes increase in GWAS and sequencing studies. With
this information, such genetic risk profiles could be applied across a population with
well-characterized environmental risk factors. Another option would be to stratify a
large genome-wide association study by one of the few common environmental fac-
tors that may be available across the whole population, such as interpregnancy inter-
val or maternal diabetes. Unfortunately, few such risk factors are likely to be reliably
assessed in these large samples.
Turning the environmental approach around, it would also be possible to study
genetic moderators of response to treatments that target putative environmental risk
factors. Promising results from early interventions have been supported in several
recent studies in ASD. For example, Dawson et al. (2010) found significant evi-
dence that, when employed prior to 30months of age, the Early Start Denver Model
(ESDM) of behavioral intervention improves IQ, language, and adaptive behavior
outcomes in children with ASD (Dawson etal., 2010). Rare variant burden or poly-
genic risk score could potentially be used to evaluate whether response to ESDM
relates to genetic profile. Since social learning is a key component to ESDM, candi-
date genes could be studied within pathways that mediate social engagement;
although these would likely require a nested replication design to achieve adequate
power in the context of multiple comparisons.
A recent series of schizophrenia research studies may serve as an example of
how to approach gene-environment interactions in the treatment context. Like
ASD, schizophrenia is a multifaceted disorder in which there is no single cause.
Past studies have shown folate deficiency as a risk factor for more severe negative
symptoms in schizophrenia (Roffman etal., 2013). Roffman and colleagues found
that low serum folate levels contributed to negative symptoms specifically in indi-
viduals with the T/T genotype of MTHFR. This gene encodes for methylenetetra-
hydrofolate reductase, which is responsible for methylation of amino acids within
the vitamin B-dependent folate pathway (James et al., 1999; Roffman, Gollub,
etal., 2008; Roffman, Weiss, etal., 2008). Carrying this work forward into an ini-
tial open label treatment study, the same group found that subjects with the MTHFR
T/T and C/T genotypes were more likely to benefit from folate and vitamin B12
supplementation, whereas subjects with the C/C genotype did not show any
improvement (Hill etal., 2011). In a follow-up randomized control trial, they found
that supplementation with folate and vitamin B12 improved negative symptoms of
schizophrenia only in the presence of common functional genetic variants related
to folate absorption and metabolism. Subjects with the FOLH1 484T>C variant
showed the strongest results (Roffman et al., 2013). A similar approach could
potentially be used with ASD treatments related to putative environmental risk fac-
tors, possibly including folate, given evidence that prenatal folate supplementation
is protective.
Independent of the approach taken, Kim and Leventhal (2015) suggest that gene-
environment interaction studies should adhere to several criteria in order to ascertain
significant results. First, the sample should be epidemiologic and population-based to
encompass the entire ASD spectrum, and to avoid getting results solely applicable to
strict autism. Second, phenotypic characterization should be categorical as well as
dimensional to ensure accuracy and specificity. Third, environmental measurements,
which are often vague and difficult to quantify, need to be measured with accuracy,
validity, and, when possible, should include biomarkers. Fourth, these studies should
use statistical methods that deal with common challenges such as population stratifi-
cation, multiple comparisons, and interactions with rare genetic variants. Fifth, animal
models should be used to test hypotheses before studies include human subjects as
well as after. Six, new methods like GWAS and environment-wide association studies
should be employed. Finally, researchers should not simply consider these interac-
tions as causes of ASD, but rather as part of the course of the disease and as outcome
modifiers (Kim & Leventhal, 2015).
While gene by environment (GxE) interaction research in ASD has a multitude

of challenges, it may be a promising path to lead to prevention or intervention strate-
gies. It may also serve as a useful model for this sort of research in developmental
psychopathology more broadly. If a high-risk group can be identified, intervening
upon an environmental factor that multiplies risk associated with a particular geno-
type could have a profound effect. ASD is a complex disorder caused by a combina-
tion of risk factors that remain largely uncertain. The explosion of genetic findings
in recent years suggests that anchoring GxE studies with robust, though uncommon,
genetic risk factors may be the initial path forward. Eventually, pharmacogenetic
studies examining treatment response may offer an experimental approach to vali-
dating GxE findings that are supported by epidemiological studies.
References
Adams, M., Lucock, M., Stuart, J., Fardell, S., Baker, K., & Ng, X. (2007). Preliminary evidence
for involvement of the folate gene polymorphism 19 bp deletion-DHFR in occurrence of
autism. Neuroscience Letters, 422(1), 2429. doi:10.1016/j.neulet.2007.05.025.
Al-Farsi, Y. M., Waly, M. I., Deth, R. C., Al-Sharbati, M. M., Al-Shafaee, M., Al-Farsi, O.,
Ouhtit, A. (2013). Low folate and vitamin B12 nourishment is common in Omani children with
newly diagnosed autism. Nutrition, 29(3), 537541. doi:10.1016/j.nut.2012.09.014.
American Psychiatric Association (2013). Diagnostic and statistical manual of mental disorders:
Dsm-5. Washington, DC: American Psychiatric Publishing Incorporated.
Anney, R., Klei, L., Pinto, D., Almeida, J., Bacchelli, E., Baird, G., Devlin, B. (2012). Individual
common variants exert weak effects on the risk for autism spectrum disorderspi. Human
Molecular Genetics, 21(21), 47814792. doi:10.1093/hmg/dds301.
Asbury, K., Almeida, D., Hibel, J., Harlaar, N., & Plomin, R. (2008). Clones in the classroom: A
daily diary study of the nonshared environmental relationship between monozygotic twin dif-
ferences in school experience and achievement. Twin Research and Human Genetics, 11(6),
586595. doi:10.1375/twin.11.6.586.
Ashley, C.T., Sutcliffe, J.S., Kunst, C.B., Leiner, H.A., Eichler, E.E., Nelson, D.L., & Warren,
S.T. (1993). Human and murine FMR-1: Alternative splicing and translational initiation down-
stream of the CGG-repeat. Nature Genetics, 4(3), 244251. doi:10.1038/ng0793-244.
Atladottir, H.O., Thorsen, P., Ostergaard, L., Schendel, D.E., Lemcke, S., Abdallah, M., & Parner,
E.T. (2010). Maternal infection requiring hospitalization during pregnancy and autism spec-
trum disorders. Journal of Autism and Developmental Disorders, 40(12), 14231430.
doi:10.1007/s10803-010-1006-y.
Bailey, A., Le Couteur, A., Gottesman, I., Bolton, P., Simonoff, E., Yuzda, E., & Rutter, M. (1995).
Autism as a strongly genetic disorder: Evidence from a British twin study. Psychological
Medicine, 25(1), 6377.
Bear, M.F., Huber, K.M., & Warren, S.T. (2004). The mGluR theory of fragile X mental retarda-
tion. Trends in Neurosciences, 27(7), 370377. doi:10.1016/j.tins.2004.04.009.
Berry-Kravis, E. (2014). Mechanism-based treatments in neurodevelopmental disorders: fragile X
syndrome. Pediatric Neurology, 50(4), 297302. doi:10.1016/j.pediatrneurol.2013.12.001.
Bolton, P. F., & Griffiths, P. D. (1997). Association of tuberous sclerosis of temporal lobes with
autism and atypical autism. Lancet, 349(9049), 392395. doi:10.1016/S0140-6736(97)80012-8.
Boris, M., Goldblatt, A., Galanko, J., & James, S.J. (2004). Association of MTHFR gene variants
with autism. Journal of the American Physicians and Surgeons, 9(4), 106108.
Brown, A.S. (2012). Epidemiologic studies of exposure to prenatal infection and risk of schizophre-
nia and autism. Developmental Neurobiology, 72(10), 12721276. doi:10.1002/dneu.22024.
Chaste, P., & Leboyer, M. (2012). Autism risk factors: Genes, environment, and gene-environment
interactions. Dialogues in Clinical Neuroscience, 14(3), 281292.
Cheslack-Postava, K., Suominen, A., Jokiranta, E., Lehti, V., McKeague, I.W., Sourander, A., &
Brown, A.S. (2014). Increased risk of autism spectrum disorders at short and long interpreg-
nancy intervals in Finland. Journal of the American Academy of Child and Adolescent
Psychiatry, 53(10), 10741081 e1074. doi:10.1016/j.jaac.2014.06.009.
Chess, S. (1971). Autism in children with congenital rubella. Journal of Autism and Childhood
Schizophrenia, 1(1), 3347.
Chess, S. (1977). Follow-up report on autism in congenital rubella. Journal of Autism and
Childhood Schizophrenia, 7(1), 6981.
Christensen, J., Gronborg, T.K., Sorensen, M.J., Schendel, D., Parner, E.T., Pedersen, L.H., &
Vestergaard, M. (2013). Prenatal valproate exposure and risk of autism spectrum disorders and
childhood autism. JAMA, 309(16), 16961703. doi:10.1001/jama.2013.2270.
Czyz, W., Morahan, J.M., Ebers, G.C., & Ramagopalan, S.V. (2012). Genetic, environmental and
stochastic factors in monozygotic twin discordance with a focus on epigenetic differences.
BMC Medicine, 10, 93. doi:10.1186/1741-7015-10-93.
Dawson, G., Rogers, S., Munson, J., Smith, M., Winter, J., Greenson, J., Varley, J. (2010).
Randomized, controlled trial of an intervention for toddlers with autism: The Early Start
Denver Model. Pediatrics, 125(1), e17e23. doi:10.1542/peds.2009-0958.
Derogatis, L.R., & Savitz, K.L. (1999). The SCL-90-R, Brief Symptom Inventory, and Matching
Clinical Rating Scales. In M.E. Maruish (Ed.), The use of psychological testing for treatment
planning and outcomes assessment (2nd ed., pp. 679724). Mahwah, NJ, USA: Lawrence
Erlbaum Associates Publishers.
Devlin, B., & Scherer, S.W. (2012). Genetic architecture in autism spectrum disorder. Current
Opinion in Genetics & Development, 22(3), 229237. doi:10.1016/j.gde.2012.03.002.
Dolen, G., & Bear, M. F. (2008). Role for metabotropic glutamate receptor 5 (mGluR5) in the
pathogenesis of fragile X syndrome. The Journal of Physiology, 586(6), 15031508.
doi:10.1113/jphysiol.2008.150722.
Dong, S., Walker, M.F., Carriero, N.J., DiCola, M., Willsey, A.J., Ye, A.Y., Sanders, S.J.
(2014). De novo insertions and deletions of predominantly paternal origin are associated with
autism spectrum disorder. Cell Reports, 9(1), 1623. doi:10.1016/j.celrep.2014.08.068.
Ehninger, D. (2013). From genes to cognition in tuberous sclerosis: Implications for mTOR
inhibitor-based treatment approaches. Neuropharmacology, 68, 97105. doi:10.1016/j.
neuropharm.2012.05.015.
Ehninger, D., Sano, Y., de Vries, P.J., Dies, K., Franz, D., Geschwind, D.H., Silva, A.J. (2012).
Gestational immune activation and Tsc2 haploinsufficiency cooperate to disrupt fetal survival
and may perturb social behavior in adult mice. Molecular Psychiatry, 17(1), 6270. doi:10.1038/
mp.2010.115.
Fatemi, S.H., Pearce, D.A., Brooks, A.I., & Sidwell, R.W. (2005). Prenatal viral infection in
mouse causes differential expression of genes in brains of mouse progeny: A potential animal
model for schizophrenia and autism. Synapse, 57(2), 9199. doi:10.1002/syn.20162.
Folstein, S., & Rutter, M. (1977). Infantile autism: A genetic study of 21 twin pairs. Journal of
Child Psychology and Psychiatry, 18(4), 297321.
Gardener, H., Spiegelman, D., & Buka, S.L. (2011). Perinatal and neonatal risk factors for autism:
a comprehensive meta-analysis. Pediatrics, 128(2), 344355. doi:10.1542/peds.2010-1036.
Gaugler, T., Klei, L., Sanders, S.J., Bodea, C.A., Goldberg, A.P., Lee, A.B., Buxbaum, J.D.
(2014). Most genetic risk for autism resides with common variation. Nature Genetics, 46(8),
881885. doi:10.1038/ng.3039.
Gillberg, I.C., Gillberg, C., & Ahlsen, G. (1994). Autistic behaviour and attention deficits in tuber-
ous sclerosis: A population-based study. Developmental Medicine and Child Neurology, 36(1),
5056.
Hall, S.S., Lightbody, A.A., Hirt, M., Rezvani, A., & Reiss, A.L. (2010). Autism in fragile X
syndrome: A category mistake? Journal of the American Academy of Child and Adolescent
Psychiatry, 49(9), 921933. doi:10.1016/j.jaac.2010.07.001.
Hallmayer, J., Cleveland, S., Torres, A., Phillips, J., Cohen, B., Torigoe, T., Risch, N. (2011).
Genetic heritability and shared environmental factors among twin pairs with autism. Archives
of General Psychiatry, 68(11), 10951102. doi:10.1001/archgenpsychiatry.2011.76.
Hessl, D., Dyer-Friedman, J., Glaser, B., Wisbeck, J., Barajas, R.G., Taylor, A., & Reiss, A.L.
(2001). The influence of environmental and genetic factors on behavior problems and autistic
symptoms in boys and girls with fragile X syndrome. Pediatrics, 108(5), E88.
Hill, M., Shannahan, K., Jasinski, S., Macklin, E.A., Raeke, L., Roffman, J.L., & Goff, D.C.
(2011). Folate supplementation in schizophrenia: a possible role for MTHFR genotype.
Schizophrenia Research, 127(13), 4145. doi:10.1016/j.schres.2010.12.006.
Howard, J.S., Sparkman, C.R., Cohen, H.G., Green, G., & Stanislaw, H. (2005). A comparison
of intensive behavior analytic and eclectic treatments for young children with autism. Research
in Developmental Disabilities, 26(4), 359383. doi:10.1016/j.ridd.2004.09.005.
Huguet, G., Ey, E., & Bourgeron, T. (2013). The genetic landscapes of autism spectrum disor-
ders. Annual Review of Genomics and Human Genetics, 14, 191213. doi:10.1146/
annurev-genom-091212-153431.
Hunt, A., & Dennis, J. (1987). Psychiatric disorder among children with tuberous sclerosis.
Developmental Medicine and Child Neurology, 29(2), 190198.
Iossifov, I., ORoak, B. J., Sanders, S. J., Ronemus, M., Krumm, N., Levy, D., Wigler, M.
(2014). The contribution of de novo coding mutations to autism spectrum disorder. Nature,
515(7526), 216221. doi:10.1038/nature13908.
James, S.J., Melnyk, S., Jernigan, S., Cleves, M.A., Halsted, C.H., Wong, D.H., Bradstreet,
J. J. (2006). Metabolic endophenotype and related genotypes are associated with oxidative
stress in children with autism. American Journal of Medical Genetics Part B: Neuropsychiatric
Genetics, 141(8), 947956.
James, S.J., Melnyk, S., Jernigan, S., Pavliv, O., Trusty, T., Lehman, S., Cleves, M.A. (2010).
A functional polymorphism in the reduced folate carrier gene and DNA hypomethylation in
mothers of children with autism. American Journal of Medical Genetics. Part B,
Neuropsychiatric Genetics, 153B(6), 12091220. doi:10.1002/ajmg.b.31094.
James, S.J., Pogribna, M., Pogribny, I.P., Melnyk, S., Hine, R.J., Gibson, J.B., Gaylor, D.W.
(1999). Abnormal folate metabolism and mutation in the methylenetetrahydrofolate reductase
gene may be maternal risk factors for Down syndrome. The American Journal of Clinical
Nutrition, 70(4), 495501.
Jeste, S.S., & Geschwind, D.H. (2014). Disentangling the heterogeneity of autism spectrum dis-
order through genetic findings. Nature Reviews. Neurology, 10(2), 7481. doi:10.1038/
nrneurol.2013.278.
Kim, Y.S., & Leventhal, B.L. (2015). Genetic epidemiology and insights into interactive genetic
and environmental effects in autism spectrum disorders. Biological Psychiatry, 77(1), 6674.
doi:10.1016/j.biopsych.2014.11.001.
Klei, L., Sanders, S.J., Murtha, M.T., Hus, V., Lowe, J.K., Willsey, A.J., Devlin, B. (2012).
Common genetic variants, acting additively, are a major source of risk for autism. Molecular
Autism, 3(1), 9. doi:10.1186/2040-2392-3-9.
Kolevzon, A., Gross, R., & Reichenberg, A. (2007). Prenatal and perinatal risk factors for autism:
A review and integration of findings. Archives of Pediatrics & Adolescent Medicine, 161(4),
326333. doi:10.1001/archpedi.161.4.326.
Kremer, E.J., Pritchard, M., Lynch, M., Yu, S., Holman, K., Baker, E., Richards, R.I. (1991).
Mapping of DNA instability at the fragile X to a trinucleotide repeat sequence p(CCG)n.
Science, 252(5013), 17111714.
Lampi, K. M., Hinkka-Yli-Salomaki, S., Lehti, V., Helenius, H., Gissler, M., Brown, A. S., &
Sourander, A. (2013). Parental age and risk of autism spectrum disorders in a Finnish national
birth cohort. Journal of Autism and Developmental Disorders, 43(11), 25262535. doi:10.1007/
s10803-013-1801-3.
Landrigan, P.J. (2010). What causes autism? Exploring the environmental contribution. Current
Opinion in Pediatrics, 22(2), 219225. doi:10.1097/MOP.0b013e328336eb9a.
Leekam, S.R., Libby, S.J., Wing, L., Gould, J., & Taylor, C. (2002). The Diagnostic Interview for
Social and Communication Disorders: Algorithms for ICD-10 childhood autism and Wing and
Gould autistic spectrum disorder. Journal of Child Psychology and Psychiatry, 43(3),
327342.
Levy, D., Ronemus, M., Yamrom, B., Lee, Y.H., Leotta, A., Kendall, J., Wigler, M. (2011). Rare
de novo and transmitted copy-number variation in autistic spectrum disorders. Neuron, 70(5),
886897. doi:10.1016/j.neuron.2011.05.015.
Libbey, J.E., Sweeten, T.L., McMahon, W.M., & Fujinami, R.S. (2005). Autistic disorder and
viral infections. Journal of Neurovirology, 11(1), 110. doi:10.1080/13550280590900553.
Lipton, J.O., & Sahin, M. (2014). The neurology of mTOR. Neuron, 84(2), 275291. doi:10.1016/j.
neuron.2014.09.034.
London, E. A. (2000). The environment as an etiologic factor in autism: A new direction for
research. Environmental Health Perspectives, 108(Suppl 3), 401404.
Mamidala, M.P., Polinedi, A., P T V, P.K., Rajesh, N., Vallamkonda, O.R., Udani, V., Rajesh,
V. (2013). Prenatal, perinatal and neonatal risk factors of Autism Spectrum Disorder: A com-
prehensive epidemiological assessment from India. Research in Developmental Disabilities,
34(9), 30043013. doi:10.1016/j.ridd.2013.06.019.
Mann, J.R., McDermott, S., Bao, H., Hardin, J., & Gregg, A. (2010). Pre-eclampsia, birth weight,
and autism spectrum disorders. Journal of Autism and Developmental Disorders, 40(5), 548
554. doi:10.1007/s10803-009-0903-4.
McPheeters, M. L., Warren, Z., Sathe, N., Bruzek, J. L., Krishnaswami, S., Jerome, R. N., &
Veenstra-Vanderweele, J.(2011). A systematic review of medical treatments for children with
autism spectrum disorders. Pediatrics, 127(5), e1312e1321. doi:10.1542/peds.2011-0427.
Nosarti, C., Reichenberg, A., Murray, R.M., Cnattingius, S., Lambe, M.P., Yin, L., Hultman,
C.M. (2012). Preterm birth and psychiatric disorders in young adult life. Archives of General
Psychiatry, 69(6), E1E8. doi:10.1001/archgenpsychiatry.2011.1374.
ORoak, B. J., Vives, L., Girirajan, S., Karakoc, E., Krumm, N., Coe, B. P., Eichler, E. E.
(2012). Sporadic autism exomes reveal a highly interconnected protein network of de novo
mutations. Nature, 485(7397), 246250. doi:10.1038/nature10989.
ODonnell, W.T., & Warren, S.T. (2002). A decade of molecular studies of fragile X syndrome.
Annual Review of Neuroscience, 25, 315338. doi:10.1146/annurev.neuro.25.112701.142909.
Ornoy, A. (2009). Valproic acid in pregnancy: How much are we endangering the embryo and
fetus? Reproductive Toxicology, 28(1), 110. doi:10.1016/j.reprotox.2009.02.014.
ORoak, B.J., Stessman, H.A., Boyle, E.A., Witherspoon, K.T., Martin, B., Lee, C., Eichler,
E.E. (2014). Recurrent de novo mutations implicate novel genes underlying simplex autism
risk. Nature Communications, 5, 5595. doi:10.1038/ncomms6595.
Patterson, P.H. (2011). Maternal infection and immune involvement in autism. Trends in Molecular
Medicine, 17(7), 389394. doi:10.1016/j.molmed.2011.03.001.
Plomin, R. (2011). Commentary: Why are children in the same family so different? Non-shared
environment three decades later. International Journal of Epidemiology, 40(3), 582592.
doi:10.1093/ije/dyq144.
Poultney, C. S., Samocha, K., Kou, Y., Liu, L., Walker, S., Singh, T., Roeder, K. (2014).
Synaptic, transcriptional and chromatin genes disrupted in autism. Nature, 515(7526), 209
215. doi:10.1038/nature13772.
Richler, J., Bishop, S.L., Kleinke, J.R., & Lord, C. (2007). Restricted and repetitive behaviors in
young children with autism spectrum disorders. Journal of Autism and Developmental
Disorders, 37(1), 7385. doi:10.1007/s10803-006-0332-6.
Richler, J., Huerta, M., Bishop, S.L., & Lord, C. (2010). Developmental trajectories of restricted
and repetitive behaviors and interests in children with autism spectrum disorders. Development
and Psychopathology, 22(1), 5569. doi:10.1017/S0954579409990265.
Roffman, J.L., Gollub, R.L., Calhoun, V.D., Wassink, T.H., Weiss, A.P., Ho, B.C., Manoach,
D. S. (2008). MTHFR 677C --> T genotype disrupts prefrontal function in schizophrenia
through an interaction with COMT 158Val --> Met. Proceedings of the National Academy of
Sciences of the United States of America, 105(45), 1757317578. doi:10.1073/pnas.0803727105.
Roffman, J.L., Lamberti, J.S., Achtyes, E., Macklin, E.A., Galendez, G.C., Raeke, L.H., Goff,
D.C. (2013). Randomized multicenter investigation of folate plus vitamin B12 supplementation
in schizophrenia. JAMA Psychiatry, 70(5), 481489. doi:10.1001/jamapsychiatry.2013.900.
Roffman, J.L., Weiss, A.P., Purcell, S., Caffalette, C.A., Freudenreich, O., Henderson, D.C.,
Goff, D.C. (2008). Contribution of methylenetetrahydrofolate reductase (MTHFR) polymor-
phisms to negative symptoms in schizophrenia. Biological Psychiatry, 63(1), 4248.
doi:10.1016/j.biopsych.2006.12.017.
Ronemus, M., Iossifov, I., Levy, D., & Wigler, M. (2014). The role of de novo mutations in the
genetics of autism spectrum disorders. Nature Reviews. Genetics, 15(2), 133141. doi:10.1038/
nrg3585.
Roth, C., Magnus, P., Schjolberg, S., Stoltenberg, C., Suren, P., McKeague, I.W., Susser, E.
(2011). Folic acid supplements in pregnancy and severe language delay in children. JAMA,
306(14), 15661573. doi:10.1001/jama.2011.1433.
Sahin, M. (2012). Targeted treatment trials for tuberous sclerosis and autism: No longer a dream.
Current Opinion in Neurobiology, 22(5), 895901. doi:10.1016/j.conb.2012.04.008.
Sallows, G.O., & Graupner, T.D. (2005). Intensive behavioral treatment for children with autism:
Four-year outcome and predictors. American Journal of Mental Retardation, 110(6), 417438.
doi:10.1352/0895-8017(2005)110[417:IBTFCW]2.0.CO;2.
Sanders, S.J., Ercan-Sencicek, A.G., Hus, V., Luo, R., Murtha, M.T., Moreno-De-Luca, D.,
State, M.W. (2011). Multiple recurrent de novo CNVs, including duplications of the 7q11.23
Williams syndrome region, are strongly associated with autism. Neuron, 70(5), 863885.
doi:10.1016/j.neuron.2011.05.002.
Sanders, S.J., Murtha, M.T., Gupta, A.R., Murdoch, J.D., Raubeson, M.J., Willsey, A.J.,
State, M. W. (2012). De novo mutations revealed by whole-exome sequencing are strongly
associated with autism. Nature, 485(7397), 237241. doi:10.1038/nature10945.
Sandin, S., Lichtenstein, P., Kuja-Halkola, R., Larsson, H., Hultman, C.M., & Reichenberg, A.
(2014). The familial risk of autism. JAMA, 311(17), 17701777. doi:10.1001/jama.2014.4144.
Satterthwaite, T.D., & Baker, J.T. (2014). How can studies of resting-state functional connectivity
help us understand psychosis as a disorder of brain development? Current Opinion in
Neurobiology, 30C, 8591. doi:10.1016/j.conb.2014.10.005.
Schizophrenia Working Group of the Psychiatric Genomics, C (2014). Biological insights from 108
schizophrenia-associated genetic loci. Nature, 511(7510), 421427. doi:10.1038/nature13595.
Schmidt, R.J., Hansen, R.L., Hartiala, J., Allayee, H., Schmidt, L.C., Tancredi, D.J., Hertz-
Picciotto, I. (2011). Prenatal vitamins, one-carbon metabolism gene variants, and risk for
autism. Epidemiology, 22(4), 476485. doi:10.1097/EDE.0b013e31821d0e30.
Schmidt, R. J., Tancredi, D. J., Ozonoff, S., Hansen, R. L., Hartiala, J., Allayee, H., Hertz-
Picciotto, I. (2012). Maternal periconceptional folic acid intake and risk of autism spectrum
disorders and developmental delay in the CHARGE (CHildhood Autism Risks from Genetics
and Environment) case-control study. The American Journal of Clinical Nutrition, 96(1),
8089. doi:10.3945/ajcn.110.004416.
Smalley, S.L., Tanguay, P.E., Smith, M., & Gutierrez, G. (1992). Autism and tuberous sclerosis.
Journal of Autism and Developmental Disorders, 22(3), 339355.
Stromland, K., Nordin, V., Miller, M., Akerstrom, B., & Gillberg, C. (1994). Autism in thalidomide
embryopathy: A population study. Developmental Medicine and Child Neurology, 36(4),
351356.
Stromswold, K. (2006). Why arent identical twins linguistically identical? Genetic, prenatal and
postnatal factors. Cognition, 101(2), 333384. doi:10.1016/j.cognition.2006.04.007.
Suren, P., Roth, C., Bresnahan, M., Haugen, M., Hornig, M., Hirtz, D., Stoltenberg, C. (2013).
Association between maternal use of folic acid supplements and risk of autism spectrum disor-
ders in children. JAMA, 309(6), 570577. doi:10.1001/jama.2012.155925.
Tordjman, S., Somogyi, E., Coulon, N., Kermarrec, S., Cohen, D., Bronsard, G., Xavier,
J.(2014). Gene x Environment interactions in autism spectrum disorders: Role of epigenetic
mechanisms. Frontiers in Psychiatry, 5, 53. doi:10.3389/fpsyt.2014.00053.
Veenstra-VanderWeele, J., & Blakely, R.D. (2012). Networking in autism: Leveraging genetic,
biomarker and model system findings in the search for new treatments. Neuropsychophar
macology, 37(1), 196212. doi:10.1038/npp.2011.185.
Verkerk, A.J., Pieretti, M., Sutcliffe, J.S., Fu, Y.H., Kuhl, D.P., Pizzuti, A., Caskey, T. (1991).
Identification of a gene (FMR-1) containing a CGG repeat coincident with a breakpoint cluster
region exhibiting length variation in fragile X syndrome. Cell, 65(5), 905914.
Volk, H.E., Kerin, T., Lurmann, F., Hertz-Picciotto, I., McConnell, R., & Campbell, D.B. (2014).
Autism spectrum disorder: Interaction of air pollution with the MET receptor tyrosine kinase
gene. Epidemiology, 25(1), 4447. doi:10.1097/EDE.0000000000000030.
Warren, Z., McPheeters, M.L., Sathe, N., Foss-Feig, J.H., Glasser, A., & Veenstra-Vanderweele,
J.(2011). A systematic review of early intensive intervention for autism spectrum disorders.
Pediatrics, 127(5), e1303e1311. doi:10.1542/peds.2011-0426.
Windham, G.C., Anderson, M.C., Croen, L.A., Smith, K.S., Collins, J., & Grether, J.K. (2011).
Birth prevalence of autism spectrum disorders in the San Francisco Bay area by demographic
and ascertainment source characteristics. Journal of Autism and Developmental Disorders,
41(10), 13621372. doi:10.1007/s10803-010-1160-2.
Xu, G., Jing, J., Bowers, K., Liu, B., & Bao, W. (2014). Maternal diabetes and the risk of autism
spectrum disorders in the offspring: A systematic review and meta-analysis. Journal of Autism
and Developmental Disorders, 44(4), 766775. doi:10.1007/s10803-013-1928-2.
Chapter 10
How Can GxE Research Help Prevent
theDevelopment ofChronic Physical
Aggression?
RichardE.Tremblay, LindaBooij, NadineProvenal, andMosheSzyf
Physical aggressions among humans are daily front page headlines. All humans fear
physical aggressions for a very good reason; they can make the difference between
life and death. The physical health professions focus on physical aggressions
because they are the cause of numerous physical health problems for the aggressed
as well as the aggressors. The mental health professions focus on emotional injuries
caused by physical aggressions, on the treatment of the aggressors, as well as the
aggressed and, occasionally, on the prevention of physical aggressions.
The American Psychiatric Association Diagnostic Statistical Manuel published
35 years ago (DSM-III, 1980) placed aggression at the center of its Conduct
Disorder (CD) diagnostic types: under-socialized aggressive, under-socialized non-
aggressive, socialized aggressive, socialized nonaggressive, atypical. Fourteen
years later, the DSM CD classification took a developmental turn. There were now
only three diagnostic types: childhood-onset, adolescent-onset, and age at onset
unknown (DSM-IV, 1994; DSM-5, 2013). However, physical aggressions remain at
the core of the diagnostic criteria and physically aggressive children are the most
feared and the most at risk of violent and nonviolent offending during adolescence
R.E. Tremblay (*)

University College Dublin, Dublin, Ireland
GRIP, Universit de Montral,
3050, douard-Montpetit Blvd., Montreal, QC, Canada, H3T 1J7
e-mail: richard.ernest.tremblay@umontreal.ca
L. Booij
Concordia University, Montreal, QC, Canada
N. Provenal
Max Planck Institute of Psychiatry, Munich, Germany
M. Szyf
McGill University, Montreal, QC, Canada

DOI10.1007/978-3-319-49227-8_10
178 R.E. Tremblay et al.
and early adulthood (Broidy etal., 2003; Fontaine, Lacourse, Vitaro, & Tremblay,
2014; Nagin & Tremblay, 1999; Pingault etal., 2013).
There is a vast body of research on the development of animal and human physi-
cal aggression. However, although the DSM took a developmental perspective
20years ago, research on the development of physical aggression using the DSM
classification is rare. Not surprisingly, there is even less research on the prevention
of physical aggression from the perspective of the DSM classification (Tremblay,
2010; Wilson & Lipsey, 2007).
This chapter first summarizes the available knowledge on the development of
human chronic physical aggression (CPA) from early childhood onwards. We then
summarize the state of knowledge on the genetic and environmental mechanisms
that may be involved in the development of CPA from conception onwards, and we
finally suggest research on the prevention of CPA that could advance both knowl-
edge of the developmental GxE mechanisms and knowledge on the very early pre-
vention of CPA.
evelopment ofChronic Physical Aggression fromEarly

D
Childhood toAdolescence
Following a survey of the international scientific literature on violence, the World

Health Organisation (WHO) concluded in 2002: The majority of young people who
become violent are adolescent-limited offenders who, in fact, show little or no evi-
dence of high levels of aggression or other problem behaviors during their childhood
(Krug, Dahlberg, Mercy, Zwi, & Lozano, 2002). This conclusion begs the question
why would humans start to use high levels of physical aggression and other prob-
lem behaviors during adolescence? Ten years before the WHO report, the US Academy
of Science Panel on Understanding Violent Behavior had given a summary of the
mechanisms which could explain the WHO conclusion: Modern psychological per-
spectives emphasize that aggressive and violent behaviors are learned responses to
frustration, that they can also be learned as instruments for achieving goals, and that
the learning occurs by observing models of such behavior. Such models may be
observed in the family, among peers, elsewhere in the neighborhood, through the
mass media (Reiss & Roth, 1993). Thus, to follow the logic of the WHO report,
humans apparently need approximately 1213years of exposure to physical aggres-
sion in their environment before they start using high levels of aggression.
This social learning of aggression hypothesis began to be explicitly formulated
in the early 1960s (Bandura, Ross, & Ross, 1961) and became the dominant expla-
nation for most forms of antisocial behavior. It also became the justification for
focusing resources on adolescents to prevent delinquent behavior in general and
physical violence in particular (Tremblay, 2006).
Interestingly, this perspective on the development of physical aggression corre-
sponds to an old scientific observation and to pervasive public opinion. From a sci-
entific perspective the age-crime curve was first described in 1831 by Adolphe
10 How Can GxE Research Help Prevent the Development of Chronic Physical 179
Rate per 100,000 population

5,000
4,500 Violent Offences
4,000 Property Offences
3,500
3,000
2,500
2,000
1,500
1,000
500
0
3 6 9 12 15 18 21 24 27 30 33 36 39 42 45 48 51 54
Fig. 10.1 Age-crime curve: Individuals accused of violent and property crimes in Canada 1999
Quetelet (1833) and has since been observed in all societies. For example, Fig. 10.1
describes the frequency of arrests for aggressions against persons and property
offenses with reference to age in peaceful Canada. We see that there is a substantial
increase with age from childhood throughout adolescence and then a substantial
decrease during adulthood.
From a public opinion perspective, in 2003 we did a survey of Canadians opin-
ion on age at which Canadians most frequently used physical aggression (Tremblay,
2003). We found (see Fig. 10.2) that Canadians perception of the age at which
physical violence was most frequent matched pretty well the age-crime curve. A
majority of Canadians also tended to believe that preventive interventions should
target adolescents and preadolescents.
Recent longitudinal studies on the development of physical aggression initiated
during early childhood in a number of industrialized countries have now seriously
questioned the idea that the frequency of physical aggressions increases with age
from childhood to late adolescence. The first systematic studies showing that the
onset of physical aggression was in early childhood were done in Canada and the
United States in the early twentieth century (Bridges, 1931, 1933; Goodenough,
1931). More recently, longitudinal studies in Canada and the United States also
showed that the mean frequency of physical aggression declined during the elemen-
tary school years and adolescence (Cairns & Cairns, 1994; Loeber & Stouthamer-
Loeber, 1998; Nagin & Tremblay, 1999). However, it was assumed that, although
most children decrease their level of physical aggression during the elementary
school years, the chronically aggressive individuals were initiating their frequent
physical aggression during adolescence. Data on the development of physical
aggression from a longitudinal study of elementary school children initiated in the
early 1980s helped question this idea when Nagin and Tremblay (1999) traced the
developmental trajectories of physical aggression from 6 to 15years of age with the
Montreal Longitudinal-Experimental Study. These developmental trajectories con-
firmed that the frequency of physical aggression was generally decreasing from
At what age boys resort most frequently to

physical aggression?*
40 37
30 24
18
% 20
9
10 5 5
2
0
0 to 4 5 to 7 8 to 11 12 to 14 15 to 17 18 + others
Age
At what age should we invest most to prevent

physical aggression?*
50
41
40 33
30
%
20
10 9
10 3 4
0
0 to 4 5 to 11 12 to 17 18 to 29 30 + others
Age
* Poll conducted by Lger-Marketing
Fig. 10.2 Perceptions of a representative sample of the Canadian population
school entry to mid-adolescence, but also that the most physically aggressive ado-
lescents were the most physically aggressive during their kindergarten year (see
Fig.1 from Nagin & Tremblay, 1999). As can be seen from Fig. 10.3, the data analy-
sis with a semi-parametric developmental trajectory approach did not identify a
significant group of children that increased their frequency of physical aggressions
from kindergarten to adolescence, although one would have expected to identify
such a trajectory with a longitudinal study based on a large sample (N=1037) of
males from 53 schools in low socioeconomic areas of a large North American city.
In order to replicate these findings from a relatively homogenous gene pool in a
relatively small culture within North America, the same analyses were done with six
large longitudinal studies from Canada, New Zealand, and the United States. The
study thus compared six samples of males and five samples of females in three differ-
ent countries. Results (Broidy etal., 2003) replicated the main finding of the Montreal
study of males from low socioeconomic environments: there was no evidence of a
significant group of children that showed low levels of physical aggression at school
entry and significantly increased their level of physical aggression during the elemen-
tary school years (see also results from a study in Italy Di Giunta etal., 2010).
Fig. 10.3 Developmental trajectories of physical aggression from 6 to 15years of age (Nagin &
Tremblay, 1999)
The following developmental question was obviously when do children learn to

aggress if they are already at their peak in frequency during kindergarten? Fortunately,
the study of social behavior development, including physical aggression, was included
in large birth cohorts during the 1990s in Canada, the United Kingdom, Norway, and
the United States. Results clearly showed that humans start to use physical aggression
towards the end of the first year after birth when they have acquired the motor coordi-
nation to push, pull, hit, kick, etc. (Alink etal., 2006; Ct, Vaillancourt, LeBlanc,
Nagin, & Tremblay, 2006; Hay etal., 2011; Naerde, Ogden, Janson, & Zachrisson,
2014; Early Child Care Research Network, 2004; Tremblay etal., 1999, 2004). Figure
10.4 illustrates results of the physical aggression developmental trajectory analyses
from 17 to 60months based on a representative sample of children from a Canadian
Province (Ct etal., 2007). We can see that the development trajectories of physical
aggression are increasing from 17 to 42months and then decreasing; hence, the peak
in frequency of physical aggressions was around 3 and a half years of age. Similar
developmental trends were observed in other large-scale longitudinal studies in
Canada and the United States (Campbell, Spieker, Burchinal, Poe, & NICHD Early
Child Care Research Network, 2006; Ct etal., 2009, 2006) and in an intensive US
observational study of childrens response in a laboratory delay of gratification situa-
tion from 18 to 48months (Cole etal., 2011).
The analyses described in Fig. 10.4 are based on prospective repeated assessments
of physical aggressions reported by mothers over 4years. From this perspective, a
developmental trajectory should be a better estimate of a chronic behavior problem
Fig. 10.4. Developmental trajectories of physical aggression from 17 to 60months (Ct etal., 2007)
than an assessment at any given point in time, even if that assessment attempts to
reconstruct past behavior. Longitudinal data has shown that, within a year, mothers do
not recall the age of onset of their childrens physical aggressions (Tremblay, 2000).
In a US clinical study of boys between 7 and 12years of age, the mean age of physical
aggression onset retrospectively reported by parents was 6.75 years (Frick et al.,
1993). Retrospective information collected in the Pittsburgh Youth Study (Loeber &
Hay, 1997; Loeber & Stouthamer-Loeber, 1998) was compared to prospective data
and highlighted the problem with retrospective assessments of the onset of physical
aggression. The participants (N = 503) represented the Pittsburgh public schools
male eighth graders and were close to 14years old (mean age=13.8; SD=0.80) at
the first data collection. The cumulative age of onset of physical aggression reported
retrospectively by the mothers and the boys at the first data collection indicated that
by age 5, less than 5% of the boys had initiated use of physical aggression and almost
none of the boys had initiated fighting. In sharp contrast, the prospective data repre-
sented in Fig. 10.4 on physical aggression from 17months after birth indicate that
children who did not initiate physical aggression before 3years of age were extremely
uncommon. These prospective studies suggest that the peak in the frequency of physi-
cal aggression for most humans is somewhere between 2 and 4years of age. The recall
problem suggests that retrospective assessments of children or adolescents does not
accurately identify the age of onset and developmental trajectories of physical aggres-
sion use or of chronic physical aggression.
From the available international data on the development of physical aggression
during childhood and adolescence, we can conclude that: (1) the vast majority of
preschool children use physical aggression; (2) the vast majority also learn with age
to use other means than physical aggression to solve problems; (3) some children
need more time than others to learn alternative solutions to physical aggression; (4)
girls learn more quickly than boys to use alternatives to physical aggression; (5) most
of the cases of chronic physical aggression during adolescence were chronic cases
since early childhood; (6) attempting to use retrospective information to determine

age of onset of physical aggression is futile because recall of a specific age is unre-
liable and in all cases, it will have been in early childhood (Tremblay, 2010). Thus
children, preadolescents, and adolescents do not learn to physically aggress; they
learn not to physically aggress. This learning starts in early childhood and appears to
be highly linked to the development of our brains capacity to inhibit primitive preda-
tory and fear impulses (Panksepp, 1998; Raine, 2013).
arly Bio-Psycho-Social Mechanisms Leading

E
totheDevelopment ofChronic Physical Aggression
The number of studies on the environmental risk factors for early childhood chronic
physical aggression is limited since there are few studies that have traced the devel-
opment of physical aggression during early childhood. However, the major studies
used large population samples with repeated assessments during early childhood
(Campbell etal., 2010; Ct etal., 2006; Hay etal., 2011; Early Child Care Research
Network, 2004; Tremblay etal., 2004).
Significant early risk factors identified by using multivariate analyses can be
grouped into four categories: maternal characteristics-lifestyle-mental health, fam-
ily characteristics, maternal parenting, and child characteristics. Maternal and fam-
ily characteristics are key for preventive interventions because they can be used to
identify at-risk pregnant women (e.g., Olds etal., 1998). Boys are always found to
be more at risk than girls. Mothers young age at birth of her child (first or target
child in the study), mothers antisocial behavior during adolescence, mothers
smoking during pregnancy, mothers depression, mothers low level of education,
and mothers hostilecoercive parenting were all found to be significant risk factors.
Family low income, family dysfunction, lack of stimulation, and the presence of
siblings were also significant risk factors. These risk factors generally correspond to
those found when older children and adolescents are studied.
Early risk factors are useful to find early causal mechanisms and guide early
preventive interventions. An interesting question is the extent to which early risk
factors for early physical aggression are similar to those for a somewhat different
diagnosis. In one of the studies described above, early risk factors for the chronic
trajectory of hyperactivity between 2 and 7years were similar to those identified
for physical aggression: maternal prenatal smoking, maternal depression, early
hostile parenting practices, and male child (Romano, Tremblay, Farhat, & Ct,
2006). These early risk factors are also similar to those found for the Conduct
Disorder trajectories in the Dunedin longitudinal study (Odgers etal., 2008): low
socioeconomic status, mothers mental health, mother Intelligent Quotient (IQ),
parental conviction, inconsistent discipline, and maltreatment (see also Moffitt &
Scott, 2008; Robinson etal., 2008).
Can we expect different factors for other forms of antisocial-CD behavior prob-
lems? Similarities in early risk factors for CD, ADHD, and physical aggression may
be due to the fact that they are all highly loaded on overt-externalizing behavior.
Unfortunately, we do not have comparative studies during early childhood. It would
be useful to know if there are prenatal characteristics that distinguish individuals on
chronic trajectories of physical aggression from those on chronic trajectories of
theft since they appear to differ in terms of cognitive development (Barker etal.,
2007). Are there common early parentfamily environment problems with all these
early chronic trajectories which become diversified because of later environmental
determinants, or do we have common environments and different genetic profiles
which put individuals on a physical aggressive versus nonaggressive track? Teasing
out the common and unique risk factors will involve differentiating pure from
comorbid groups. As discussed below, experimental preventive interventions may
be the best approach to understand the links between the risk factors and the differ-
ent types of developmental trajectories.
Genetic Risk Factors
There is good evidence from quantitative genetic studies (mainly twin studies; e.g.,
Arseneault et al., 2003; Dionne, Tremblay, Boivin, Laplante, & Prusse, 2003;
Lacourse etal., 2014) that genetic factors are associated with the use of frequent
aggression during early childhood. For example, a longitudinal study of physical
aggression development with a large sample of twins from 18 to 50months of age
(Lacourse et al., 2014) concluded that genetic factors explained between 50 and
63% of the variance in frequency of physical aggression. The effects of genetic fac-
tors at 20 months substantially decreased over time, while new genetic effects
appeared at 32 and 50 months. Two separate sets of uncorrelated genetic factors
accounted for the variation in initial level and change over time.
There is also an increasing large body of evidence from molecular genetic studies
that genes related to the functioning of the serotonergic system (e.g., the serotonin
transporter gene; SLC6A4; Monoamine Oxydase A, MAOA) are implicated in the
development of individuals with CPA problems during adolescence and early adult-
hood. Following studies with monkeys (Barr & Discroll, 2014) which inspired the
first developmental study of human males (Caspi et al., 2002), numerous studies
investigated the statistical interaction between childhood adversity and monoaminergic-
related genes. A number of these studies suggest that males with a genotype confer-
ring low levels of monoamine oxidase (MAOA-LPR) are more at risk of different
forms of behavior problems (including violence) during adolescence and adulthood
when brought up in an adverse environment (Laucht, Brandeis, & Zohsel, 2014).
Interestingly, the adverse environments tested included prenatal toxic exposure such
as smoking during pregnancy. Wakschlag et al. (2010) showed that women who
smoked during pregnancy were more likely to have male offspring with conduct dis-
order during adolescence if their child had a genotype conferring low levels of
MAOA-L, but they were more likely to have female offspring with conduct disorder
during adolescence if their child had a genotype conferring high levels of MAOA-H.
Together with the early development of aggression studies and the twin studies
these statistical GxE studies clearly point to the importance of the very early envi-
ronment in the developmental chain of events leading to chronic physical aggres-
sion. However, to elaborate and implement effective early preventive interventions
we need to go beyond statistical interactions to bio-psycho-social interactions.
Epigenetic Mechanisms
New evidence from DNA expression studies (epigenetics) suggests that the numer-
ous bio-psycho-social environmental risk factors related to the mother may start to
have their impact on the childs developing brain and eventual self-control problems
during fetal life and soon after, through their impact on gene expression. The first
study to point in that direction was done on maternal behavior of rats a decade ago
(Weaver et al., 2004). We now have evidence that the quality of the bio-psycho-
social prenatal and early postnatal environment has an impact on the expression of
genes that are essential for the normal development of our brain (Booij, Tremblay,
Szyf, & Benkelfat, 2015; Meaney, 2010). Offspring of women who have a history
of behavior problems, who smoke, drink alcohol, and are exposed to abuse, are at
high risk of modifying the DNA expression programming of their offspring during
the prenatal and early postnatal period. It could be hypothesized that this modifica-
tion in gene expression programming, in turn, can modify brain development and
self-control programming. In this section, we discuss two developmental neurobio-
logical pathways that could be affected by prenatal and early postnatal epigenetic
effects of adverse bio-psycho-social environments.
The Serotonergic System
Following numerous studies showing that 5-HT-related genes like MAOA and
SCL6A4, in interaction with early adversity affects behavior like aggression, an
emerging number of studies now have shown that early adversity was associated with
altered levels of methylation in 5-HT-related genes like SLC6A4 and MAOA (see
Booij, Wang, Levesque, Tremblay, & Szyf, 2013). A few studies focused on the con-
sequences for aggression. For instance, Beach, Brody, Todorov, Gunter, and Philibert
(2011) showed that women victims of child sexual abuse had overall hypermethyl-
ation of the SLC6A4 (5HTT) promoter region (Beach etal., 2011). In addition, they
observed a significant association between DNA methylation in the SLC6A4 pro-
moter with symptoms of antisocial personality disorder (ASPD) in women that was
also partly mediated by 5-HTTLPR polymorphism (Beach etal., 2011). These find-
ings may suggest that exposure to sexual abuse in childhood may create long-lasting
epigenetic changes in the SLC6A4 gene promoter and lead to female antisocial
behavior. In one of our own studies, we showed that SLC6A4 promoter methylation
in white blood cells of adults was associated with higher childhood physical aggres-
sion. Notably, methylation in those CpG sites that were significantly associated with
childhood aggression, were also associated with lower brain 5-HT synthesis in the
orbitofrontal cortex (Wang etal., 2012), a brain region previously been found to be
associated with aggression (Booij etal., 2010). Although causality cannot be estab-
lished in any of these studies, given the important role of SLC6A4 in brain develop-
mental processes and development of psychopathology, it can be hypothesized that
early adversity may alter methylation levels of the SLC6A4 promoter, with conse-
quences for brain development and emotion regulation (Booij, Tremblay, etal., 2015).
This was further supported by very recent reports that greater SLC6A4 methylation,
assessed in whole blood DNA, was associated with lower hippocampal volume, a
brain region with rich 5-HT innervation (Booij, Szyf, etal., 2015) and with altered
neural responses in the limbic system during processing of emotional stimuli (Frodl
etal., 2015). Similarly, we recently observed greater methylation in the MAOA pro-
moter in individuals with ASPD relative to controls (Checknita etal., 2015).
Hence, taken together these findings suggest that, DNA methylation in 5-HT
genes may be a physiological mechanism of how gene and the early environment
interact. Given that the 5-HT system starts to develop very early in development (i.e.,
in the first trimester of human pregnancy), it could be hypothesized that in utero and
early postnatal environmental adversity can disrupt the 5-HT system in early devel-
opment (Booij, Tremblay, etal., 2015). Taking into account its crucial role in brain
developmental processes, such disruption in 5-HT homeostasis, in turn, may predis-
pose the individual to structural and functional alterations in brain circuits such as the
frontal cortex and the amygdala, previously identified as key regions in the modula-
tion of aggression (Davidson, Putnam, & Larson, 2000), social-cognitive processes,
and prosocial behaviors (Eisenberger, 2013; Emonds, Declerck, Boone, Seurinck, &
Achten, 2014; Rameson, Morelli, & Lieberman, 2012). Although it is clear that the
influence of 5-HT and associated proteins on brain development depend on its inter-
action with other biological systems, it could be speculated that early alterations in
these brain circuits, direct or indirectly resulting from a disruption in 5-HT homeo-
stasis, may be partly responsible for an increased difficulty to learn to use alternatives
for physical aggression in the first years of life, and/or increase the susceptibility to
express enduring patterns of aggressive behaviors. Longitudinal cohort studies,
incorporating repeated measures of brain development and molecular processes in
the first years of life, are needed to further test such hypotheses.
The Immune System
Using peripheral blood cells DNA from monocytes and T cells, we recently reported
an association between childhood CPA in men and differential DNA methylation in
regulatory regions of cytokine and transcription factor genes (Provenal et al.,
2013). Moreover, these cytokines were also shown to be repressed in men with CPA
compared to men on normative developmental trajectory of aggressive behavior
(Provenal etal., 2013). Interestingly, one of these downregulated cytokine in men
with CPA, Interleukin-6 (IL-6), was previously shown to be involved in aggressive
behavior in mice since its knockout (IL-6 (/)) resulted in increased aggressive
behavior phenotype in these mice (Alleva etal., 1998). In humans, a growing body
of research also suggests that inflammatory cytokines might have systemic effects
in addition to their traditional roles in the immune response. Indeed, recent studies
have shown that cytokines are associated with various behavioral disorders such as
anxiety, depression, suicide, childhood mood disorder, and posttraumatic stress dis-
order (Bauer, Wieck, Lopes, Teixeira, & Grassi-Oliveira, 2010; Dowlati etal., 2010;
Groer & Morgan, 2007; Hoge etal., 2009; Janelidze, Mattei, Westrin, Trskman-
Bendz, & Brundin, 2010; Kawamura etal., 2001; Koo & Duman, 2008; OBrien,
Scott, & Dinan, 2004 Pace, Hu, & Miller, 2007; Smith etal., 2011; von Kanel etal.,
2007) as well as aggression (Marsland, Prather, Petersen, Cohen, & Manuck, 2008;
Suarez, Lewis, & Kuhn, 2002). Moreover, early life stress such as social isolation
and prenatal anxiety has been found to alter the immune system (Barreau, Cartier,
Ferrier, Fioramonti, & Bueno, 2004; Danese, Pariante, Caspi, Taylor, & Poulton,
2007; OConnor, Moynihan, & Caserta, 2014; Powell et al., 2013; Sloan et al.,
2007). Previous studies from our group and others that have examined associations
of genome-wide DNA methylation profiles with adverse exposures have pointed to
immune pathways, both in the brain and in the periphery. Maternal deprivation in
rhesus macaques (Provenal etal., 2012), early life socioeconomic position (Borghol
etal., 2012; Kieseppa, Partonen, Haukka, Kaprio, & Lonnqvist, 2004), child abuse
(Suderman etal., 2014) and PTSD (Mehta etal., 2013; Smith etal., 2011; Uddin
etal., 2010), all found DNA methylation associations in promoters regulating genes
in the immune response pathways. Together, these results suggest that immunoregu-
lators are responsive to early life stress and might be involved in aggression where
DNA methylation could be one of the mechanisms that mediate this association.
These immunoregulators could influence brain circuitry and behavior through a
wide variety of other biological systems, including the 5-HT system and the HPA-
axis. For instance, cytokines have been shown to influence 5-HT synthesis and
transporter expression (Capuron & Miller, 2011). Effects of immunoregulators also
occur through their action on the hypothalamic-pituitary-adrenal (HPA) axis previ-
ously shown to play a role in aggression (Birger etal., 2003; Craig, 2007; Murni
etal., 2010).
The HPA axis is considered to be the most important system in stress regulation.
Upon its activation corticotrophin releasing hormone (CRH) and vasopressin
(AVP) are released from the hypothalamus and stimulate adrenocorticotropic hor-
mone (ACTH) release from the pituitary into the blood. This results in cortisol
secretion from the adrenal cortex. The cellular actions of the cortisol are mediated
by its binding to the glucocorticoid receptor (GR) and the mineralocorticoid recep-
tor that act as transcription factors (Larsson, Thorbert-Mros, Rymo, & Berglundh,
2012; Sanchez, Arnt, Hyttel, & Moltzen, 1993) and are expressed in most tissues.
Once activated, GR and MR translocate into the nucleus where they can exert their
function as transcription factors regulating adaptive responses to stress, including

metabolism, immune activation, and cell proliferation and differentiation. At mul-
tiple levels of the HPA axis, the activation of the GR will initiate a negative feed-
back loop that is responsible for terminating the stress response and therefore the
secretion of cortisol. A decrease in GR expression/activation is generally associ-
ated with an increase in the response to stress due to an impaired negative feed-
back. In addition, there is strong evidence of a crosstalk between the immune
system and the brain through the HPA axis. It is well known that increases in glu-
cocorticoid levels in response to activation of the HPA axis results in a profound
silencing of gene expression of pro-inflammatory proteins and cytokines. Also, it
was shown that early life social class can affect the expression of genes bearing
response elements to transcription factors regulating immune genes such as CREB/
ATF, NFKB, and GR (Miller etal., 2009). Thus, the effects observed on the immune
system in relation to aggression could be due to a dysregulated HPA axis and there-
fore alterations in the cortisol release and actions.
In general, correlations have been found between reduced cortisol levels and
increased aggression levels in adolescents and young men (Loney, Butler, Lima,
Counts, & Eckel, 2006; Popma etal., 2007; Shirtcliff, Granger, Booth, & Johnson,
2005). In contrast, one study showed that boys with conduct disorder (CD) had
elevated salivary cortisol levels compared to those without CD (van Bokhoven
et al., 2005), and boys with an aggressive form of CD had even higher cortisol
levels. A strong correlation was also observed between reactive aggression and
elevated cortisol. In this context, it is important to note that maltreatment in child-
hood also leads to low basal cortisol in association with conduct and aggressive
disorders (Tarullo & Gunnar, 2006). Together, these results indicate that both
hyper- and hypoactive HPA axis might explain childrens aggression, where hyper-
activity may be involved in reactive aggression and hypoactivity may be involved
in proactive aggression.
Prenatal stress exposure to high levels of glucocorticoids was also shown to pro-
mote aggressive behavior (Glover, 2011). In chicken, in ovo injection of high dose of
cortisol during embryonic development was shown to increase aggressive behaviors
through alteration of the HPA axis and serotonin system (Ahmed, Ma, Ni, Zhou, &
Zhao, 2014). Reduced hypothalamic levels of GR protein and CRH mRNA levels
accompanied by an increase in DNA methylation in the GR and CRH gene promot-
ers were observed in the chicks. Here, prenatal cortisol exposure caused epigenetic
reprogramming of critical genes that in turn, altered the HPA axis and enhanced
aggressive behavior.
In rats, exposure to early adverse life experiences was shown to induce high and
sustained rates of increased aggressive behavior in adulthood (Marquez et al.,
2013). This study also showed that peripubertal exposure to stress (fear-induction
experiences) induces pathological aggression in male rats. These peripubertal
stressed rat also exhibited hyperactivity in the amygdala and hypoactivity in the
medial orbitofrontal cortex after exposure to social challenge. Interestingly, these
neuroimaging brain activity data were accompanied by a sustained increase in
MAOA expression in the PFC of stressed animals that is likely to be explained by
epigenetic modulation. Indeed, they found an increase in histone 3, but not histone
4, acetylation levels in the promoter of the MAOA gene. Histones acetylation are
known to promote gene transcription by increasing the accessibility to active tran-
scription regulators binding (Kuo & Allis, 1998) and especially histone 3 acetyla-
tion have been shown to play a role in regulating long-term changes in gene
expression (Tsankova, Renthal, Kumar, & Nestler, 2007). Together, these finding
with the previous work on MAOA gene support the hypothesis that either MAOA
hypo- or hyperactivity contribute to pathological aggression (Nelson & Trainor,
2007) possibly through epigenetic programming.
The epigenetic association studies presented above were mainly focused on can-
didate genes that were either suspected or were previously shown to be involved in
aggression. We also used an unbiased genome-wide approach (Mehta etal., 2013) to
analyze men T cell genomes and identified significant associations of DNA methyla-
tion levels with childhood CPA in 448 distinct gene promoters involved in biological
pathways related to behavior and immune function, and their colocalization in
genomic clusters (Provenal etal., 2014). Interestingly, some of these differentially
methylated genes, such as the AVP receptor 1A (AVPR1A), SLC6A3 (dopamine
transporter), and serotonin receptor 1D (HTR1D), were previously associated with
aggressive phenotype in humans (Guo, Roettger, & Shih, 2007; Vage etal., 2010;
Vaughn, Delisi, Beaver, & Wright, 2009) and animals (Ferris etal., 2006; Hammock,
Lim, Nair, & Young, 2005). As anticipated from our previous study in cytokine genes
(Provenal etal., 2013), the inflammatory and immune biological function with spe-
cific signalling pathway such as cytokines signalling between immune cells, IL-6
and IL-10 signalling were found enriched with genes differentially methylated in
men with CPA.Specific cytokines and receptors involved in these pathways were
previously shown to be involved in aggression and human mood disorders such as
IL1R1 and IL1RN (Pesce et al., 2011). Together, these findings suggest a well-
defined, genome-wide epigenetic pattern associated with chronic physical aggres-
sion in men.
In another study, we observed similar DNA methylation signatures associated
with childhood CPA in women (n=430 promoters) as seen in men where 31 gene
promoters were significantly associated in both sexes (Guillemin et al., 2014).
Interestingly, a significant portion of this overlap is due to identical genomic sites
being differentially methylated in a gender-independent fashion. The almost perfect
overlap between functional categories represented by both men and women signa-
tures provides further evidence for these signatures to be, at least in part, associated
with aggression rather than confounding factors. Here also, specific genes involved
in serotonin metabolism and HPA axis regulation, previously shown to be involved
in aggression, were found differentially methylated in women with childhood
CPA.These HPA-regulating genes (NR3C1 and CRHBP) were only found differen-
tially methylated in women with CPA.This may be explained in part by the fact that
the HPA axis negative feedback control have been shown to be more sensitive in
females than in males (Keck etal., 2002). These sex-specific and sex-independent
components of the epigenetic signature are consistent with the existence of sex dif-
ferences and similarities observed in human physical aggression.
I mplications forPrevention ofChronic Physical Aggression

andOther Behavior Problems (e.g., ADHD, ODD, CD)
In one of the first modern longitudinal psychopathology study on conduct problems,

Lee Robins (1966) showed that individuals who had serious behavior problems as
adults had behavior problems during childhood. Twenty-six years later, she wrote
There is probably no area of behavior or psychiatric disorder riper for an experi-
mental design than conduct problems. We know the population at high risk: it is
boys living in poor areas of inner cities in their first school years Conduct prob-
lems tend to be long-lasting, enduring often into adulthood, and almost always
beginning by the midteens, so evaluation can take place at any time from age 10 to
15 or 16. We also have an abundance of hypothetical causal agents to consider as
possible candidates for prevention. (Robins, 1992).
Robins and her colleague Joan McCord had been highly supportive of such an
experiment that was initiated in 1984 when physically aggressive/hyperactive kin-
dergarten boys from a longitudinal study of families living in poor areas of a North
American inner city (Montreal, Canada) were randomly allocated to an intensive 2
year preventive intervention (Tremblay, Pagani-Kurtz, Msse, Vitaro, & Pihl, 1995).
The intervention involved parent training sessions during home visits every 2 weeks
and regular social skills training at school in small peer groups where two disruptive
boys were joined by three to four highly prosocial boys. The boys from the experi-
mental and control groups as well as all the other boys recruited for the longitudinal
study were assessed yearly until age 15 years and then at longer intervals from
17years onwards. Comparisons of the preventive intervention group and the control
group up to age 24years showed important significant impacts: better school adjust-
ment during elementary school, less delinquency during adolescence (physical
aggression, vandalism, and theft), less alcohol and drug abuse during adolescence,
more boys from the prevention group finished high school, and less had criminal
records from 18 to 24years of age (Boisjoli, Vitaro, Lacourse, Barker, & Tremblay,
2007; Castellanos-Ryan, Sguin, Vitaro, Parent, & Tremblay, 2013; Lacourse etal.,
2002; Tremblay etal., 1995; Vitaro, Barker, Brendgen, & Tremblay, 2012).
Thus, Lee Robins advice to experiment preventive interventions at school entry
helped provide evidence that intensive interventions during the first school years for
boys from poor areas of inner cities can deflect developmental trajectories leading
to juvenile delinquency, substance abuse, school failure, and criminal behavior.
In this chapter, we provided an overview of the evidence accumulated over the past
two decades that suggests preventive interventions should start much earlier than the
early school years because by the time of pregnancy the risk factors can be identified
and the underlying mechanisms appear to be well in place. The evidence that Lee
Robins was using in 1992 was essentially social and behavioral. We have now linked
this social and behavioral evidence to evidence from genetic studies, brain develop-
ment studies, immune system development studies and, totally unexpected in 1992,
one of the basic intergenerational mechanism appears to be the impact of the bio-
psycho-social environment on gene expression from conception onwards.
We can now reformulate Lee Robins 1992 challenge by saying: There is prob-
ably no area of behavior or psychiatric disorder riper for an experimental design
during pregnancy than conduct problems. We know the population at high risk: it
is pregnant women with a history of behavior problems and low levels of education,
they mated with a man from a similar background, they have poor marital relations,
they are young at their first pregnancy, often depressive, stressed, malnourished, con-
tinue to smoke during pregnancy, and generally live in poor areas of inner cities
their childrens behavior problems have started by the end of the first year after birth
and tend to be long-lasting, enduring often into adulthood so evaluation of the
effectiveness of the intervention can take place at any time from age 12months to
early adulthood at numerous bio-psycho-social levels. We also have an abundance
of hypothetical causal agents to consider as possible candidates for prevention.1
Furthermore, the content of the preventive interventions most likely to be effec-
tive for the at-risk pregnant women and their family during early childhood have
been well tested over the past 35years.
Nurse home visitation: Between 1978 and 1980, the Elmira Nurse Home Visitation
program randomly distributed to four treatment conditions 400 young pregnant
women who had no previous live birth, were unmarried or poor. Results showed
numerous significant impacts on the mothers and the children from early childhood
to early adulthood (Eckenrode et al., 2010; Olds et al., 1998). One of the most
encouraging long-term impacts was the reduction of early adulthood adjustment
problems for the female offspring (less arrests, less convictions, and less childbirth;
Eckenrode et al., 2010). This is an indication that an intervention starting during
pregnancy has an impact on the third generation through its impact on female chil-
dren, suggesting that the intergenerational reduction of male behavior problems
needs to start by the reduction of female problems.
One of the limits of the Nurse Home Visitation Program described above is probably
the fact that the home visits stop when the children are 24months. As described in the
first part of this chapter, childrens behavior problems reach their peak between 18 and
42months. Parents need intensive support during this period. One form of effective sup-
port appears to be placement of their children in quality daycare (Ct etal., 2007).
Participation in quality daycare: Between 1962 and 1965, the High/Scope Preschool
preventive intervention program was implemented with low IQ 34 year olds
(N=123) living in poverty with parents who had low education. Impressive reduc-
tions of antisocial behavior during adolescence and adulthood were observed
(Heckman, Pinto, & Savelyev, 2013; Schweinhart etal., 2005). Second, in 1973 the
Mauritius Child Health Project randomly distributed 3 year olds (N = 200) with
varying risk for schizophrenia to enrich and control nursery schools. Follow-up
1
Robins had of course anticipated the importance of starting preventive interventions earlier than
at school entry. See: Robins, L. N., & Earls, F. J. (1986). A program for preventing antisocial
behavior for high-risk infants and preschoolers: A research prospectus. In R. L. Hough, P. A.
Gongla, V.B. Brown, & S.E. Goldston (Eds.), Psychiatric epidemiology and prevention: The pos-
sibilities (pp.7384). Los Angeles, CA: Neuropsychiatric Institute.
assessments at 17 and 23years showed impressive reductions of antisocial behavior

and schizotypal personality (Raine, Mellingen, Liu, Venables, & Mednick, 2003;
Raine etal., 2001; Venables, Raine, Mednick, Schulsinger, & Dalais, 2006).
Thus, the preventive intervention programs from pregnancy to school entry for
children from low socioeconomic environments have been created and shown to
have long-term impacts on antisocial behavior. However, there are many gaps to fill
with reference to the development of chronic physical aggression. First, none of
these experimental studies monitored the development of physical aggression dur-
ing early childhood, and the monitoring was sketchy during childhood, adolescence,
and adulthood, in part because these interventions were not designed to prevent
chronic physical aggression (Sidora-Arcoleo etal., 2010). Secondly, the children
that were targeted were not necessarily at high risk of chronic physical aggression.
The families were from low socioeconomic environments, but the parents were not
chosen because they had a disturbed behavior background which is characteristic of
children with chronic physical aggression. Finally, with a few notable exceptions
(Bentley etal., 2013; Raine, Liu, Venables, & Mednick, 2006) there is an almost
complete absence of biological information on the children and their parents,
including brain developmental, genetic and epigenetic information. Hopefully,
researchers interested in the development and prevention of chronic physical aggres-
sion will now take advantage of the bio-psycho-social knowledge and technologies
accumulated over the past four decades to take the big leap forward and use them in
randomized control trials during pregnancy and early childhood.
uggestions fortheNext Generation ofPreventive

S
Intervention Experiments WithinaGxE Framework
Randomized control trials of preventive interventions can serve two purposes: (a)
they can evaluate the short- and long-term impacts of preventive interventions; (b)
they can test causal hypotheses. Randomized evaluations of preventive interven-
tions are essential to test to what extent the intervention is achieving its goals, but
they are also the best mean of testing to what extent causal hypotheses are founded.
Whereas animal researchers do not need to frame their experiments within a preven-
tion context, for obvious ethical reasons, experimental tests of causal hypotheses
with humans at high risk of health problems must be framed in the context of a
preventive intervention. Here, we give examples of experiments designed to prevent
the development of chronic physical aggression that can both test impacts of the
environment on DNA methylation and their developmental consequences.
Example of a prevention hypothesis: Nurse home visitation during pregnancy fol-
lowed by quality child care during early childhood will have more of an impact on
the prevention of chronic physical aggression in children of mothers with a history
of behavior problems compared to interventions with only one component and com-
pared to no intervention.
Causal (mechanistic) hypothesis: (a) Nurse home visitation and child care will each
have impacts on a cascade of biological developments: DNA methylation, the
immune system, the serotonergic system, brain morphology, and brain functioning;
(b) the biological impacts will be associated with impacts on a wide range of behav-
ior problems (chronic physical aggression, ODD, CD, ADHD).
We clearly have the intervention tools and the assessment tools to test the preven-
tion hypothesis as well as the causal hypothesis. Prevention studies in most areas of
physical and mental health are generally done with singletons. The basic design to
test the prevention and causal hypotheses can be used with singletons and with
twins. We first describe the singleton design so that it will be easier to understand
the advantages of using monozygotic twins.
Preventive experiments with singletons: Ideally, primiparous pregnant women at the
beginning of their pregnancy who have at least three risk factors associated to
chronic physical aggression during early childhood would be allocated to the fol-
lowing experimental groups: (a) nurse home visitation up to age 3years+center-
based child care starting at 12months until school entry; (b) nurse home visitation
up to age 3years only; (c) center-based child care starting at 12months until school
entry only; (d) control group receiving care as usual.
The impact of the interventions would be assessed with evaluations of mothers
lifestyle (eating, smoking, drinking, stress, etc.) and mental health, ultrasound track-
ing of the childs physical growth, DNA methylation of child from birth onwards
(every 6months), development of the serotonergic and immune system, brain-imag-
ing measures, repeated regularly at critical phases during early development to
understand the neurobiological underpinnings of problem behavior and efficacy of
preventive intervention.
One of the limitations of this study would be the lack of control over the childs
genotype. This limit could be overcome by statistically taking the genotype into
account or by adding the genotype of the parents in the randomization procedure.
However, this is where the monozygotic twin design can help since both twins are
genetically identical, providing the best control for the other twin.
Preventive experiments with monozygotic twins: It appears that researchers in the
field of space flights have understood that MZ twins offer the best experimental
approach to test the effects of living in outer space. Both twins are trained for the
space flight but one member of the pair travels while the other is used as a control
for any effect of the space flight (Smith etal., 2014; Zwart, Morgan, & Smith, 2013)
(http://science.nasa.gov/science-news/science-at-nasa/2014/10apr_twins/). This
approach should be used more often in health research, and we believe that it is one
of the best designs to study the effect of the prenatal environment (another form of
space capsule!) as well as the postnatal environments on DNA methylation, brain
development, and behavior problems.
There are many different angles that can be taken with the MZ experimental design.
Lets first take the simplest design using the approach described above for singletons.
Primiparous pregnant women of monozygotic twins who have at least 34 risk factors
associated to chronic physical aggression during early childhood would be allocated
to the following experimental groups: (a) nurse home visitation up to age 3years+cen-
ter-based child care starting at 12months until school entry for both twins; (b) nurse
home visitation up to age 3years only and center-based child care starting at 12months
until school entry for only one of the twins; (c) control group receiving care as usual.
We would expect that when the intervention is the same for both twins they will be
more similar on all the outcomes than when the intervention is different. A variant to
this basic approach would be to use the frequent birth weight discordance among MZ
twins (which is generally related to an in utero environmental difference, e.g., blood
flow, and is often related to worse outcome in terms of physical and mental health,
Asbury, Dunn, & Plomin, 2006; Lehn etal., 2007; Lewi etal., 2007) to verify to what
extent the more at-risk twin (the smaller one) will benefit from an enriched postnatal
environment (e.g., child care) and catch up with his identical twin in terms of physical
and mental health. For example, the smaller twin at birth would be the one who
receives the day care enrichment. To the extent that the quality of the social environ-
ment is highly important for bio-psycho-social development, we would hypothesize
that the smaller twin in the quality day care environment will have a better bio-psy-
cho-social development than the larger twin at birth.
Beside the genetic control advantage of using MZ twins for preventive interven-
tions, there is also the social advantage. Parents of twins, especially those with the
risk factors for behavior problems, are much more likely to suffer from the burden
of taking care of two infants, than parents of singletons. This situation should help
both to find funds for the experiment and support from ethic committees.
Breast-feeding is another target that could be used in an experimental design.
Breast-feeding has been shown to be associated with many positive behaviors,
including increased maternal sensitivity and increased attachment security (e.g.,
Tharner etal., 2012). One possible mechanism for these outcomes is the stimulating
effects of breast-feeding on the production of the hormone oxytocin (see, e.g., Febo,
Numan, & Ferris, 2005). Oxytocin has also been shown to have a positive influence
on child development (Carter, 2003; Insel, 2010) and experimental placebo con-
trolled studies using intranasal oxytocin has shown a positive effect of oxytocin on
prosocial behaviors (MacDonald & MacDonald, 2010). An important question to
address would be; (a) to what extent breast-feeding can enhance the efficacy of
preventive interventions for aggression problems in at-risk populations?; (b) to what
extent the mechanism of breast-feeding impact involves the serotonin system and
immune regulation and HPA-axis functioning through DNA methylation.
Twins are less likely to be breast-fed (Lutsiv etal., 2013; McDonald etal., 2012;
Ostlund, Nordstrm, Dykes, & Flacking, 2010) and mothers of twins that do not
breast-feed or do not breast-feed for extended periods share characteristics of moth-
ers with chronically aggressive children: young, low education, and smoking during
pregnancy (e.g., Ostlund et al., 2010). In cases of important birth weight discor-
dance, the smaller twin is probably still less likely to be breast-fed.
Thus, this at-risk populations could be targeted for an RCT of prenatal and postna-
tal support to breast-feeding. The breast-feeding support could be randomized between
pairs as well as within pairs. In cases where breast-feeding is not possible the babies
could be bottle fed by their mothers with breast or formula milk (Rizzi etal., 2013).
These experimental designs could very well address one of the outstanding chal-
lenges in current epigenetic studies of behavior in general and aggressive behavior in
particular: defining causation between particular DNA methylation states and aggres-
sion. Studies to date examined adult DNA.The outstanding questions are whether DNA
methylation preceded or followed development of chronic aggressive behavior as well
as whether these DNA methylation changes are a cause or effect of aggressive behavior.
A related question is whether the DNA methylation changes that have been previously
shown to differentiate chronic aggressive adults from other groups are triggered by
genetic differences between the groups. Addressing these questions is critical for devel-
oping new epigenetic-based approaches to prediction, diagnosis, prognosis, and inter-
ventions in chronic aggression (e.g., Williamson, Harris, Beynon, & Jenkins, 2015). The
twin pair design proposed above will enable establishing DNA methylation markers of
aggression that are genetically independent. Although causation is extremely difficult
to establish in human studies the experimental-longitudinal design will establish whether
DNA methylation markers of aggression appear early and whether they are predictive of
adult chronic aggression. These studies will provide the first proof of principle for early
DNA methylation predictors of aggression. The RCT intervention design will allow us
to test whether DNA methylation markers of aggression are reversed by interventions
that reverse chronic aggression thus establishing a causal link between DNA methyla-
tion and aggression outcomes as well as allowing the delineation of prognostic DNA
methylation biomarkers. An epigenetic underpinning for aggression that would
emerge from these studies will provide proof of principle for applying epigenetic
behavioral and therapeutic interventions to alleviate the burden of aggression and for
developing DNA methylation markers that will provide tools for prediction as well as
proximal objective markers for following up interventions, particularly epigenetically
based interventions (e.g., Williamson etal., 2015).
To summarize, the challenges for a research agenda that would advance the field of
gene-environment bio-psycho-social interactions (GEX) with reference to the devel-
opment of chronic physical aggression and probably many other related problems
(e.g., conduct disorder, oppositional-defiant disorder, attention-deficit hyperactivity
disorder) are:
(a) To enrol in longitudinal studies, large samples of primiparous pregnant women
with a history of behavior problems (ideally singletons and twin pregnancies)
(b) To integrate in these longitudinal studies, preventive RCTs with intensive pre-
ventive interventions that are likely to change these womens lifestyle during
pregnancy and their maternal care from birth onwards
(c) To genotype these women and their spouse
(d) To epigenotype these women before and throughout pregnancy
(e) To regularly monitor the family environment from the prenatal period onwards
(f) To regularly monitor sibling and peer environments during early childhood
(g) To regularly monitor parenting behavior of both parents from birth onwards
(h) To regularly monitor offspring (female and male) development with a specific
focus on the epigenomic profile, brain development through imaging, physical
growth, cognitive development, and behavior development.
Thus, we suggest that we will substantially advance our understanding of the

mechanisms that lead to chronic physical aggression if we regularly monitor, from
pregnancy through early childhood, the changes in epigenetic profiles, brain struc-
ture and processes, cognition, language, as well as changes in socialemotional
behavior. This monitoring should of course include the changes in the childrens
environment such as parenting practices, family functioning, and peer relationships.
The information obtained from these brain-imaging studies could be used to opti-
mize early cognitive and behavioral preventive intervention strategies. These early
preventive interventions, in turn, would potentially alter functional and structural
brain development at the time of greatest neuroplasticity.
Conclusions
In a recent comment (Lahey, 2015) on a longitudinal study showing that 916year

olds with psychopathology between age 9 and 16years were at risk of psychopa-
thology and dysfunction during adulthood (Copeland, Wolke, Shanahan, & Costello,
2015), Lahey suggested that such studies could be explained by three mechanisms.
First, psychopathology in children and adults could have different causes, but child-
hood psychopathology could directly or indirectly increase the risk for adult psy-
chopathology. Second, that early life may partly or entirely cause childhood and
adulthood psychopathology. Third, that association between psychopathology in
childhood and adulthood could both be due to exposure to environmental conditions
that are present during childhood and adulthood.
We suggest that we will advance knowledge on the causes of these associations
mainly by using an intergenerational preventive intervention design initiated during
pregnancy at the latest because the genetic and environmental factors leading to
psychopathology are generally highly correlated. Parents who had a childhood psy-
chopathology carry with them pervasive high risk environmental conditions (e.g.,
low education, low income, poor neighborhoods, smoking) which can impact child-
hood and adulthood psychopathology through many channels including impacts on
the childrens DNA methylation. We cannot see how studies which do not start close
to conception would disentangle the many causal pathways involved. From this
perspective, the search for causes and preventive interventions of complex psycho-
pathology that start much later than conception appear to be an important waste of
money and brain power.
It is interesting to note that the importance of taking an intergenerational per-
spective concerning preventive interventions comes very close to the conclusion
that was reached 65years ago by Lucien Bovet (1951), a Swiss child psychiatrist, in
his World Health Organisation report on juvenile delinquency: the prevention of
antisocial behavior should start by supporting girls who have behavior problems
and will become the mothers of the next generation of delinquent boys.
Bovet reached that conclusion from the available research at the time which showed
the intergenerational patterns of juvenile delinquency (e.g., Glueck & Glueck, 1950;
Healy & Bronner, 1926). The advances in epigenetic research reported above simply
start describing the mechanisms by which the intergenerational problems are trans-
mitted. There are clearly genetic effects involved in these mechanisms, but the life-
style of the parents, especially the mothers during pregnancy appears to have a
biological impact on gene expression which in turn has an impact on brain develop-
ment, the main instrument a child has to learn to use alternatives to the hard wired
physical aggression response that has been inherited throughout evolution to survive
in the jungle (termed the original sin from a religious perspective, see Tremblay
(2010)). If brain development is compromised (e.g., through epigenetic effects) dur-
ing pregnancy and if the childs parents also lack self-control, it is easy to understand
that the child will lack the cognitive and environmental support needed to develop the
self-control that is essential to learn to use alternatives to physical aggression. It is also
important to note that children born to mothers with a history of mental health prob-
lems are more likely to have a father with similar problems and live in a physical and
social environment which is less likely to provide the support to the development of
self-control that is lacking at home.
We thus conclude that developmental trajectories of physical aggression from
early childhood to adulthood are the consequence of the bio-psycho-social transac-
tions between genetic and environmental endowments. The early environment is
created by the parents own developmental history and has a major impact on physi-
cal aggression development through its impact on gene expression and brain devel-
opment. Mothers are likely to have the greatest impact on early gene expression
because their lifestyle during pregnancy has direct biological impacts on the childs
development. As children grow older, their peers also have an impact on how well
they learn to control physical aggression. Most of these risk factors can be identified
prior to or at the start of pregnancy: mothers behavior problems during adoles-
cence, poor education, first pregnancy at a young age, depression, smoking, dys-
functional relations with the father, and poverty.
Being of the male sex is clearly one of the most robust predictor of chronic physi-
cal aggression from early childhood to early adulthood. However, we need to funda-
mentally revisit our thinking about the early prevention of chronic physical aggression
because, although males are much more affected, females (pregnant women) need to
be our prime target to prevent a new generation of males and females with chronic
physical aggression and similar intergenerational problems.
There is increasing evidence that the perinatal bio-psycho-social environment
which has an impact on gene expression is very largely related to pregnant womens
health status and lifestyle. This epigenetic perspective suggests that successful pre-
vention of many physical and mental health problems may be easier to achieve by
ameliorating the pre and early postnatal environment, rather than by chasing bad
genes (Bernet, Vnencak-Jones, Farahany, & Montgomery, 2007; Gluckman,
Hanson, Cooper, & Thornburg, 2008) or trying to change an individuals environ-
ment once he clearly has physical or mental health problems.
Finally, to be well understood, we have to emphasize that mothers, fathers, and
children should not be blamed for the genes and the environment they inherit at con-
ception and onwards. We are arguing that victims of intergenerational transmission
of physical and mental health problems need to be given intensive support when they
become pregnant to help the new generation achieve better control over their behav-
ior and be able to prevent the next generation of victims.
This chapter was adapted from:
Booij, L., Tremblay, R. E., Szyf, M., Benkelfat, C. (2015). Genetic and early
environmental influences on the serotonin system: Consequences for brain devel-
opment and risk for psychopathology. Journal of Psychiatry and Neuroscience,
40(1), 518. Oct 7. doi: 10.1503/jpn.140099.
Booij, L., Wang, D., Lvesque, M.L., Tremblay, R.E, & Szyf, M. (2013). Looking
beyond the DNA Sequence: the relevance of DNA methylation processes for the
stress-diathesis model of depression. Philosophical Transactions of the Royal
Society B-Biological Sciences, 368(1615), 116. doi: 10.1098/rstb.2012.0251
Provenal, N., Booij, L., & Tremblay, R.E. (2015). The developmental origins
of chronic physical aggression: Early life adversity, epigenetics and impact on
other biological systems. Journal of Experimental Biology, 2018(1), 123133.
doi: 10.1242/jeb.111401
Tremblay, R. E. (2010). Developmental origins of disruptive behaviour prob-
lems: The original sin hypothesis, epigenetics and their consequences for preven-
tion. Journal of Child Psychology and Psychiatry, 51(4), 341367.
References
Ahmed, A.A., Ma, W.Q., Ni, Y.D., Zhou, Q., & Zhao, R.Q. (2014). Embryonic exposure to cor-
ticosterone modifies aggressive behavior through alterations of the hypothalamic pituitary
adrenal axis and the serotonergic system in the chicken. Hormones and Behavior, 65(2),
97105. doi:10.1016/j.yhbeh.2013.12.002.
Alink, L.R., Mesman, J., van Zeijl, J., Stolk, M.N., Juffer, F., Koot, H.M., van Ijzendoorn,
M.H. (2006). The early childhood aggression curve: Development of physical aggression in
10- to 50-months-old children. Child Development, 77(4), 954966.
Alleva, E., Cirulli, F., Bianchi, M., Bondiolotti, G.P., Chiarotti, F., De Acetis, L., & Panerai, A.E.
(1998). Behavioural characterization of interleukin-6 overexpressing or deficient mice during
agonistic encounters. European Journal of Neuroscience, 10(12), 36643672.
doi:10.1046/j.1460-9568.1998.00377.x.
Arseneault, L., Moffit, T.E., Caspi, A., Taylor, A., Rijsdijk, F.V., Jaffee, S.R., Measelle, J.R.
(2003). Strong genetic effects on cross-situational antisocial behaviour among 5-year-old chil-
dren according to mothers, teachers, examiner-observers, and twins self-reports. Journal of
Child Psychology and Psychiatry, 44(6), 832848.
Asbury, K., Dunn, J.F., & Plomin, R. (2006). Birthweight-discordance and differences in early
parenting relate to monozygotic twin differences in behaviour problems and academic achieve-
ment at age 7. Developmental Science, 9(2), F22F31.
Bandura, A., Ross, D., & Ross, S. A. (1961). Transmission of aggression through imitation of
aggressive models. Journal of Abnormal and Social Psychology, 63, 575582.
Barker, E.D., Sguin, J.R., White, H.R., Bates, M.E., Lacourse, E., Carbonneau, R., & Tremblay,
R.E. (2007). Developmental trajectories of male physical violence and theft: Relations to neu-
rocognitive performance. Archives of General Psychiatry, 64(5), 592599. doi:10.1001/
archpsyc.64.5.592.
Barr, C.S., & Discroll, C. (2014). Neurogenetics of aggressive behavior: Studies in primates. In
K. A. Miczek, & A. Meyer-Lindenberg (Eds.), Neuroscience of aggression (pp. 4571).
Heidelberg, Germany: Springer.
Barreau, F., Cartier, C., Ferrier, L., Fioramonti, J., & Bueno, L. (2004). Nerve growth factor medi-
ates alterations of colonic sensitivity and mucosal barrier induced by neonatal stress in rats.
Gastroenterology, 127(2), 524534. doi:10.1053/s0016-5085(04)00865-0.
Bauer, M. E., Wieck, A., Lopes, R. P., Teixeira, A. L., & Grassi-Oliveira, R. (2010). Interplay
between neuroimmunoendocrine systems during post-traumatic stress disorder: A minireview.
Neuroimmunomodulation, 17(3), 192195. doi:10.1159/000258721.
Beach, S.R. H., Brody, G.H., Todorov, A.A., Gunter, T.D., & Philibert, R.A. (2011). Methylation
at 5HTT mediates the impact of child sex abuse on women's antisocial behavior: An examina-
tion of the Iowa Adoptee Sample. Psychosomatic Medicine, 73(1), 8387. doi:10.1097/
PSY.0b013e3181fdd074.
Bentley, M.J., Lin, H.Q., Fernandez, T.V., Lee, M., Yrigollen, C.M., Pakstis, A.J., Leckman,
J.F. (2013). Gene variants associated with antisocial behaviour: A latent variable approach.
Journal of Child Psychology and Psychiatry, 54(10), 10741085. doi:10.1111/jcpp.12109.
Bernet, W., Vnencak-Jones, C.L., Farahany, N., & Montgomery, S.A. (2007). Bad nature, bad
nurture, and testimony regarding MAOA and SLC6A4 genotyping at murder trials. Journal of
Forensic Sciences, 52(6), 13621371.
Birger, M., Swartz, M., Cohen, D., Alesh, Y., Grishpan, C., & Kotelr, M. (2003). Aggression: The
testosterone-serotonin link. The Israel Medical Association Journal, 5(9), 653658.
Boisjoli, R., Vitaro, F., Lacourse, E., Barker, E.D., & Tremblay, R.E. (2007). Impact and clinical
significance of a preventive intervention for disruptive boys: 15-year follow-up. British Journal
of Psychiatry, 191(5), 415419. doi:10.1192/bjp.bp.106.030007.
Booij, L., Szyf, M., Carballedo, A., Frey, N.N., Morris, D., Dymov, S., Frodl, T. (2015). DNA
methylation of the serotonin transporter gene in peripheral cells and stress-related changes in
hippocampal volume: A study in depressed patients and healthy controls. PLoS One, 10(3),
114 (e0119061). doi:10.1371/journal.pone.0119061.
Booij, L., Tremblay, R.E., Leyton, M., Sguin, J.R., Vitaro, F., Gravel, P., Benkelfat, C. (2010).
Brain serotonin synthesis in adult males characterized by physical aggression during child-
hood: A 21-Year Longitudinal Study. PLoS One, 5(6), 19 (e11255). doi:10.1371/journal.
pone.0011255.
Booij, L., Tremblay, R.E., Szyf, M., & Benkelfat, C. (2015). Genetic and early environmental
influences on the serotonin system: Consequences for brain development and risk for psycho-
pathology. Journal of Psychiatry and Neuroscience, 40(1), 518. doi:10.1503/jpn.140099.
Booij, L., Wang, D.S., Levesque, M.L., Tremblay, R.E., & Szyf, M. (2013). Looking beyond the
DNA sequence: The relevance of DNA methylation processes for the stress-diathesis model of
depression. Philosophical Transactions of the Royal Society, B: Biological Sciences, 368(1615),
116. doi:10.1098/rstb.2012.0251.
Borghol, N., Suderman, M., McArdle, W., Racine, A., Hallett, M., Pembrey, M., Szyf, M.
(2012). Associations with early-life socio-economic position in adult DNA methylation.
International Journal of Epidemiology, 41(1), 6274. doi:10.1093/ije/dyr147.
Bovet, L. (1951). Psychiatric aspects of juvenile delinquency. Geneva, Switzerland: World Health
Organization.
Bridges, K.M. B. (1931). The social and emotional development of the pre-school child. London:
Kegan Paul.
Bridges, K.M. B. (1933). A study of social development in early infancy. Child Development, 4(1),
3649.
Broidy, L.M., Nagin, D.S., Tremblay, R.E., Bates, J.E., Brame, B., Dodge, K.A., Vitaro, F.
(2003). Developmental trajectories of childhood disruptive behaviors and adolescent delin-
quency: A six site, cross national study. Developmental Psychology, 39(2), 222245.
Cairns, R.B., & Cairns, B.D. (1994). Life lines and risks: Pathways of youth in our time. NewYork,
NY: Cambridge University Press.
Campbell, S.B., Spieker, S., Burchinal, M., Poe, M.D., & NICHD Early Child Care Research
Network (2006). Trajectories of aggression from toddlerhood to age 9 predict academic and
social functioning through age 12. Journal of Child Psychology and Psychiatry, and Allied
Disciplines, 47(8), 791800.
Campbell, S.B., Spieker, S., Vandergrift, N., Belsky, J., Burchinal, M., & NICHD Early Child
Care Research Network (2010). Predictors and sequelae of trajectories of physical aggression
in school-age boys and girls. Development and Psychopathology, 22(1), 133150. doi:10.1017/
s0954579409990319.
Capuron, L., & Miller,A.H. (2011). Immune system to brain signaling: Neuropsychopharmacological
implications. Pharmacology & Therapeutics, 130(2), 226238. doi:10.1016/j.
pharmthera.2011.01.014.
Carter, C. S. (2003). Developmental consequences of oxytocin. Physiology & Behavior, 79(3),
383397. doi:10.1016/s0031-9384(03)00151-3.
Caspi, A., McClay, J., Moffitt, T.E., Mill, J., Martin, J., Craig, I.W., Poulton, R. (2002). Role of
genotype in the cycle of violence in maltreated children. Science, 297(5582), 851854.
doi:10.1126/science.1072290.
Castellanos-Ryan, N., Sguin, J.R., Vitaro, F., Parent, S., & Tremblay, R.E. (2013). Impact of a
2-year multimodal intervention for disruptive 6-year-olds on substance use in adolescence:
randomised controlled trial. British Journal of Psychiatry, 203(3), 188195. doi:10.1192/bjp.
bp.112.123182.
Checknita, D., Maussion, G., Labont, B., Comai, S., Tremblay, R.E., Vitaro, F., Turecki, G.
(2015). Monoamine oxidase A gene promoter methylation and transcriptional downregulation
in an offender population with antisocial personality disorder. British Journal of Psychiatry,
206(3), 216222. doi:10.1192/bjp.bp.114.144964.
Cole, P.M., Tan, P.Z., Hall, S.E., Zhang, Y.Y., Crnic, K.A., Blair, C.B., & Li, R.Z. (2011).
Developmental changes in anger expression and attention focus: learning to wait. Developmental
Psychology, 47(4), 10781089. doi:10.1037/a0023813.
Copeland, W.E., Wolke, D., Shanahan, L., & Costello, E.J. (2015). Adult functional outcomes of
common childhood psychiatric problems: A prospective, longitudinal study [Online]. JAMA
Psychiatry. doi:10.1001/jamapsychiatry.2015.0730.
Ct, S.M., Boivin, M., Liu, X., Nagin, D.S., Zoccolillo, M., & Tremblay, R.E. (2009). Depression
and anxiety symptoms: Onset, developmental course and risk factors during early childhood.
Journal of Child Psychology and Psychiatry, 50(10), 12011208.
Ct, S.M., Boivin, M., Nagin, D.S., Japel, C., Xu, Q., Zoccolillo, M., & Tremblay, R.E. (2007).
The role of maternal education and non-maternal care services in the prevention of childrens
physical aggression. Archives of General Psychiatry, 64(11), 13051312.
Ct, S.M., Vaillancourt, T., LeBlanc, J.C., Nagin, D.S., & Tremblay, R.E. (2006). The develop-
ment of physical aggression from toddlerhood to pre-adolescence: A nation wide longitudinal
study of Canadian children. Journal of Abnormal Child Psychology, 34(1), 7185.
Craig, I.W. (2007). The importance of stress and genetic variation in human aggression. Bioessays,
29(3), 227236. doi:10.1002/bies.20538.
Danese, A., Pariante, C.M., Caspi, A., Taylor, A., & Poulton, R. (2007). Childhood maltreatment
predicts adult inflammation in a life-course study. Proceedings of the National Academy of
Davidson, R.J., Putnam, K.M., & Larson, C.L. (2000). Dysfunction in the neural circuitry of
emotion regulation- a possible prelude to violence. Science, 289(5479), 591594. doi:10.1126/
science.289.5479.591.
Di Giunta, L., Pastorelli, C., Eisenberg, N., Gerbino, M., Castellani, V., & Bombi, A.S. (2010).
Developmental trajectories of physical aggression: prediction of overt and covert antisocial
behaviors from self- and mothers reports. European Child & Adolescent Psychiatry, 19(12),
873882. doi:10.1007/s00787-010-0134-4.
Dionne, G., Tremblay, R.E., Boivin, M., Laplante, D., & Prusse, D. (2003). Physical aggression
and expressive vocabulary in 19month-old twins. Developmental Psychology, 39(2), 261273.
Dowlati, Y., Herrmann, N., Swardfager, W., Liu, H., Sham, L., Reim, E. K., & Lanctot, K. L.
(2010). A meta-analysis of cytokines in major depression. Biological Psychiatry, 67(5), 446
457. doi:10.1016/j.biopsych.2009.09.033.
Eckenrode, J., Campa, M., Luckey, D.W., Henderson, C.R., Cole, R., Kitzman, H., Olds, D.
(2010). Long-term effects of prenatal and infancy nurse home visitation on the life course of
youths: 19-year follow-up of a randomized trial. Archives of Pediatrics & Adolescent Medicine,
164(1), 915.
Eisenberger, N.I. (2013). An empirical review of the neural underpinnings of receiving and giving
social support: Implications for health. Psychosomatic Medicine, 75(6), 545556. doi:10.1097/
PSY.0b013e31829de2e7.
Emonds, G., Declerck, C.H., Boone, C., Seurinck, R., & Achten, R. (2014). Establishing coopera-
tion in a mixed-motive social dilemma. An fMRI study investigating the role of social value
orientation and dispositional trust. Social Neuroscience, 9(1), 1022. doi:10.1080/17470919.2
013.858080.
Febo, M., Numan, M., & Ferris, C.F. (2005). Functional magnetic resonance imaging shows oxy-
tocin activates brain regions associated with mother-pup bonding during suckling. Journal of
Neuroscience, 25(50), 1163711644. doi:10.1523/jneurosci.3604-05.2005.
Ferris, C.F., Lu, S.F., Messenger, T., Guillon, C.D., Heindel, N., Miller, M., Simon, N.G.
(2006). Orally active vasopressin V1a receptor antagonist, SRX251, selectively blocks aggres-
sive behavior. Pharmacology Biochemistry and Behavior, 83(2), 169174. doi:10.1016/j.
pbb.2006.01.001.
Fontaine, N.M. G., Lacourse, E., Vitaro, F., & Tremblay, R.E. (2014). Links between trajectories
of self-reported violent and nonviolent offending and official offending during adolescence and
adulthood. Criminal Behaviour and Mental Health, 24(4), 277290.
Frick, P.J., Lahey, B.B., Loeber, R., Tannenbaum, L., Vanhorn, Y., Christ, M.A. G., Hanson, K.
(1993). Oppositional defiant disorder and conduct disorder: A meta-analytic review of factor-
analyses and cross-validation in a clinic sample. Clinical Psychology Review, 13(4), 319340.
Frodl, T., Szyf, M., Carballedo, A., Ly, V., Dymov, S., Vaisheva, F., Booij, L. (2015). DNA
methylation of serotonin transporter gene (SLC6A4) associates with insular cortex function
involved in awareness for emotional stimuli [Online]. Journal of Psychiatry & Neuroscience.
doi:10.1503/jpn.140180.
Glover, V. (2011). Prenatal stress and the origins of psychopathology: An evolutionary perspective.
JournalofChildPsychologyandPsychiatry, 52(4),356367.doi:10.1111/j.1469-7610.2011.02371.x.
Gluckman, P.D., Hanson, M.A., Cooper, C., & Thornburg, K.L. (2008). Effect of in utero and early-
life conditions on adult health and disease. New England Journal of Medicine, 359(1), 6173.
Glueck, S., & Glueck, E. (1950). Unraveling juvenile delinquency. Cambridge: Harvard University
Press.
Goodenough, F.L. (1931). Anger in young children. Westport, CT: Greenwood Press.
Groer, M.W., & Morgan, K. (2007). Immune, health and endocrine characteristics of depressed post-
partum mothers. Psychoneuroendocrinology, 32(2), 133139. doi:10.1016/j.psyneuen.2006.11.007.
Guillemin, C., Provenal, N., Suderman, M., Ct, S. M., Vitaro, F., Hallett, M., Szyf, M.
(2014). DNA methylation signature of childhood chronic physical aggression in T cells of both
men and women. PLoS One, 9(1), 116 (e86822). doi:10.1371/journal.pone.0086822.
Guo, G., Roettger, M.E., & Shih, J.C. (2007). Contributions of the DAT1 and DRD2 genes to
serious and violent delinquency among adolescents and young adults. Human Genetics, 121(1),
125136. doi:10.1007/s00439-006-0244-8.
Hammock, E.A. D., Lim, M.M., Nair, H.P., & Young, L.J. (2005). Association of vasopressin 1a
receptor levels with a regulatory microsatellite and behavior. Genes, Brain and Behavior, 4(5),
289301. doi:10.1111/j.1601-183X.2005.00119.x.
Hay, D.F., Mundy, L., Roberts, S., Carta, R., Waters, C.S., Perra, O., van Goozen, S. (2011).
Known risk factors for violence predict 12-month-old infants aggressiveness with peers.
Psychological Science, 22(9), 12051211. doi:10.1177/0956797611419303.
Healy, W., & Bronner, A. F. (1926). Delinquents and criminals: Their making and unmaking.
NewYork, NY: Macmillan.
Heckman, J., Pinto, R., & Savelyev, P. (2013). Understanding the mechanisms through which an
influential early childhood program boosted adult outcomes. American Economic Review,
103(6), 20522086. doi:10.1257/aer.103.6.2052.
Hoge, E.A., Brandstetter, K., Moshier, S., Pollack, M.H., Wong, K.K., & Simon, N.M. (2009).
Broad spectrum of cytokine abnormalities in panic disorder and posttraumatic stress disorder.
Depression and Anxiety, 26(5), 447455. doi:10.1002/da.20564.
Insel, T.R. (2010). The challenge of translation in social neuroscience: A review of oxytocin, vaso-
pressin, and affiliative behavior. Neuron, 65(6), 768779. doi:10.1016/j.neuron.2010.03.005.
Janelidze, S., Mattei, D., Westrin, A., Trskman-Bendz, L., & Brundin, L. (2010). Cytokine levels
in the blood may distinguish suicide attempters from depressed patients. Brain, Behavior, and
Immunity, 25(2), 335339. doi:10.1016/j.bbi.2010.10.010.
Kawamura, A., Yoshikawa, T., Takahashi, T., Hayashi, T., Takahashi, E., Ogawa, S. (2001).
Randomized two of phosphodiesterase inhibitors versus catecholamines in patients with
acutely decompensated heart failure. Japanese Circulation Journal-English Edition, 65(10),
858862. doi:10.1253/jcj.65.858.
Keck, M.E., Wigger, A., Welt, T., Muller, M.B., Gesing, A., Reul, J., Neumann, I.D. (2002).
Vasopressin mediates the response of the combined dexamethasone/CRH test in hyper-anxious
rats: Implications for pathogenesis of affective disorders. Neuropsychopharmacology, 26(1),
94105. doi:10.1016/s0893-133x(01)00351-7.
Kieseppa, T., Partonen, T., Haukka, J., Kaprio, J., & Lonnqvist, J.(2004). High concordance of
bipolar I disorder in a nationwide sample of twins. American Journal of Psychiatry, 161(10),
18141821. doi:10.1176/appi.ajp.161.10.1814.
Koo, J.W., & Duman, R.S. (2008). IL-1 beta is an essential mediator of the antineurogenic and
anhedonic effects of stress. Proceedings of the National Academy of Sciences of the United
States of America, 105(2), 751756. doi:10.1073/pnas.0708092105.
Krug, E.G., Dahlberg, L.L., Mercy, J.A., Zwi, A.B., & Lozano, R. (2002). World report on vio-
lence and health, 2002. Accessed from www.who.int/violence_injury_prevention/violence/
world_report/wrvh1/en
Kuo, M.H., & Allis, C.D. (1998). Roles of histone acetyltransferases and deacetylases in gene
regulation. Bioessays, 20(8), 615626. doi:10.1002/(sici)1521-1878(199808)20:8<615::aid-
bies4>3.0.co;2-h.
Lacourse, E., Boivin, M., Brendgen, M., Petitclerc, A., Girard, A., Vitaro, F., Tremblay, R.E.
(2014). A longitudinal twin study of physical aggression in early childhood: Evidence for a
developmentally dynamic genome. Psychological Medicine, 44(12), 26172627. doi:10.1017/
S0033291713003218.
Lacourse, E., Ct, S., Nagin, D. S., Vitaro, F., Brendgen, M., & Tremblay, R. E. (2002). A
longitudinal-experimental approach to testing theories of antisocial behavior development.
Development and Psychopathology, 14(4), 909924. doi:10.1017/S0954579402004121.
Lahey, B.B. (2015). Why are children who exhibit psychopathology at high risk for psychopathology
and dysfunction in adulthood? [Online]. JAMA Psychiatry. doi:10.1001/jamapsychiatry.2015.0798.
Larsson, L., Thorbert-Mros, S., Rymo, L., & Berglundh, T. (2012). Influence of epigenetic modi-
fications of the interleukin-10 promoter on IL10 gene expression. European Journal of Oral
Sciences, 120(1), 1420. doi:10.1111/j.1600-0722.2011.00917.x.
Laucht, M., Brandeis, D., & Zohsel, K. (2014). Gene-environment interactions in the etiology of
human violence. In K.A. Miczek, & A.Meyer-Lindenberg (Eds.), Neuroscience of aggression
(pp.267295). Heidelberg, Germany: Springer.
Lehn, H., Derks, E.M., Hudziak, J.J., Heutink, P., van Beijsterveldt, T., & Boomsma, D.I. (2007).
Attention problems and attention-deficit/hyperactivity disorder in discordant and concordant
monozygotic twins: Evidence of environmental mediators. Journal of the American Academy
of Child and Adolescent Psychiatry, 46(1), 8391. doi:10.1097/01.chi.0000242244.00174.d9.
Lewi, L., Cannie, M., Blickstein, I., Jani, J., Huber, A., Hecher, K., Deprest, J.(2007). Placental
sharing, birthweight discordance, and vascular anastomoses in monochorionic diamniotic twin
placentas. American Journal of Obstetrics and Gynecology, 197(6), 587.e581587.e588.
doi:10.1016/j.ajog.2007.05.009.
Loeber, R., & Hay, D.F. (1997). Key issues in the development of aggression and violence from
childhood to early adulthood. Annual Review of Psychology, 48, 371410. doi:10.1146/
annurev.psych.48.1.371.
Loeber, R., & Stouthamer-Loeber, M. (1998). Development of juvenile aggression and violence.
Some common misconceptions and controversies. American Psychologist, 53(2), 242259.
Loney, B. R., Butler, M. A., Lima, E. N., Counts, C. A., & Eckel, L. A. (2006). The relation
between salivary cortisol, callous-unemotional traits, and conduct problems in an adolescent
non-referred sample. Journal of Child Psychology and Psychiatry, 47(1), 3036.
doi:10.1111/j.1469-7610.2005.01444.x.
Lutsiv, O., Pullenayegum, E., Foster, G., Vera, C., Giglia, L., Chapman, B., McDonald, S. D.
(2013). Womens intentions to breastfeed: a population-based cohort study. BJOG: An International
Journal of Obstetrics and Gynaecology, 120(12), 14901499. doi:10.1111/1471-0528.12376.
MacDonald, K., & MacDonald, T.M. (2010). The peptide that binds: A systematic review of oxy-
tocin and its prosocial effects in humans. Harvard Review of Psychiatry, 18(1), 121.
doi:10.3109/10673220903523615.
Marquez, C., Poirier, G.L., Cordero, M.I., Larsen, M.H., Groner, A., Marquis, J., Sandi, C.
(2013). Peripuberty stress leads to abnormal aggression, altered amygdala and orbitofrontal
reactivity and increased prefrontal MAOA gene expression. Translational Psychiatry, 3, e216.
doi:10.1038/tp.2012.144.
Marsland, A.L., Prather, A.A., Petersen, K.L., Cohen, S., & Manuck, S.B. (2008). Antagonistic
characteristics are positively associated with inflammatory markers independently of trait nega-
tive emotionality. Brain, Behavior, and Immunity, 22(5), 753761. doi:10.1016/j.bbi.2007.11.008.
McDonald, S.D., Pullenayegum, E., Chapman, B., Vera, C., Giglia, L., Fusch, C., & Foster, G.
(2012). Prevalence and predictors of exclusive breastfeeding at hospital discharge. Obstetrics
and Gynecology, 119(6), 11711179. doi:10.1097/AOG.0b013e318256194b.
Meaney, M.J. (2010). Epigenetics and the biological definition of gene x environment interactions.
Child Development, 81(1), 4179.
Mehta, D., Klengel, T., Conneely, K.N., Smith, A.K., Altmann, A., Pace, T.W., Binder, E.B.
(2013). Childhood maltreatment is associated with distinct genomic and epigenetic profiles in
posttraumatic stress disorder. Proceedings of the National Academy of Sciences of the United
States of America, 110(20), 83028307. doi:10.1073/pnas.1217750110.
Miller, G.E., Chen, E., Fok, A.K., Walker, H., Lim, A., Nicholls, E.F., Kobor, M.S. (2009).
Low early-life social class leaves a biological residue manifested by decreased glucocorticoid
and increased proinflammatory signaling. Proceedings of the National Academy of Sciences of
the United States of America, 106(34), 1471614721. doi:10.1073/pnas.0902971106.
Moffitt, T.E., & Scott, S. (2008). Conduct disorders of childhood and adolescence. In M.Rutter,
D.Bishop, D.Pine, S.Scott, J.Stevenson, E.Taylor, & A.Thapar (Eds.), Rutters child and
adolescent psychiatry (5th ed., pp.543564). Oxford: Blackwell.
Murni, E., Ponsuksili, S., DEath, R. B., Turner, S. P., Kurt, E., Evans, G., Wimmers, K.
(2010). Association of HPA axis-related genetic variation with stress reactivity and aggressive
behaviour in pigs. BMC Genetics, 11(74), 111. doi:10.1186/1471-2156-11-74.
Naerde, A., Ogden, T., Janson, H., & Zachrisson, H.D. (2014). Normative development of physical
aggression from 8 to 26months. Developmental Psychology, 50(6), 17101720. doi:10.1037/
a0036324.
Nagin, D., & Tremblay, R.E. (1999). Trajectories of boys physical aggression, opposition, and
hyperactivity on the path to physically violent and nonviolent juvenile delinquency. Child
Development, 70(5), 11811196.
Nelson, R. J., & Trainor, B. C. (2007). Neural mechanisms of aggression. Nature Reviews
Neuroscience, 8(7), 536546.
Early Child Care Research Network, N.I. C.H. D. (2004). Trajectories of physical aggression
from toddlerhood to middle school: predictors, correlates, and outcomes. SRCD Monographs,
69(4, 278), 1146.
OBrien, S.M., Scott, L.V., & Dinan, T.G. (2004). Cytokines: Abnormalities in major depression
and implications for pharmacological treatment. Human Psychopharmacology: Clinical and
Experimental, 19(6), 397403. doi:10.1002/hup.609.
OConnor, T.G., Moynihan, J.A., & Caserta, M.T. (2014). Annual Research Review: The neuro-
inflammation hypothesis for stress and psychopathology in children- developmental psycho-
neuroimmunology. Journal of Child Psychology and Psychiatry, 55(6), 615631. doi:10.1111/
jcpp.12187.
Odgers, C.L., Moffitt, T.E., Broadbent, J.M., Dickson, N., Hancox, R.J., Harrington, H., Caspi,
A. (2008). Female and male antisocial trajectories: From childhood origins to adult outcomes.
Development and Psychopathology, 20(2), 673716. doi:10.1017/s0954579408000333.
Olds, D., Henderson, C.R., Cole, R., Eckenrode, J., Kitzman, H., Luckey, D., Powers, J.(1998).
Long-term effects of nurse home visitation on children's criminal and antisocial behavior:
15-year follow-up of a randomized controlled trial. Journal of the American Medical
Association, 280(14), 12381244.
Ostlund, A., Nordstrm, M., Dykes, F., & Flacking, R. (2010). Breastfeeding in preterm and term
twins-maternal factors associated with early cessation: A population-based study. Journal of
Human Lactation, 26(3), 235241. doi:10.1177/0890334409359627.
Pace, T.W. W., Hu, F., & Miller, A.H. (2007). Cytokine-effects on glucocorticoid receptor func-
tion: Relevance to glucocorticoid resistance and the pathophysiology and treatment of major
depression. Brain, Behavior, and Immunity, 21(1), 919. doi:10.1016/j.bbi.2006.08.009.
Panksepp, J. (1998). Affective neuroscience: The foundations of human and animal emotions.
NewYork, NY: Oxford University Press.
Pesce, M., Speranza, L., Franceschelli, S., Ialenti, V., Patruno, A., Febo, M.A., Grilli, A. (2011).
Biological role of Interleukin-1 beta in defensive aggressive behavior. Journal of Biological
Regulators and Homeostatic Agents, 25(3), 323329.
Pingault, J.-B., Ct, S. M., Lacourse, E., Galra, C., Vitaro, F., & Tremblay, R. E. (2013).
Childhood hyperactivity, physical aggression and criminality: A 19year prospective population-
based study. PLoS One, 8(5), 17.
Popma, A., Vermeiren, R., Geluk, C., Rinne, T., van den Brink, W., Knol, D.L., Doreleijers, T.A.
H. (2007). Cortisol moderates the relationship between testosterone and aggression in delinquent
male adolescents. Biological Psychiatry, 61(3), 405411. doi:10.1016/j.biopsych.2006.06.006.
Powell, N.D., Sloan, E.K., Bailey, M.T., Arevalo, J.M. G., Miller, G.E., Chen, E., Cole, S.W.
(2013). Social stress up-regulates inflammatory gene expression in the leukocyte transcriptome
via beta-adrenergic induction of myelopoiesis. Proceedings of the National Academy of
Provenal, N., Suderman, M.J., Caramaschi, D., Wang, D.S., Hallett, M., Vitaro, F., Szyf, M.
(2013). Differential DNA methylation regions in cytokine and transcription factor genomic loci
associate with childhood physical aggression. PLoS One, 8(8), 119 (e71691). doi:10.1371/
journal.pone.0071691.
Provenal, N., Suderman, M. J., Guillemin, C., Massart, R., Ruggiero, A., Wang, D., Szyf, M.
(2012). The signature of maternal rearing in the methylome in Rhesus Macaque prefrontal cortex
andT cells. Journal of Neuroscience, 32(44), 1562615642. doi:10.1523/JNEUROSCI.1470-12.2012.
Provenal, N., Suderman, M.J., Guillemin, C., Vitaro, F., Ct, S.M., Hallett, M., Szyf, M.
(2014). Association of childhood chronic physical aggression with a DNA methylation signa-
ture in adult human T cells. PLoS One, 9(4), 110 (e89839). doi:10.1371/journal.pone.0089839.
Quetelet, A. (1833). Research on the propensity for crime at different ages. Bruxels: M.Hayez [Quetelet,
A. (1833). Recherches sur le penchant au crime aux diffrents ges. Bruxelles: M.Hayez].
Raine, A. (2013). The anatomy of violence: The biological roots of crime. NewYork, NY: Pantheon.
Raine, A., Liu, J.H., Venables, P., & Mednick, S.A. (2006). Preventing crime and schizophrenia
using early environmental enrichment. In A.Raine (Ed.), Crime and Schizophrenia: Causes
and cures (pp.249265). Hauppauge, NY: Nova Science Publishers, Inc..
Raine, A., Mellingen, K., Liu, J.H., Venables, P., & Mednick, S.A. (2003). Effects of environmen-
tal enrichment at ages 3-5years on schizotypal personality and antisocial behavior at ages 17
and 23years. American Journal of Psychiatry, 160(9), 16271635.
Raine, A., Venables, P.H., Dalais, C., Mellingen, K., Reynolds, C., & Mednick, S.A. (2001). Early
educational and health enrichment at age 3-5years is associated with increased autonomic and
central nervous system arousal and orienting at age 11 years: Evidence from the Mauritius
Child Health Project. Psychophysiology, 38(2), 254266.
Rameson, L.T., Morelli, S.A., & Lieberman, M.D. (2012). The neural correlates of empathy:
Experience, automaticity, and prosocial behavior. Journal of Cognitive Neuroscience, 24(1),
235245.
Reiss, A.J., & Roth, J.A. (Eds.) (1993). Understanding and preventing violence. Washington, DC:
National Academy Press.
Rizzi, T.S., van der Sluis, S., Derom, C., Thiery, E., van Kesteren, R.E., Jacobs, N., Posthuma,
D. (2013). FADS2 genetic variance in combination with fatty acid intake might alter composi-
tion of the fatty acids in brain. PLoS One, 8(6), 110 (e68000). doi:10.1371/journal.
pone.0068000.
Robins, L.N. (1966). Deviant children grown up. Baltimore, MD: Williams & Wilkins.
Robins, L.N. (1992). The role of prevention experiments in discovering causes of childrens anti-
social behavior. In J. McCord, & R. E. Tremblay (Eds.), Preventing antisocial behavior:
Interventions from birth to adolescence (pp.318). NewYork, NY: The Guilford Press.
Robinson, M., Oddy, W.H., Li, J.H., Kendall, G.E., de Klerk, N.H., Silburn, S.R., Mattes, E.
(2008). Pre- and postnatal influences on preschool mental health: a large-scale cohort study. Journal
of Child Psychology and Psychiatry, 49(10), 11181128. doi:10.1111/j.1469-7610.2008.01955.x.
Romano, E., Tremblay, R.E., Farhat, A., & Ct, S. (2006). Development and prediction of hyper-
active symptoms from 2 to 7years in a population-based sample. Pediatrics, 117(6), 2101
2110. doi:10.1542/peds.2005-0651.
Sanchez, C., Arnt, J., Hyttel, J., & Moltzen, E.K. (1993). The role of serotonergic mechanisms in
inhibition of isolation-induced aggression in male mice. Psychopharmacology, 110(12),
5359. doi:10.1007/bf02246950.
Schweinhart, L., Montie, J., Xiang, Z., Barnett, W.S., Belfield, C.R., & Nores, M. (2005). Lifetime
effects: The High/Scope Perry Preschool study through age 40 (vol. 14). Ypsilanti, MI: High/
Scope Press.
Shirtcliff, E.A., Granger, D.A., Booth, A., & Johnson, D. (2005). Low salivary cortisol levels and
externalizing behavior problems in youth. Development and Psychopathology, 17(1), 167184.
doi:10.1017/s0954579405050091.
Sidora-Arcoleo, K., Anson, E., Lorber, M., Cole, R., Olds, D., & Kitzman, H. (2010). Differential
effects of a nurse home-visiting intervention on physically aggressive behavior in children.
Journal of Pediatric Nursing: Nursing Care of Children & Families, 25(1), 3545. doi:10.1016/j.
pedn.2008.07.011.
Sloan, E.K., Capitanio, J.P., Tarara, R.P., Mendoza, S.P., Mason, W.A., & Cole, S.W. (2007).
Social stress enhances sympathetic innervation of primate lymph nodes: Mechanisms and
implications for viral pathogenesis. Journal of Neuroscience, 27(33), 88578865. doi:10.1523/
jneurosci.1247-07.2007.
Smith, A.K., Conneely, K.N., Kilaru, V., Mercer, K.B., Weiss, T.E., Bradley, B., Ressler, K.J.
(2011). Differential immune system DNA methylation and cytokine regulation in post-
traumatic stress disorder. American Journal of Medical Genetics Part B: Neuropsychiatric
Genetics, 156B(6), 700708. doi:10.1002/ajmg.b.31212.
Smith, S.M., Zwart, S.R., Heer, M., Hudson, E.K., Shackelford, L., & Morgan, J.L. L. (2014).
Men and women in space: Bone loss and kidney stone risk after long-duration spaceflight.
Journal of Bone and Mineral Research, 29(7), 16391645. doi:10.1002/jbmr.2185.
Suarez, E.C., Lewis, J.G., & Kuhn, C. (2002). The relation of aggression, hostility, and anger to
lipopolysaccharide-stimulated tumor necrosis factor (TNF)-alpha by blood monocytes from nor-
mal men. Brain, Behavior, and Immunity, 16(6), 675684. doi:10.1016/s0889-1591(02)00019-3.
Suderman, M., Borghol, N., Pappas, J.J., Pereira, S.M. P., Pembrey, M., Hertzman, C., Szyf,
M. (2014). Childhood abuse is associated with methylation of multiple loci in adult DNA. BMC
Medical Genomics, 7, 13. doi:10.1186/1755-8794-7-13.
Tarullo, A. R., & Gunnar, M. R. (2006). Child maltreatment and the developing HPA axis.
Hormones and Behavior, 50(4), 632639. doi:10.1016/j.yhbeh.2006.06.010.
Tharner, A., Luijk, M., Raat, H., Ijzendoorn, M.H., Bakermans-Kranenburg, M.J., Moll, H.A.,
Tiemeier, H. (2012). Breastfeeding and Its Relation to Maternal Sensitivity and Infant
Attachment. Journal of Developmental and Behavioral Pediatrics, 33(5), 396404. doi:10.1097/
DBP.0b013e318257fac3.
Tremblay, R. E. (2000). The origins of youth violence (ISUMA). Canadian Journal of Policy
Research, 1(2), 1924.
Tremblay, R. E. (2003). Public opinion and violence prevention. Accessed from http://www.
excellence-jeunesenfants.ca/documents/BulletinVol2No1Avril03ANG.pdf
Tremblay, R.E. (2006). Prevention of youth violence: Why not start at the beginning? Journal of
Abnormal Child Psychology, 34(4), 481487.
Tremblay, R.E. (2010). Developmental origins of disruptive behaviour problems: The original
sin hypothesis, epigenetics and their consequences for prevention. Journal of Child Psychology
and Psychiatry, 51(4), 341367. doi:10.1111/j.1469-7610.2010.02211.x.
Tremblay, R.E., Japel, C., Prusse, D., McDuff, P., Boivin, M., Zoccolillo, M., & Montplaisir,
J. (1999). The search for the age of onset of physical aggression: Rousseau and Bandura
revisited. Criminal Behavior and Mental Health, 9(1), 823.
Tremblay, R.E., Nagin, D., Sguin, J.R., Zoccolillo, M., Zelazo, P.D., Boivin, M., Japel, C.
(2004). Physical aggression during early childhood: Trajectories and predictors. Pediatrics,
114(1), e43e50.
Tremblay, R.E., Pagani-Kurtz, L., Msse, L.C., Vitaro, F., & Pihl, R.O. (1995). A bimodal preven-
tive intervention for disruptive kindergarten boys: Its impact through mid-adolescence. Journal
of Consulting and Clinical Psychology, 63(4), 560568. doi:10.1037//0022-006X.63.4.560.
Tsankova, N., Renthal, W., Kumar, A., & Nestler, E.J. (2007). Epigenetic regulation in psychiatric
disorders. Nature Reviews Neuroscience, 8(5), 355367. doi:10.1038/nrn2132.
Uddin, M., Aiello, A.E., Wildman, D.E., Koenen, K.C., Pawelec, G., de los Santos, R., Galea,
S. (2010). Epigenetic and immune function profiles associated with posttraumatic stress disor-
der. Proceedings of the National Academy of Sciences of the United States of America, 107(20),
94709475. doi:10.1073/pnas.0910794107.
Vage, J., Wade, C., Biagi, T., Fatjo, J., Amat, M., Lindblad-Toh, K., & Lingaas, F. (2010).
Association of dopamine- and serotonin-related genes with canine aggression. Genes, Brain
and Behavior, 9(4), 372378. doi:10.1111/j.1601-183X.2010.00568.x.
van Bokhoven, I., Van Goozen, S.H. M., van Engeland, H., Schaal, B., Arseneault, L., Sguin,
J.R., Tremblay, R.E. (2005). Salivary cortisol and aggression in a population-based longi-
tudinal study of adolescent males. Journal of Neural Transmission, 112(8), 10831096.
doi:10.1007/s00702-004-0253-5.
Vaughn, M.G., Delisi, M., Beaver, K.M., & Wright, J.P. (2009). DAT1 and 5HTT are associated
with pathological criminal behavior in a nationally representative sample of youth. Criminal
Justice and Behavior, 36(11), 11131124. doi:10.1177/0093854809342839.
Venables, P.H., Raine, A., Mednick, S.A., Schulsinger, F., & Dalais, C. (2006). The Mauritius
Child Health Project: Its origins, procedures, ethical issues, and outcome. In A.Raine (Ed.),
Crime and schizophrenia: Causes and cures (pp. 315334). New York: Hauppauge, Nova
Science Publishers.
Vitaro, F., Barker, E.D., Brendgen, M., & Tremblay, R.E. (2012). Pathways explaining the reduc-
tion of adult criminal behaviour by a randomized preventive intervention for disruptive kin-
dergarten children. Journal of Child Psychology and Psychiatry, 53(7), 748756.
doi:10.1111/j.1469-7610.2011.02517.x.
von Kanel, R., Hepp, U., Kraemer, B., Traber, R., Keel, M., Mica, L., & Schnyder, U. (2007). Evidence
for low-grade systemic proinflammatory activity in patients with posttraumatic stress disorder.
Journal of Psychiatric Research, 41(9), 744752. doi:10.1016/j.jpsychires.2006.06.009.
Wakschlag, L.S., Kistner, E.O., Pine, D.S., Biesecker, G., Pickett, K.E., Skol, A.D., Cook,
E.H. (2010). Interaction of prenatal exposure to cigarettes and MAOA genotype in pathways
to youth antisocial behavior. Molecular Psychiatry, 15(9), 928937. doi:10.1038/mp.2009.22.
Wang, D., Szyf, M., Benkelfat, C., Provenal, N., Caramaschi, D., Ct, S.M., Booij, L. (2012).
Peripheral SLC6A4 DNA methylation is associated with in vivo measures of human brain
serotonin synthesis and childhood physical aggression. PLoS One, 7(6), 18 (e39501).
doi:10.1371/journal.pone.0039501.
Weaver, I.C. G., Cervoni, N., Champagne, F.A., DAlessio, A.C., Sharma, S., Seckl, J.R.,
Meaney, M.J. (2004). Epigenetic programming by maternal behavior. Nature Neuroscience,
7(8), 847854. doi:10.1038/nn1276.
Williamson, J.S., Harris, D.A., Beynon, J., & Jenkins, G.J. S. (2015). Review of the development
of DNA methylation as a marker of response to neoadjuvant therapy and outcomes in rectal
cancer. Clinical Epigenetics, 7(1), 70. doi:10.1186/s13148-015-0111-3.
Wilson, S.J., & Lipsey, M.W. (2007). School-based interventions for aggressive and disruptive
behavior: Update of a meta-analysis. American Journal of Preventive Medicine, 33(2),
S130S143.
Zwart, S.R., Morgan, J.L. L., & Smith, S.M. (2013). Iron status and its relations with oxidative
damage and bone loss during long-duration space flight on the International Space Station.
American Journal of Clinical Nutrition, 98(1), 217223. doi:10.3945/ajcn.112.056465.
Part III
Intervention Research: Implications for
Gene Environment Transaction
Framework
Chapter 11
Using Genetically Informed Prevention Trials
toTest GeneEnvironment Hypotheses
GeneH.Brody
Introduction
Historically, residing in rural communities in the southern United States has pro-
tected African-American preadolescents and adolescents from the use of alcohol and
other drugs. Recent epidemiological data, however, indicate that African-American
youths in rural areas use drugs at rates equal to or exceeding those of youths in
densely populated inner cities (National Center on Addiction and Substance Abuse,
2000). Escalation of drug use has prognostic significance for rural African-American
youths educational and occupational opportunities and attainment, involvement
with the criminal justice system, and physical health (Centers for Disease Control
and Prevention, 1999). Prevention programs designed to deter drug use are scarce in
the rural South. These circumstances and the resulting need for prevention program-
ming led to the development of the Strong African American Families (SAAF) pro-
gram for preadolescents, the Strong African American FamiliesTeen (SAAFT)
program for adolescents, and the Adults in the Making (AIM) program for young
adults. These family-centered prevention programs were developed, implemented,
and tested in longitudinal, randomized efficacy trials. As part of the programs evalu-
ations, candidate genes that have been implicated in drug use and dependence pat-
terns were genotyped from saliva samples.
This chapter describes (a) the SAAF, SAAFT, and AIM programs, along with
their efficacy; (b) the ways in which these prevention programs were used to test
G.H. Brody (*)

Center for Family Research, University of Georgia,
1095 College Station Road, Athens, GA 30602-4527, USA
e-mail: gbrody@uga.edu

DOI10.1007/978-3-319-49227-8_11
212 G.H. Brody
geneenvironment (GE) hypotheses; (c) the way forward with geneintervention

research; and (d) caveats about the use of genetic data to select participants for pre-
ventive interventions.
ThePrevention Programs
On the basis of prior research demonstrating the importance of protective caregiving

in fostering competence and resilience in rural African-American youths (Brody
etal., 2004), culturally and ecologically valid family skills training programs were
developed. This approach was selected based on several considerations. It was par-
ticularly attractive to the rural families who participated in focus groups. Evidence
also suggests that the integration of youth, caregiver, and family curricula may pro-
duce better outcomes than does targeting only youths or only caregivers in preven-
tion programming (Foxcroft, Ireland, Lister-Sharp, Lowe, & Breen, 2003; Spoth,
Redmond, Trudeau, & Shin, 2002). The family skills format has also outperformed
parent-only interventions in facilitating protective caregiving (Spoth, Redmond, &
Shin, 1998). This model also permits program implementation by facilitators who,
although trained to conduct the intervention, do not have postsecondary education or
clinical certification. Few African-Americans in rural areas have these credentials.
Structure of the Family Interventions. These programs are designed to accommodate
groups of 813 families. Single-parent, two-parent, and foster-parent families are all
welcome to participate. Single parents are invited to ask an extended family member
who assists in rearing the child (e.g., grandparent, aunt, or uncle) to attend (Molgaard
& Spoth, 2001). Caregivers and youths attend a structured program that is composed
of weekly 2-h meetings: seven meetings for SAAF, six meetings for AIM, and five
meetings for SAAFT.In the first hour, caregivers and youths meet separately; in the
second hour, families practice the skills they learned in their separate sessions in a
joint caregiveryouth session. Youth and caregiver sessions have parallel content dur-
ing most meetings, with the family session providing reinforcement and skills prac-
tice. Because the family skills approach is highly structured, a detailed curriculum
manual is used and all activities are timed. Program content for the caregivers ses-
sions is delivered by narrators on DVDs that also depict family interactions illustrat-
ing targeted behaviors. DVDs are used in some youth sessions, showing
age-appropriate models discussing typical high-risk situations and ways of dealing
with temptation and peer pressure issues. Structured activities, games, role-playing,
and group discussions are also part of the youth sessions. Group leaders facilitate
activities, guide group discussions, organize role-playing and other interactive activi-
ties, and answer participants questions. Leaders are provided with materials designed
to help them correctly execute the session protocol; these materials include an outline
of each session, a checklist of the materials necessary for each activity, the specific
theme of each task, and forms for in-session notes. This carefully structured approach
standardizes much of the content and facilitates fidelity in implementation.
11 Using Genetically Informed Prevention Trials to Test Gene Environment 213
AnOverview oftheSAAF Randomized Trial
Participants in the original SAAF randomized prevention trial were African-

American youths (mean age at pretest=11.2years) and their primary caregivers,
who resided in rural Georgia. Families in this area live in small towns and commu-
nities in which poverty rates are among the highest in the nation and unemployment
rates are above the national average (Proctor & Dalaker, 2003). The intervention
group included 369 families, and the control group included 298 families. Families
were oversampled into the intervention to ensure that at least 350 would take part in
SAAF.From lists that schools provided, 11-year-old African-Americans residing in
nine rural counties were randomly selected. Their families were contacted and
invited to participate; the recruitment rate of 64% exceeds rates commonly reported
for prevention trials. Recruitment procedures and the conceptual underpinnings of
SAAF are described in detail in Brody etal. (2004). To preserve the random nature
of the group assignments, the analyses included all families in the intervention con-
dition who completed all assessments regardless of the number of prevention ses-
sions they actually attended, a procedure called an intent-to-treat analysis. At each
data collection point, one home visit lasting 2 h was made to each family. Field
researchers administered self-report questionnaires to caregivers and youths in an
interview format.
Caregivers in the prevention condition were taught involvedvigilant parenting,
which includes the consistent use of warm and communicative parenting practices
along with high levels of monitoring and control, adaptive racial socialization
strategies, strategies for communication about sex, and the establishment of clear
expectations about alcohol use. Program content for the youth sessions focused on
the importance of caring family relationships and compliance with household
rules, peer pressure and resistance efficacy strategies, and development of a posi-
tive racial identity.
Testing the Efficacy and Validity of SAAF and Its Theoretical Underpinnings. The
SAAF causative model included the hypotheses that (a) participation in the inter-
vention would increase involvedvigilant parenting and youth intrapersonal pro-
tective processes (e.g., self-control, future orientation, and racial identity) and (b)
intervention-induced changes in youth intrapersonal processes would be mediated
through the interventions effects on involvedvigilant parenting. As hypothesized,
parents in SAAF reported greater changes from pretest to posttest in involved
vigilant parenting than did control parents (=.33, p<.01), and youths in SAAF
demonstrated greater changes in protective self-regulatory processes than did con-
trol youths (=.23, p<.01). Of particular interest was the finding that, in control
families, both parent and youth protective processes declined, whereas the same
processes increased in families randomly assigned to participate in SAAF.Thus,
across the relatively short period of 7 months that separated the pretest and post-
test, protective processes among control families began to wane during a develop-
mental period when youths begin to spend more time away from home and with
peers. A hypothesis was tested that changes in involvedvigilant parenting would
214 G.H. Brody
mediate the effect of treatment condition on changes in youth protective factors.

The analyses met all of Baron and Kennys (1986) criteria for mediation: (a) group
assignment was significantly associated with increases in involvedvigilant parent-
ing after controlling for pretest levels; (b) group assignment was significantly asso-
ciated with increases in youth protective factors after controlling for pretest levels;
(c) changes in involvedvigilant parenting were significantly associated with
changes in youth protective factors; and (d) the impact of assignment to SAAF or
the control group on youth protective factors was not significant in the presence of
the indicator, involvedvigilant parenting.
In addition to evaluating effects on alcohol use, we ran a test to determine
whether SAAF-induced increases in youth self-regulatory processes from the pre-
test to posttest assessments would mediate the influence of SAAF on alcohol use at
the first long-term follow-up, 2 years after participating in SAAF (Brody et al.,
2006). Our results revealed that, compared with adolescents in the control condi-
tion, fewer SAAF participants initiated alcohol use, and those who did increased
their use at a slower rate over time. The analysis also confirmed the proposed medi-
ational hypothesis: Youth self-regulation at posttest, with pretest levels controlled,
was associated with decreases from pretest to long-term follow-up in youth alcohol
use, and the impact of group assignment was not significant in the presence of
SAAF-induced increases from pretest to posttest in youth protective factors.
The second analysis addressed SAAFs continuing efficacy more than 5years
after program participation (Brody, Chen, Kogan, Murry, & Brown, 2010). Prior to
this analysis, the literature had not addressed the long-term efficacy of family-
centered preventive interventions designed to deter alcohol use among African-
American youths in general or those living in rural areas in particular. The results
showed that, 65 months after the pretest, youths who participated in SAAF reported
drinking half as often during the previous month than did those in the control
condition.
AnOverview oftheSAAFT Randomized Trial
SAAFT builds on other family-centered intervention programsparticularly

SAAFthat have been found to enhance parent and youth competence while inhib-
iting substance use, delinquent activity, and other co-occurring problems among
youths in elementary school and middle school (Brody et al., 2004; Kumpfer,
Alvarado, & Whiteside, 2003; Leigh & Stahl, 1993). For broad public health impact,
however, family-centered prevention programs must be available for youths at all
developmental stages. Unfortunately, no family-centered prevention programs for
rural African-American adolescents had been developed and tested despite epide-
miologic research that highlights the emergence and escalation of substance use,
conduct problems, and depressive symptoms around the time of high school entry
(Brody, Chen, & Kogan, 2010; OConnell, Boat, & Warner, 2009). The development
of these problems could be addressed with timely intervention. The development and
evaluation of SAAFT was designed to meet the need for family-centered prevention
programs for adolescents in general and for African-American adolescents in
particular.
Participants in the trial included 502 African-American families in rural Georgia;
in each family, an adolescent who was 16years of age at recruitment (51% girls)
and the adolescents primary caregiver (in most cases the biological mother) pro-
vided data. Mean household monthly gross income was $1482.50. Although the
caregivers worked an average of 41.5h per week (SD=20.36), 63.8% of the fami-
lies lived below federal poverty standards and another 18% lived within 150% of
the poverty threshold; they could be described as working poor (Boatright, 2005).
Recruitment procedures and SAAFT conceptual underpinnings are described in
Brody, Chen, Kogan etal. (2012). Of the families recruited, 252 were assigned ran-
domly to SAAFT and 250 to an attention control group. Of the families who pro-
vided data at pretest, 478 (95 %) provided data 22 months later at long-term
follow-up: 237in the intervention group and 241in the control group. The use of an
attention control group is unique in evaluations of the efficacy of family-centered
interventions. Typically, control groups in such evaluations receive either no treat-
ment or minimal information. These designs do not control for nonspecific factors,
such as social support from intervention trainers and other group members, which
arguably could be responsible for observed intervention effects. To provide a more
stringent efficacy evaluation, all families, whether assigned to SAAFT or the atten-
tion control condition, attended a five-session, 10-h group prevention program held
at community facilities.
The SAAFT program followed the structure of the SAAF program, including
separate caregiver and adolescent skill-building curricula and a family curriculum
presented during 1-h concurrent training sessions and a 1-h family session. During
the implementation of SAAFT, the attention control group participated in a five-
session family-centered intervention structurally similar to SAAFT designed to
promote healthful behaviors among adolescents by encouraging good nutrition,
exercise, and informed consumer behavior.
In SAAFT, caregivers were taught developmentally appropriate use of consis-
tent monitoring and control, adaptive racial socialization approaches that included
guidance for dealing with discrimination, establishment of clear norms and expecta-
tions about adolescent substance use, provision of academic support, and coopera-
tive caregiveradolescent problem-solving. Adolescents were taught the importance
of having and following household rules, strategies to use when encountering rac-
ism, the importance of academic success, goal formation, and strategies for attain-
ing educational and occupational goals. To preserve the random nature of the group
assignments, data analyses included all families regardless of the number of pro-
gram sessions they attended, an intent-to-treat analysis. Both SAAFT and control
families attended an average of four of the five program sessions.
The study hypothesis proposed that adolescents in the SAAFT condition
would evince fewer conduct problems, less frequent substance use, fewer sub-
stance use problems, and fewer depressive symptoms than would those in the
attention control condition across the 22 months between the pretest and long-term
216 G.H. Brody
follow-up assessments. The analyses for changes in the frequencies of conduct

problems, substance use, and substance use problems revealed consistent results.
SAAFT participation caused significant and robust prevention effects (all
ps<.001), such that participation in SAAFT, compared with participation in the
attention control program, was associated with a 36% decrease in the frequency of
conduct problems, a 32% decrease in substance use, and a 47% decrease in sub-
stance use problems. SAAFT participants also experienced fewer depressive
symptoms over time than did attention control participants (p < .01). SAAFT
participants evinced a 4.5% decrease in depressive symptoms relative to partici-
pants in the attention control program.
AnOverview oftheAIM Randomized Trial
The transition from adolescence to emerging adulthood involves pervasive contex-

tual and social role changes that can increase young peoples involvement in health-
risk behaviors, including marijuana use, excessive drinking, and sexual intercourse
while using substances (Schulenberg, Sameroff, & Cicchetti, 2004). For example,
after leaving high school, rural African-American emerging adults confront envi-
ronments that provide minimal opportunities and resources for employment or con-
tinuing education to help them embark on beneficial life paths. Some who see no
pathway to adequate subsistence or attainment of life goals may cope by drinking
excessively, using marijuana, or engaging in risk behaviors (Fergus & Zimmerman,
2005). Not all emerging adults, however, evince increases in risk behaviors when
they confront sociodemographic and contextual stressors that include opportunity
limitations.
The AIM program was designed to buffer the negative impact of life stress and
contextual stress (e.g., racial discrimination, conflicted family relationships, deviant
companions) on rural African-Americans and to prevent increases in risk behavior
by increasing the provision of family-based buffering processes such as emotional
and instrumental support, vocational coaching and advocacy, and racial socializa-
tion. Thus, AIM was hypothesized to reduce the impact of life stressors on increases
in risk behaviors.
Participants were 347 African-American students in the last 2years of secondary
school (M age=17.7years) who resided in 6 rural counties in Georgia. We targeted
age 17 because it marks the beginning of the transition into emerging adulthood, a
time when rural African-Americans increase risk behaviors that carry negative
health consequences (French, Finkbiner, & Duhamel, 2002). These families, repre-
sentative of the areas in which they lived (Boatright, 2005), can be described as
working poor. Recruitment procedures and the conceptual underpinnings of AIM
are described in Brody, Yu, Chen, Kogan, and Smith (2012).
The AIM program consists of six consecutive weekly meetings, held at commu-
nity facilities and structured in the same manner as SAAF and SAAFT.Parents and
youths receive 12h of prevention training. Parents in the prevention condition were
taught to provide developmentally appropriate emotional and instrumental support,

to provide ongoing racial socialization that included strategies for dealing with dis-
crimination, to provide occupational and educational mentoring, to promote auton-
omy and adult responsibility, and to encourage responsible decisions about risk
behaviors. Youths in the prevention condition were taught to develop a future orien-
tation, to plan to meet goals, to identify people in their communities who could help
them attain goals, to cope with barriers and racial discrimination, and to formulate
self-care strategies.
Latent growth modeling (LGM) was used to test the efficacy of AIM (Brody,
Chen, Kogan, Smith, & Brown, 2010; Brody, Yu etal., 2012). Contextual risk, a
composite formed from family conflict, affiliation with deviant companions, and
exposure to racial discrimination, was regressed on the intercept of alcohol use
(alcohol use prevalence at pretest); participation in AIM, contextual risk, and the
AIMcontextual risk interaction were regressed on the slope of alcohol use 2years
after AIM participation and on the slope of substance use problems. Contextual risk
was positively associated with the intercept of alcohol use, indicating that partici-
pants who experienced high levels of contextual risk were likely to report alcohol
use at the beginning of the study. A significant AIMcontextual risk interaction
predicted the slope of alcohol use. As predicted, the interaction suggests that partici-
pants in the control group who experienced high levels of contextual risk were more
likely than AIM participants to increase their alcohol use over time. No significant
differences between the control and AIM groups emerged for alcohol use when
contextual risk was low.
Contextual risk was also positively associated with the intercept of substance use
problems, indicating that participants who experienced high levels of contextual
risk were also likely to report a substance use problem at the beginning of the study.
A significant AIMcontextual risk interaction predicted the slope of substance use
problems. Unpacking the interaction demonstrated, as predicted, that participants in
the control group who experienced high contextual risk reported a greater increase
over time in substance use problems than did participants in the AIM group. Again,
no differences emerged between participants in the control and intervention groups
when contextual risk was low.
sing Prevention Trials toTest GeneEnvironment

U
Hypotheses
Etiological models of drug use disorder, psychopathology, and the studies they have
sponsored have focused primarily on social (e.g., family-, peer-, and community-
level processes) and psychological (e.g., temperament and self-regulation) determi-
nants. However, such models are incomplete. Vulnerabilities to tobacco, alcohol, and
marijuana use and dependence are likely to be influenced by a combination of envi-
ronmental and genetic factors, mediated in part through psychological processes
(Kreek, Nielsen, Butelman, & LaForge, 2005; Rutter, Moffitt, & Caspi, 2006). These
218 G.H. Brody
contributions include genetic main effects, GeneEnvironment interactions (GE),

and geneenvironment correlations (rGE), collectively called geneenvironment
(GE) interplay (Rutter etal., 2006). GE can be seen either as the moderation of a
genetic effect by environmental influence or as the moderation of environmental
influence by genotype. For example, human children and primates whose genotypes
predispose them to high impulsivity may develop poorly in adverse environments but
normally in nurturing environments (Barr etal., 2004; Pauli-Pott, Friedl, Hinney, &
Hebebrand, 2009). Moreover, rGE occur (a) when genetic factors contribute to indi-
vidual differences in exposure to positive or negative life events (such as when
genetically influenced characteristics such as sociability or irritability evoke posi-
tive or negative responses from others) or (b) when genetically influenced behavior
(such as risk-taking propensities) affects the individuals choice of environmental
experiences.
Most existing findings on GE interplay in humans as it relates to drug use have
come from genetic epidemiology, a branch of science that seems ideal for demon-
strating that GE and rGE exist in nature and affect etiology. Prevention science
has less often been considered as an important source of basic information about
the impact of genetic variation on drug use and psychological adjustment out-
comes. This oversight is unfortunate because prevention science has considerable
potential to refine GE hypotheses and to investigate causal mechanisms that are
difficult to explicate in traditional genetic epidemiological designs. We propose
that, through manipulated environments in randomized prevention trials, preven-
tive interventions permit a more facile disentangling of environments from genetic
influences and therefore greater flexibility in characterizing the nature of GE
interplay. Thus, the use of randomized intervention designs brings the power of
experimental manipulation to the study of GE interplay, advancing understanding
of drug use and abuse, and thereby increasing the power of future prevention efforts
(Howe, Reiss, & Yuh, 2002).
Through the implementation of prevention trials, a causal relationship between
an environmental manipulation and the alteration of a targeted outcome can be
identified (Rutter, 2005). Randomized prevention trials rule out rGE as rival expla-
nations. Experimental random assignment of participants to a prevention or control
condition eliminates biases that reflect rGE.For example, youths with certain gen-
otypes may select deviant peers (active rGE), or parents with specific genotypes
that their children share produce particular kinds of family environments (passive
rGE). Accordingly, random assignment has the advantage of ruling out these
potential rGE confounds that, in epidemiological designs, may be mistaken for
pure environmental effects. In addition, the testing of GE hypotheses using ran-
domized prevention trials may enhance statistical power as much as fivefold over
epidemiological genetic approaches (Bakermans-Kranenburg & van IJzendoorn,
2015; van IJzendoorn, 2015); consequently, fewer participants may be needed to
detect a GE interaction in a randomized trial. Of the most importance, testing
GE hypotheses in the context of prevention trials broadens the conceptual mod-
els guiding such trials, contributing to progress within the Institute of Medicine
prevention development cycle (OConnell etal., 2009).
First-Generation GeneIntervention (GI) Research. Existing prevention trials

can serve as experimental contexts in which genetic assessments are used to deter-
mine whether youths are more susceptible to prevention efforts for genetic reasons.
We term this first-generation geneintervention (GI) research. My colleagues
and I have conducted several first-generation studies involving the SAAF, SAAFT,
and AIM prevention trials. These studies have provided initial evidence of the utility
of prevention trials in circumventing the issues inherent in epidemiological GE
studies. In the SAAF randomized trial, Brody, Beach, Philibert, Chen, and Murry
(2009) investigated a GI interaction involving the serotonin transporter gene at
the 5-HTTLPR.Two strands of research occasioned a focus on this variable nucleo-
tide polymorphism (VNTR) in the promoter region of SLC6A4. Genetic association
studies indicate that carrying a short allele at the 5-HTTLPR was a risk factor for
alcohol dependence and other drug use among adults (Kreek etal., 2005). Other
studies examined associations between a short allele at the 5-HTTLPR and indica-
tors of low self-control, a risk factor for a cluster of risk behaviors that included
alcohol use and early sexual activity (Moffitt, Poulton, & Caspi, 2013). The purpose
of this study was to test the GI hypothesis that random assignment to the SAAF
prevention group versus a control group would interact with genetic risk to predict
youths risk behavior initiation. Specifically, we predicted that (a) youths who car-
ried one or two copies of the short allele at the 5-HTTLPR and were assigned ran-
domly to the control condition would initiate more risk behaviors than would youths
with the same genotype who were assigned randomly to the SAAF prevention con-
dition; (b) youths who carried a short allele and were assigned to the control condi-
tion would initiate more risk behaviors than would youths without genetic risk
assigned randomly to the prevention or control condition; and (c) youths who car-
ried a short allele and were assigned to the prevention condition would not initiate
more risk behaviors than would youths without genetic risk who were assigned to
either the prevention or control condition. Data in Fig. 11.1 depict the results. Across
the 29months that separated the pretest and longitudinal follow-up, control youths
carrying one or two copies of the short allele at the 5-HTTLPR initiated risk behav-
iors at twice the rate of SAAF youths who carried a short allele and youths in either
condition who did not carry a short allele.
Three of our subsequent first-generation studies focused on the dopamine recep-
tor 4 gene (DRD4). The long allele of DRD4 (seven or more repeats) is associated
with reduced gene expression (Schoots & Van Tol, 2003) and altered functioning
(Asghari et al., 1994, 1995). This gene is also associated with behavioral self-
control problems such as alcoholism (Laucht, Becker, Blomeyer, & Schmidt,
2007), use of alcohol and other drugs (Brody, Chen, Yu etal., 2012), and impulsiv-
ity (Eisenberg etal., 2007). Prior to our investigations with DRD4, other investiga-
tors found intervention efficacy to be moderated by the 7-repeat version of DRD4.
Specifically, toddlers who carried this allele showed a greater reduction in disrup-
tive behavior after parenting skill intervention than did children who did not carry
it (Bakermans-Kranenburg, van IJzendoorn, Pijlman, Mesman, & Juffer, 2008). In
another experiment, kindergarten students with this genotype were affected more
positively than were those without it when randomly assigned to play computer
220 G.H. Brody
2
Risk Behavior Initiation Index
1 .5 1.5
SAAF Control
Group Assignment
No genetic risk Genetic risk
Fig. 11.1 Mean risk behavior initiation for each Prevention GroupGenetic Risk condition at
long-term follow-up, adjusted for pretest values. Reprinted from Child Development, 2009, Vol.
80, No. 3, p.654. Used with permission from John Wiley and Sons
games designed to enhance their phoneme awareness skills (Kegel, Bus, & van
IJzendoorn, 2011).
Each of our DRD4 first-generation analyses tested the hypothesis that partici-
pants who carried at least one long allele of DRD4 and were assigned to the control
condition would evince more drug use over time than would (a) carriers of a long
allele who were assigned to the prevention condition and (b) carriers of two short
alleles (less than seven repeats) who were assigned to either condition. The first
study in this series (Beach, Brody, Lei, & Philibert, 2010) tested the DRD4 genetic
moderation hypothesis using four waves of data that spanned 29months, taken from
the SAAF prevention trial. Youths carrying a 7-repeat allele showed a greater
response to intervention than did those with two 4-repeat alleles. Control youths,
but not SAAF youths, with a 7-repeat allele, reported increases in past-month sub-
stance use across the 29-month study period, but this pattern did not emerge for
those with two 4-repeat alleles. These findings are depicted in Fig. 11.2. Using data
from the SAAFT trial, Brody etal. (2014) investigated a DRD4 genetic modera-
tion effect. Adolescents were assigned to SAAFT or to an attention control condi-
tion and then were followed for 22months. The results revealed that male adolescents
who carried at least one long DRD4 allele and were assigned to the control condi-
tion evinced more substance use across 22months than did (a) carriers of the long
allele who were assigned to SAAFT or (b) adolescents assigned to either condition
who carried two short alleles. Figure 11.3 presents these findings.
The study addressing DRD4 genetic moderation with AIM focused on young
adults living in high-risk family contexts (Brody, Yu, & Beach, 2015). We proposed
that young adults who carried at least one long allele of DRD4, were assigned to the
1.20
Estimated Mean Past Month Substance Use
1.00
0.80
0.60
0.40
0.20
0.00
11 12 13 14
AGE
DRD4, youth 4-R and Control DRD4, youth 7-R and Control
DRD4, youth 4-R and SAAF DRD4, youth 7-R and SAAF
Fig. 11.2 Youth genotype at DRD4 moderates SAAF effects on past-month substance use and
parent genotype at DRD4 influences SAAF effects on intervention-targeted parenting. Reprinted
from the Journal of Family Psychology, 2010, Vol. 24, No. 5, p.518. Used with permission from
the American Psychological Association
control condition, and lived in a high-risk family context would evince more drug use
over time than would (a) carriers of at least one long allele who were assigned to
AIM and lived in a high-risk family, (b) young adults assigned to either condition
who carried two short alleles, or (c) young adults who did not live in a high-risk fam-
ily context. The results supported these hypotheses, as evinced by the data depicted
in Fig. 11.4. Taken together, these results extend previous findings that carrying a
long allele of DRD4 increases sensitivity to intervention or prevention programs not
only among toddlers (Bakermans-Kranenburg etal., 2008) and kindergarten children
(Kegel etal., 2011) but also among preadolescents (Beach, Brody, Lei etal., 2010)
and adolescents (Brody etal., 2014).
Finally, my colleagues and I conducted a study using data pooled from the SAAF
and SAAFT trials to create a more powerful sample with which to extend knowl-
edge about G E hypotheses (Brody, Chen, & Beach, 2013). The study was
designed to answer two questions. First, do common variations in the dopaminergic
genes (DRD2, DRD2/ANKK1, TaqI A, and DRD4) and GABAergic genes (GABRG1
and GABRA2) forecast increases in alcohol use for youths assigned randomly to the
pooled control condition? Presumably, youths in the control condition represent
normative alcohol use patterns and therefore provide an opportunity to determine
whether putative risk/sensitivity genes for alcohol use are actually associated with
increases in use across 2years. Second, can participation in efficacious prevention
programming moderate the risk that dopaminergic and GABAergic genes confer for
222 G.H. Brody
Predicted Increases in Frequency of Substance Use from

3.0
2.5
2.0
Pretest to Long-term Follow-up
1.5
1.0
0.5
Control SAAF-T
Intervention Status
DRD4 Status
DRD4 6 or fewer repeat (s) DRD4 7 or more repeat (l)
alleles alleles
Fig. 11.3 Intervention statusDRD4 status effect on male youths predicted substance use. The
vertical lines represent 95% confidence intervals. Reprinted from Health Psychology, 2014, Vol.
33, No. 2, p.187. Used with permission from the American Psychological Association
increases in youths alcohol use across 2years? Prior to this study, evaluations by
our group and others of intervention program efficacy in moderating genetic risk
have been limited by their reliance on single genetic loci (predominantly DRD4)
despite the likelihood that many youths carry multiple risk-conferring loci (Rutter,
2005). This study was unique in combining data from two large prevention trials
involving more than 900 youths to test hypotheses about gene intervention
interactions.
The results revealed that dopaminergic and GABAergic candidate genes forecast
increases in alcohol use across 2years. This finding established the validity of these
genes designation as risk/sensitivity conferring because each gene forecasts
increases in alcohol use. Second, the risk conferred by the six genes was moderated
by participation in a prevention program. Youths who carried the risk variations of
the dopaminergic and GABAergic genes who were assigned to the control condition
evinced more alcohol use over time than did (a) those who were assigned to the pre-
vention condition or (b) those assigned to either condition who carried the nonrisk
genetic variants. Of the six significant geneprevention interactions, three remained
significant (p<.05) after significance levels were adjusted for multiple comparisons:
GABRG1, Block 2; GABRA2, Block 1; and DRD2. These results are presented in Fig.
11.5. The results support Belsky, Bakermans-Kranenburg, and van IJzendoorns
(2007) differential susceptibility hypothesis in which variants of specific genes,
1.0
Predicted Probabaility of Any Drug Use
0.8
0.6
0.4
0.2
0.0
age 17 age 18 age 19 age 20
0.5
Predicted Probabaility of Regular Drug Use
0.4
0.3
0.2
0.1
0.0
age 17 age 18 age 19 age 20
Low family risk, DRD4 s, control Low family risk, DRD4 s, AIM
High family risk, DRD4 s, control High family risk, DRD4 s, AIM
Low family risk, DRD4 l, control Low family risk, DRD4 l, AIM
High family risk, DRD4 l, control High family risk, DRD4 l, AIM
Fig. 11.4 Growth in probability of past-month drug use by family risk, AIM assignment, and
DRD4 status. Low family risk: 1 SD below the mean; high family risk: 1 SD above the mean.
Reprinted from Development and Psychopathology, 2015, Vol. 27, No. 1, p.44. Used with permis-
sion from Springer Science+Business Media
including dopaminergic and GABAergic genes, are proposed to render individuals

more susceptible to the surrounding environment whether it is characterized by high
positivity or high risk. The finding that, after exposure to the protective processes that
the prevention programs offered, carriers of risk/sensitivity genes evinced less alco-
224 G.H. Brody
0.0 0.2 0.4 0.6 0.8 1.0
0.0 0.2 0.4 0.6 0.8 1.0

Predicted Alcohol Use

Non Risk Risk Non Risk Risk
GABRG1 Block 2 GABRG2 Block 2
Invervention condition Invervention condition
Control Intervention Control Intervention
0.2 0.4 0.6 0.8 1.0
Non Risk Risk

DRD2
Invervention condition
Control Intervention
Fig. 11.5 Geneprevention interaction for dopaminergic and GABAergic genes, with standard
error bars included. Alcohol use is defined as occasions on which alcohol was consumed during the
past month, rated on a scale of 0 (never) to 6 (30 or more times). Reprinted from the Journal of
Child Psychology and Psychiatry, 2013, Vol. 54, No. 8, p.868. Used with permission from John
Wiley and Sons
hol use over time than did similar youths in the control condition supports differential
susceptibility predictions. These results extended the existing literature to previously
unstudied genes. The finding that participation in prevention moderated genetic risk,
and the retention of significance by three of those genes after controls for multiple
comparisons were applied, supports the importance of prevention efforts.
Second-Generation Gene Intervention Research. Only recently have prevention
scientists begun to examine the processes that account for or mediate first-generation
GI findings. These investigations can be termed second-generation GI research.
The aforementioned study (Brody et al., 2014) using the SAAFT trial, and the
Brody et al. (2015) study, included analyses designed to determine the processes
through which GI interactions o perate to influence drug use trajectories. The study
by Brody et al. (2014) tested a mediated moderation hypothesis in which parents
assigned to SAAFT whose children carried a long allele of DRD4 would evince
greater increases in intervention-targeted protective parenting processes (e.g., moni-
toring, reciprocal communication, positive problem-solving, and a clear articulation
of drug use norms) than would parents in the other DRD4group assignment condi-
tions. Youths in the SAAFT condition who carried at least one long allele of DRD4
were conjectured to be more responsive than were other youths to intervention-
induced changes in protective parenting processes. This sensitivity, in turn, was
expected to reinforce parents use of these practices, resulting in a decrease in youths

substance use over time. The results, depicted in Fig. 11.6, show that the effect of the
SAAFTDRD4 interaction on changes in substance use became nonsignificant,
changing from =.72, p<.01, when the interaction effect on intervention-targeted
protective parenting was omitted from the analysis, to =.39, p>.05, when this
interaction effect was included in the analysis. As expected, increases in interven-
tion-targeted parenting forecast decreases in substance use, =.56, p<.01.
In the Brody etal. (2015) study, we hypothesized a GI interaction in which
young adults who lived in high-risk family contexts, were assigned to the control
condition, and carried a long allele of DRD4 would evince increases in vulnerability
cognitions for drug use, a proximal risk factor for escalation. We further proposed
that this interaction would account for the Family RiskAIM ParticipationDRD4
interaction. The theoretical basis for this mediated moderation hypothesis was
drawn from the premise that African-American emerging adults who experience
high levels of family stress may come to believe that they have little to lose by aban-
doning planful, conventional orientations in favor of a present orientation that pro-
motes living in the moment. These young adults are at heightened risk of
developing cognitions that increase their likelihood of drug use, such as intentions
to use drugs, willingness to use them, and positive prototypes or images of drug-
using peers. Thus, we formed a construct labeled vulnerability cognitions that com-
prised willingness, intentions, and prototypes; we hypothesized that this construct
would serve as a mediator.
A parallel growth model (Muthn & Curran, 1997) was executed to test the medi-
ated moderation hypothesis that AIM effects on increases in vulnerability cognitions
Pretest Posttest
SAAFTeen
-.041
DRD4 status -.12
.22* Intervention-
SAAFTeen DRD4
status Targeted
Parenting
.91***
Intervention- .70 .79 .71 .64

Targeted
Problem Reciprocal Substance use
Parenting Monitoring
solving communication expectations
.58 .64 .68 .55

2(50) = 88.02, p = .001; CFI = .94; RMSEA = .06 (.04, .08)
Problem Reciprocal Substance use
Monitoring
solving communication expectations *p < .05. ***p < .001.
Fig. 11.6InterventionDRD4 effects on increases in intervention-targeted parenting for male

youths. Socioeconomic and neighborhood risks were controlled. Standardized regression coeffi-
cients were used. Reprinted from Health Psychology, 2014, Vol. 33, No. 2, p.188. Used with per-
mission from the American Psychological Association
226 G.H. Brody
for youths who lived in high-risk families and carried a long allele of DRD4 would
account for its efficacy in reducing drug use among this vulnerable group. The models
depicted in Fig. 11.7 demonstrated that (a) the Family RiskAIM ParticipationDRD4
Status interaction effect on vulnerability cognitions found in previous analyses also
emerged in the mediation model (Path A), (b) the path from the slope for vulnerability
cognitions to the slope for drug use (Path B) was positive and significant, and (c) the
path from the Family RiskAIM ParticipationDRD4 Status interaction to the slope
for drug use became nonsignificant when the effect of the three-way interaction on
vulnerability cognitions was included in the model (Path C: =.209, p<.02 without
the growth of vulnerability cognitions in the model; Path C: =.004, ns with the
growth of vulnerability cognitions in the model). Thus, AIMs efficacy in preventing
increases in drug use among rural African-American youths living in high-risk
Pretest Wave 2 Wave 3 Wave 4
Intercept, Slope,
Family risk Drug Use Drug Use
DRD4 status
AIM
Path C -0.004
Family risk DRD4
Path C -0.209*
Family risk AIM
DRD4 AIM Path B
2.116***
0.394***
Family risk DRD4

AIM
Path A
Intercept, -0.097* Slope,
Vulnerability Vulnerability
Cognitions Cognitions
Pretest Wave 2 Wave 3 Wave 4
Fig. 11.7 Vulnerability cognitions as a mediator for the effect of AIMDRD4 statusrisky family
on probability of past-month drug use with gender and SES risk index controlled. Dashed lines indi-
cate nonsignificant paths. *p<.05, ***p<.001. Reprinted from Development and Psychopathology,
2015, Vol. 27, No. 1, p.46. Used with permission from Springer Science+Business Media
families and carrying a long allele of DRD4 occurred through its effect on deterring
the development of cognitions that make the use of drugs attractive (indirect effect of
Path APath B=.204, SE=.098, p=.037).
TheWay Forward
On the basis of information gained from etiological GE studies of drug use, drug
abuse, and psychopathology, and from first- and second-generation GI studies,
research can be designed to increase the precision of preventive interventions by
addressing questions concerning nonspecific environmental effects. Various envi-
ronmental risk mechanisms have nonspecific effects on the use and abuse of alcohol
and other drugs and on the intermediate phenotypes that forecast them. For exam-
ple, exposure to family conflict and violence, harsh parenting, stressful life events
involving loss or threat, and economic insecurity all forecast both internalizing and
externalizing problem phenotypes as well as drug use and abuse (Rutter, 2006). A
potential explanation for these nonspecific effects may be that particular adverse
childhood events are connected to different intermediate phenotypes and different
drug use and psychopathology outcomes through different G E pathways. For
example, it has been shown that early adversity forecasts depressive symptoms
among carriers of the short allele at the 5-HTTLPR and antisocial behavior among
carriers of high-activity alleles of MAOA. These findings show that exposure to
early adversity in combination with different gene systems culminates in different
adjustment phenotypes (Beach, Brody, Gunter etal., 2010; Caspi etal., 2002).
Identifying protective mechanisms in individuals, families, schools, and com-
munities; linking their beneficial effects to various intermediate phenotypes that
forecast drug use initiation and escalation, emotional problems, and behavior prob-
lems; and insuring that they constrain the combined impact of genomic and environ-
mental effects will provide important information for building a new generation of
preventive interventions that target multiple positive outcomes. If genetically sus-
ceptible or vulnerable subgroups can be identified for analysis, modest associations
may prove to be stronger and more specific than was previously believed. It will be
important in building this next generation of preventive interventions to describe the
extent to which developmental stage, gender, and significant contextual factors
modify the impact of nonspecific protective factors.
ACaveat About Using Genetic Data toSelect Participants
The discussion thus far has focused on the ways in which the integration of genetic
data into prevention trials can enhance understanding of etiology and the design of
more effective interventions. It is natural, therefore, to ask whether such preventive
interventions would also be more efficacious and cost-effective if only those indi-
viduals who responded most strongly were selected to receive them. That is, if
228 G.H. Brody
genes moderate response to preventive interventions, should we identify and select

individuals for preventive interventions based on genotypes? Targeting interven-
tions by genotype is problematic for several reasons, several of which follow.
Targeted Interventions andthePrevention Paradox
A major concern about targeting preventive interventions to individuals at high risk,

whatever risk mechanisms are involved, is the so-called prevention paradox. This
phenomenon is a seemingly contradictory situation in which the majority of cases
of a complex disorder come from a population at low or moderate risk for that dis-
order, with only a minority of cases coming from the high-risk population (Rose,
1981). Aiming prevention only at those at high relative risk might reduce their indi-
vidual risk, but it would do little to reduce the total prevalence of the disorder in the
population. This concern is well worth considering before genotype is used as a
selection criterion for preventive intervention, even in the case of selective preven-
tion programs.
Potential forDiscrimination andStigmatization
Another major concern in identifying and selecting individuals for preventive inter-
ventions is the possibility of stigmatization or discrimination arising from such
selection (Offord, Kraemer, Kazdin, Jensen, & Harrington, 1998; Robson, Storm,
Weitzel, Wollins, & Offit, 2010). Asymptomatic individuals or their family mem-
bers may be treated differentially based on real or presumed genotype (Billings
etal., 1992). Conversely, and similarly to the prevention paradox, negative effects
could also accrue to those not selected. Children or parents not selected for preven-
tion programming might feel inappropriately invulnerable to the risk of drug-
related, emotional, or conduct problems. These concerns are reasons to avoid the
adoption of genetically targeted interventions.
Growing Impracticality ofGenotypic Targeting
We anticipate that the number of genotypes found to confer susceptibility, vulnera-

bility, or risk at various developmental stages and in different contexts will continue
to increase, particularly as candidate gene approaches are replaced with gene-
network approaches. As the list of relevant genes becomes longer, the direct use of
genotypes in the selection of individuals for participation in prevention programs
will become increasingly impractical for several reasons. First, given a sufficient
number of relevant genes, the number of risk-free individuals will become
negligible. Second, even if one assumes that multiple, independently distributed

genes contribute by themselves to susceptibility, vulnerability, and risk, the distribu-
tion of risk genes is likely to take on the shape of a normal (Gaussian) distribution,
confounding efforts to create clear risk and nonrisk groups. In such circum-
stances, creation of genetic risk groups will be arbitrary, with most of the general
population having intermediate risk status. To the extent that selection for risk is
desirable in the context of prevention trials, it is likely that high-risk status will be
captured best by giving attention to the intermediate phenotypes and intervening
processes that genes influence and moderate rather than to discrete genotypes.
Recommendations
I believe that broadly delivering both universal and selective preventive interven-
tions without targeting individuals based on genetic vulnerability may reach a larger
segment of the genetically vulnerable population (Offord etal., 1998) and, conse-
quently, be more effective in affecting population-level prevalence of these common
and costly public health problems. Nevertheless, I believe that incorporation of GE
interplay processes will benefit prevention programs by expanding understanding of
the intermediate phenotypes and processes that account for preventive intervention
program success, enabling program designers to incorporate new targets and
enhancing overall impact in the general population.
Looking Forward
Substantial individual differences exist in response to prevention programming: Many

individuals who experience it do not go on to use and abuse drugs while others do so.
As we implied previously, one important goal in going forward with genepreven-
tion research should be to (a) account for heterogeneity in response to constructed
resilience environments (prevention programs) at the genetic level, with the goal of
(b) understanding better the neurobiological mechanisms underlying this variable
response. Our initial research, described previously, documented that genes involved
in neurotransmission moderated individuals responses to prevention programming.
Subsequent research has focused on the possibility of an interaction of genes that are
implicated in the stress response, such as the oxytocin receptor gene, with prevention
programming to foster biological resilience to adversity as manifested in relatively
longer telomeres (Smearman , Yu, & Brody, 2016). The goal is to use genepreven-
tion designs to suggest the importance of the contributions that neurotransmission
properties could make to the psychosocial and health benefits that exposure to resil-
ience-promoting experiences confers. Epidemiologists and neuroscientists can use
this evidence to illuminate hidden associations among genes, risk and resilience
processes, and disorders. Progress, however, requires the characterization of genetic
230 G.H. Brody
vulnerability to move beyond single genetic polymorphisms. New approaches may

incorporate information about genetic pathways that enhances explanatory power in
gene prevention program designs. Novel loci also may be identified by using
sequencing and genome-wide association methodologies in studies of youths who
respond to prevention programming. It is clear that future efforts to use prevention
programs to understand the role of genes in the etiology of drug use and other
disorders will require transdisciplinary teams of prevention scientists, psychiatric
geneticists, and epidemiologists.
References
Asghari, V., Sanyal, S., Buchwaldt, S., Paterson, A., Jovanovic, V., & An Tol, H.H. M. (1995).
Modulation of intracellular cyclic AMP levels by different human dopamine D4 receptor vari-
ants. Journal of Neurochemistry, 65, 11571165.
Asghari, V., Schoots, O., Van Kats, S., Ohara, K., Jovanovic, V., Guan, H.-C., Van Tol, H.H. M.
(1994). Dopamine D4 receptor repeat: Analysis of different native and mutant forms of the
human and rat genes. Molecular Pharmacology, 46, 364373.
Bakermans-Kranenburg, M.J., & van IJzendoorn, M.H. (2015). The hidden efficacy of interven-
tions: Gene environment experiments from a differential susceptibility perspective. Annual
Review of Psychology, 66, 381409.
Bakermans-Kranenburg, M.J., van IJzendoorn, M.H., Pijlman, F.T. A., Mesman, J., & Juffer, F.
(2008). Experimental evidence for differential susceptibility: Dopamine D4 receptor polymor-
phism (DRD4 VNTR) moderates intervention effects on toddlers externalizing behavior in a
randomized controlled trial. Developmental Psychology, 44, 293300.
Baron, R.M., & Kenny, D.A. (1986). The moderatormediator variable distinction in social psy-
chological research: Conceptual, strategic, and statistical considerations. Journal of Personality
and Social Psychology, 51, 11731182.
Barr, C.S., Newman, T.K., Lindell, S., Shannon, C., Champoux, M., Lesch, K.P., Higley, J.D.
(2004). Interaction between serotonin transporter gene variation and rearing condition in
alcohol preference and consumption in female primates. Archives of General Psychiatry, 61,
11461152.
Beach, S.R. H., Brody, G.H., Gunter, T.D., Packer, H., Wernett, P., & Philibert, R.A. (2010).
Child maltreatment moderates the association of MAOA with symptoms of depression and
antisocial personality disorder. Journal of Family Psychology, 24, 1220.
Beach, S.R. H., Brody, G.H., Lei, M.-K., & Philibert, R.A. (2010). Differential susceptibility to
parenting among African American youths: Testing the DRD4 hypothesis. Journal of Family
Psychology, 24, 513521.
Belsky, J., Bakermans-Kranenburg, M. J., & van IJzendoorn, M. H. (2007). For better and for
Billings, P.R., Kohn, M.A., de Cuevas, M., Beckwith, J., Alper, J.S., & Natowicz, M.R. (1992).
Discrimination as a consequence of genetic testing. American Journal of Human Genetics, 50,
476482.
Boatright, S.R. (2005). The Georgia county guide (24th ed.). Athens: Center for Agribusiness and
Economic Development.
Brody, G.H., Beach, S.R. H., Philibert, R.A., Chen, Y.-F., & Murry, V.M. (2009). Prevention
effects moderate the association of 5-HTTLPR and youth risk behavior initiation: Gene envi-
ronment hypotheses tested via a randomized prevention design. Child Development, 80,
645661.
Brody, G. H., Chen, Y.-F., & Beach, S. R. H. (2013). Differential susceptibility to prevention:
54, 863871.
Brody, G.H., Chen, Y.-F., Beach, S.R. H., Kogan, S.M., Yu, T., Diclemente, R.J., Philibert,
R. A. (2014). Differential sensitivity to prevention programming: A dopaminergic
polymorphism-enhanced prevention effect on protective parenting and adolescent substance
use. Health Psychology, 33, 182191.
Brody, G.H., Chen, Y.-F., & Kogan, S.M. (2010). A cascade model connecting life stress to risk
behavior among rural African American emerging adults. Development and Psychopathology,
22, 667678.
Brody, G.H., Chen, Y.-F., Kogan, S.M., Murry, V.M., & Brown, A.C. (2010). Long-term effects
of the Strong African American Families program on youths alcohol use. Journal of Consulting
and Clinical Psychology, 78, 281285.
Brody, G.H., Chen, Y.-F., Kogan, S.M., Smith, K., & Brown, A.C. (2010). Buffering effects of a
family-based intervention for African American emerging adults. Journal of Marriage and
Family, 72, 14261435.
Brody, G.H., Chen, Y.-F., Kogan, S.M., Yu, T., Molgaard, V.K., Diclemente, R.J., & Wingood,
G. M. (2012). Family-centered program to prevent substance use, conduct problems, and
depressive symptoms in Black adolescents. Pediatrics, 129, 108115.
Brody, G.H., Chen, Y.-F., Yu, T., Beach, S.R. H., Kogan, S.M., Simons, R.L., Philibert, R.A.
(2012). Life stress, the dopamine receptor gene, and emerging adult drug use trajectories: A
longitudinal, multilevel, mediated moderation analysis. Development and Psychopathology,
24, 941951.
Brody, G.H., Murry, V.M., Gerrard, M., Gibbons, F.X., Molgaard, V., McNair, L.D., Neubaum-
Carlan, E. (2004). The Strong African American Families program: Translating research into
prevention programming. Child Development, 75, 900917.
Brody, G.H., Murry, V.M., Kogan, S.M., Gerrard, M., Gibbons, F.X., Molgaard, V., Wills,
T.A. (2006). The Strong African American Families program: A cluster-randomized preven-
tion trial of long-term effects and a mediational model. Journal of Consulting and Clinical
Brody, G.H., Yu, T., & Beach, S.R. H. (2015). A differential susceptibility analysis reveals the
who and how about adolescents responses to preventive interventions: Tests of first- and
second-generation Gene Intervention hypotheses. Development and Psychopathology, 27,
3749.
Brody, G.H., Yu, T., Chen, Y.-F., Kogan, S.M., & Smith, K. (2012). The Adults in the Making
program: Long-term protective stabilizing effects on alcohol use and substance use problems
for rural African American emerging adults. Journal of Consulting and Clinical Psychology,
80, 1728.
Centers for Disease Control and Prevention. (2000). Youth risk behavior surveillanceUnited
States, 1999. Morbidity and Mortality Weekly Report, 49, 196.
Eisenberg, D.T. A., MacKillop, J., Modi, M., Beauchemin, J., Dang, D., Lisman, S.A., Wilson,
D. S. (2007). Examining impulsivity as an endophenotype using a behavioral approach: A
DRD2 Taq1 A and DRD4 48-bp VNTR association study. Behavioral and Brain Functions, 3, 2.
Fergus, S., & Zimmerman, M.A. (2005). Adolescent resilience: A framework for understanding
healthy development in the face of risk. Annual Review of Public Health, 26, 399419.
Foxcroft, D.R., Ireland, D., Lister-Sharp, D.J., Lowe, G., & Breen, R. (2003). Longer-term
primary prevention for alcohol misuse in young people: A systematic review. Addiction, 98,
397411.
French, K., Finkbiner, R., & Duhamel, L. (2002). Patterns of substance use among minority youth
and adults in the United States: An overview and synthesis of national survey findings. Fairfax,
VA: Department of Health and Human Services.
232 G.H. Brody
Howe, G. W., Reiss, D., & Yuh, J. (2002). Can prevention trials test theories of etiology?
Development and Psychopathology, 14, 673694.
Kegel, C.A. T., Bus, A.G., & van IJzendoorn, M.H. (2011). Differential susceptibility in early
literacy instruction through computer games: The role of the dopamine D4 receptor gene
(DRD4). Mind, Brain, and Education, 5, 7178.
Kreek, M.J., Nielsen, D.A., Butelman, E.R., & LaForge, K.S. (2005). Genetic influences on
impulsivity, risk taking, stress responsivity and vulnerability to drug abuse and addiction.
Kumpfer, K. L., Alvarado, R., & Whiteside, H. O. (2003). Family-based interventions for sub-
stance use and misuse prevention. Substance Use and Misuse, 38, 17591787.
Laucht, M., Becker, K., Blomeyer, D., & Schmidt, M. H. (2007). Novelty seeking involved in
mediating the association between the dopamine D4 receptor gene exon III polymorphism and
heavy drinking in male adolescents: Results from a high-risk community sample. Biological
Leigh, B.C., & Stahl, R. (1993). Substance use and risky sexual behavior for exposure to HIV:
Issues in methodology, interpretation, and prevention. American Psychologist, 48, 10351045.
Moffitt, T.E., Poulton, R., & Caspi, A. (2013). Lifelong impact of early self-control. American
Scientist, 101, 352359.
Molgaard, V., & Spoth, R. (2001). The Strengthening Families Program for young adolescents:
Overview and outcomes. Residential Treatment for Children and Youth, 18, 1529.
Muthn, L.K., & Curran, P.J. (1997). General longitudinal modeling of individual differences in
experimental designs: A latent variable framework for analysis and power estimation.
Psychological Methods, 2, 371402.
National Center on Addiction and Substance Abuse. (2000). No place to hide: Substance abuse in
mid-size cities and rural America. NewYork: Columbia University.
OConnell, M.E., Boat, T., & Warner, K.E. (2009). Preventing mental, emotional, and behavioral
disorders among young people: Progress and possibilities. Washington, DC: National
Academies Press.
Offord, D.R., Kraemer, H.C., Kazdin, A.E., Jensen, P.S., & Harrington, R. (1998). Lowering the
burden of suffering from child psychiatric disorder: Trade-offs among clinical, targeted, and
universal interventions. Journal of the American Academy of Child and Adolescent Psychiatry,
37, 686694.
Pauli-Pott, U., Friedl, S., Hinney, A., & Hebebrand, J.(2009). Serotonin transporter gene polymor-
phism (5-HTTLPR), environmental conditions, and developing negative emotionality and fear
in early childhood. Journal of Neural Transmission, 116, 503512.
Proctor, B. D. & Dalaker, J. (2003). Poverty in the United States: 2002 (Current Population
Reports, P60-222). Washington, DC: US Census Bureau.
Robson, M.E., Storm, C.D., Weitzel, J., Wollins, D.S., & Offit, K. (2010). American Society of
Clinical Oncology policy statement update: Genetic and genomic testing for cancer suscepti-
bility. Journal of Clinical Oncology, 28, 893901.
Rose, G. (1981). Strategy of prevention: Lessons from cardiovascular disease. British Medical
Journal, 282, 18471851.
Rutter, M.L. (2005). Environmentally mediated risks for psychopathology: Research strategies
and findings. Journal of the American Academy of Child and Adolescent Psychiatry, 44, 318.
Rutter, M. L. (2006). Genes and behavior: Nature-nurture interplay explained. Malden, MA:
Blackwell.
Rutter, M.L., Moffitt, T.E., & Caspi, A. (2006). Gene-environment interplay and psychopathol-
ogy: Multiple varieties but real effects. Journal of Child Psychology and Psychiatry, 47,
226261.
Schoots, O., & Van Tol, H. H. M. (2003). The human dopamine D4 receptor repeat sequences
modulate expression. Pharmacogenetics Journal, 3, 343348.
Schulenberg, J.E., Sameroff, A.J., & Cicchetti, D. (2004). The transition to adulthood as a critical
juncture in the course of psychopathology and mental health. Development and Psychopathology,
16, 799806.
Smearman, E. L., Yu, T., & Brody, G. H. (2016). Variation in the oxytocin receptor gene moderates
the protective effects of a family-based prevention program on telomere length. Brain and
Behavior, 6(2), e00423. doi:10.1002/brb3.423.
Spoth, R.L., Redmond, C., & Shin, C. (1998). Direct and indirect latent-variable parenting out-
comes of two universal family-focused preventive interventions: Extending a public health-
oriented research base. Journal of Consulting and Clinical Psychology, 66, 385399.
Spoth, R.L., Redmond, C., Trudeau, L., & Shin, C. (2002). Longitudinal substance initiation out-
comes for a universal preventive intervention combining family and school programs.
Psychology of Addictive Behaviors, 16, 129134.
van IJzendoorn, M. H. (2013). Genetic differential susceptibility experiments. In Jacobs
Foundation Conference, Genetic Moderation of Intervention Effects. Marbach Conference,
hningen, Germany.
Chapter 12
Developmental Mechanisms
inGene-by-Intervention (GxI) Effects
intheFast Track Trial
KennethA.Dodge
The first decade of this century witnessed extraordinary progress in our understand-
ing of the role of the environment in the development of psychopathology (e.g.,
Caspi etal., 2002), how interventions can be mounted to prevent psychopathology
(e.g., Conduct Problems Prevention Research Group, 2002), and how individual
differences marked by genetic factors moderate the impacts of the environment and
intervention (e.g., Brody, Beach, Philibert, Chen, & Murry, 2009). The past several
years have built on this foundation with progress in understanding the mechanisms
through which these environmental and intervention effects, and their interaction
with genetic factors, occur. This chapter describes some of this research directed
toward understanding the mechanisms through which early stress and harsh parent-
ing have an effect on the development of externalizing psychopathology, the mecha-
nisms through which one preventive intervention called Fast Track operates, and the
mechanisms through which interaction with one gene, the glucocorticoid receptor
gene NR3C1, has on regulating the response to the Fast Track intervention. The
chapter will focus on developmental epidemiological studies from the Child
Development Project and the randomized controlled trial evaluating the impact of
the Fast Track intervention to prevent chronic violence.
The intervention experiment is a complement to the epidemiological study, a
useful tool not only for examining the efficacy of an intervention but also for study-
ing the processes through which problem behavior outcomes develop. Most studies
of psychosocial processes are fraught with problems of interpretation of bivariate
correlations. An interpretation that the process causes the outcome is threatened by
alternate interpretations that the process was self-selected by the individual or both
the purported causal process and the outcome are actually caused by a third v ariable.
K.A. Dodge (*)

Duke University, Box 90545, 302 Towerview Drive, Durham, NC, USA
e-mail: dodge@duke.edu

DOI10.1007/978-3-319-49227-8_12
236 K.A. Dodge
The intervention experiment is essentially a laboratory manipulation of a process

that allows stronger confidence in the causal interpretation.
Study of gene-by-environment and gene-by-intervention effects are ordinarily
subject to the same problems, and the intervention experiment solves the selection
and third-variable problems in the same manner.
ascade Model oftheDevelopment ofChronic

C
Externalizing Disorder
Longitudinal studies following community samples of children from early life into
adulthood have charted the path that leads from the early experience of trauma, fam-
ily stress, and harsh discipline by parents to early conduct problems at the time of
formal school entry. From that age, these studies have found remarkable predictabil-
ity of adult externalizing psychopathology and associated indicators of criminal
arrest, substance abuse, aggression toward spouses and offspring, and antisocial per-
sonality disorder, from early indicators of conduct problems in preschool and pri-
mary grades. The path through which early disobedience of parents and teachers and
minor aggressive conflicts with siblings and classroom peers grow into more serious
violence and antisocial behaviors is one that has been characterized as a developmen-
tal cascade (Dodge, Greenberg, Malone, & Conduct Problems Prevention Research
Group, 2008). Once conduct problems take hold, transactions between the young
child and her or his social environment become increasingly filled with conflict and
resistant to change.
Dodge etal. (2009) followed a community sample of 585 boys and girls of the
Child Development Project from preschool through young adulthood and used par-
tial least squares analysis to identify the contingent relations between early experi-
ences and a succession of later life events and individual outcomes. The cascade that
they identified (depicted in Fig. 12.1) involves a reciprocal relation between a
childs behavior and the life events that the behavior elicits. These life events, turn,
lead the child to behave in different ways. This pattern repeats across development,
leading to a cascade in which early minor problems lead to serious out comes by
adulthood.
The cascade begins with the nexus between very early difficult temperament
(perhaps reflecting genetic factors) and an adverse sociocultural context of poverty
and stress. Dodge et al. (2009) found that this combination made it difficult for
parents to provide consistent and non-harsh discipline for children during preschool.
Early harsh parenting, in turn, increased the odds that a child would come to school
displaying conduct problems with authority and peers. These conduct problems are
likely to elicit rejection from peers and disciplinary action from school authorities.
Parents respond to the conflict and trouble initiated by the child with declining sup-
port and reduced monitoring, even though increased monitoring would be most
12 Developmental Mechanisms in Gene-by-Intervention (GxI) 237
Genetic Risk
Social Failure at:

Information School, Adolescent Adult
Processing Peers, Problems Disorder
Deficits Authority
Early Trauma,
Stress,
Prob Parenting
Fig. 12.1 Cascade model of the development of chronic xternalizing disorder
helpful in steering the child back on to the mainstream track. Most likely, these
parents have fatigued from conflict with the child and are withdrawing their support.
The decreased parental monitoring opens the door for the child to gravitate toward
deviant peer groups, where the child is exposed to new models of antisocial behav-
ior and is reinforced through a process called deviancy training. Ultimately, the
adolescents deviant and antisocial behavior intensifies, becomes chronic, and lasts
into adulthood.
The cascade is one in which minor perturbations early in life elicit environmental
events that reinforce the childs deviant tendencies, which elicit stronger environ-
mental reactions and then increased deviant behavior by the child until the pattern
becomes chronic and difficult to change. It suggests the pivotal importance of genetic
and environmental factors, as well as interventions, early in life. Although the pro-
cess is one of environmental reinforcement and affordance unfolding across develop-
ment, it points toward the need for early intervention to interrupt and reroute the
cascade away from a life of chronic antisocial behavior.
The empirical findings identifying cascades that start early in life suggest that
interventions may be most successful if they begin early and interrupt the cascading
process before it becomes too self-reinforcing.
ocial Information Processing Mechanisms inAntisocial

S
Behavior
The dynamic reciprocal exchange between the developing child and the peer environ-
ment suggests the role of the childs cognitive processing of information about peers
and the childs interpretation of peer behaviors as a factor that perpetuates and exacer-
bates the cascade. Indeed, models of social information processing posit a pattern of
238 K.A. Dodge
defensive processing that grows out of experiences of early physical maltreatment

(Dodge, Bates, & Pettit, 1990) and peer social rejection (Dodge etal., 2003) that leads
to growth in a childs peer-directed aggressive behavior over time (Dodge, 2006).
The pattern of defensive processing starts with hypervigilance to threat cues
from the peer environment. The child who has been mistreated by peers and adults
is likely to anticipate further rejection and comes to be highly vigilant for cues indi-
cating threat (Dodge etal., 1990). When these cues are detected, the child is likely
to interpret them as an indication of hostile intent by the peer, called a hostile attri-
butional bias. This hostile attribution leads the child to experience anxiety and anger
and to adopt self-defensive goals. These goals are transformed into reactive aggres-
sive behavior, which serves to ward off any immediate threat.
From the perspective of the peer and third-party observers, the childs defen-
sive processing and reactive aggression are unwarranted. They view the child as
unstable and inappropriately aggressive, leading them to reject the child even fur-
ther and to act toward the child with increased, actual aggression. The child inter-
prets the peers response as evidence that his or her original interpretation was
accurate, and the gulf between the child and the peer group becomes even wider
and more difficult to change. The child cascades toward permanent separation
from the mainstream peer group, with each side confident that the other side is the
primary cause of the conflict. As noted earlier, once the cascade deepens, it is dif-
ficult to interrupt. Early intervention seems a more hopeful path than after-the-
fact treatment.
Correlations withAggressive Behavior
Evidence has accumulated to support the hypothesis that defensive processing is

correlated with aggressive behavior. Numerous studies show that, relative to nonag-
gressive peers, aggressive children attend to more threatening social cues, interpret
those cues as hostile, experience anxiety and anger, react with increased heart rate,
release more testosterone, adopt self-defensive goals, generate more aggressive
responses from memory, and evaluate those responses as justified and instrumen-
tally effective in solving the current challenge (Dodge & Pettit, 2003).
Prediction ofGrowth inAggression
Longitudinal evidence from the Child Development Project shows that this defen-
sive processing pattern predicts growth in aggressive behavior across time, control-
ling for prior levels of aggressive behavior (Pettit, Lansford, Malone, Dodge, &
Bates, 2010). Thus, the defensive pattern is not merely correlated with aggression,
it is responsible for the cascade from earlier conduct problems to later more serious
and chronic antisocial behavior.
Acquisition ofDefensive Processing Patterns
Social information processing patterns are both inherited and acquired through life
experiences. Early stressful experiences, particularly physical abuse and harsh dis-
cipline from parents and chronic social rejection by peers, have been found to pre-
dict the development of defensive processing styles in children.
In the Child Development Project, Dodge etal. found that harsh discipline in the
first 5years of life predicts defensive processing, including hypervigilance to threat
cues, hostile attributional biases, and aggressive problem-solving (Dodge, Pettit, Bates,
& Valente, 1995). In the same sample, peer social rejection also predicts defensive
processing, specifically selective attention to threat cues, hostile attributional biases,
and evaluations that aggressive behavior leads to instrumentally successful outcomes
(Dodge etal., 2003).
enetic Factors That Moderate theImpact

G
oftheEnvironment
Although early stressful experiences of physical maltreatment and peer social rejec-
tion exert main effects on increasing risk for maladjustment outcomes in adulthood,
not all children who experience these stressors suffer these outcomes. Genetic fac-
tors have been found to moderate the effect of life experiences, both protecting
certain children from long-term harm and exacerbating risk for vulnerable children.
The Child Development Project has identified several genetic factors that interact
with early stressful experiences.
Monoamine Oxidase A(MAOA)
One genetic factor is a functional polymorphism in the promotor region of the

monoamine oxidizing gene monoamine oxidase A (MAOA). MAOA degrades
monoamines and thus regulates behavior in response to stressors. The upstream
variable number of tandem repeat (uVNTR) markers influences the development of
aggressive behavior by altering how an individual responds to stressors. MAOA is
located on the X chromosome and is involved in the catabolism of monoamines
such as dopamine, serotonin, and noradrenaline.
Building on the seminal findings of Caspi etal. (2002), Edwards etal. (2010)
hypothesized that children with low-activity alleles would be more susceptible to
the adverse impact of early harsh discipline than would other children. They identi-
fied 31.2% of the Child Development Project male sample as hemizygous for a
low-activity allele and contrasted this group with the remainder of the sample.
Using a continuous measure of harsh physical discipline in the first 5years of life
based on mother and father behavior as reported on the Conflict Tactics Scale, they
240 K.A. Dodge
C Delinquent Behavior
5
Mean Delinquency Score
3
High MAOA
Low MAOA
2
0
0 4 8 12 16 20 24
Physical Discipline
Fig. 12.2 The interaction between MAOA and harsh physical discipline on childrens delinquency
(Edwards etal., 2010, Journal of Child Psychology & Psychiatry)
found a linear effect of early harsh physical discipline on a latent construct defined
by both mother and teacher reports of the childs delinquent behavior on the Child
Behavior Checklist collected in 12 consecutive years between ages 6 and 17. In sup-
port of their hypothesis, they found that this relation was stronger (with a steeper
slope) for children with the MAOA risk allele than for children without the risk
allele (shown in Fig. 12.2).
amma-Aminobutyric Acid Receptor Subunit Alpha-2

G
(GABRA2)
Gamma-aminobutyric acid is the major inhibitory neurotransmitter. Variations in

GABRA2 OMIM137140 have been associated with adult alcohol dependence
(Edenberg etal., 2004) and childhood conduct disorder (Dick etal., 2006) in the
Collaborative Study on the Genetics of Alcoholism (COGA) project.
Dick etal. (2009) hypothesized that this gene could moderate the impact of paren-
tal monitoring and supervision on the development of externalizing problems in the
children of the Child Development Project, specifically, that children with two copies
of a single-nucleotide polymorphism in GABRA2 would be especially vulnerable to
the effects of poor parental supervision and monitoring. Externalizing problems were
measured by growth mixture modeling of a class of elevated and persistent problems
based on annual reports between ages 12 and 22 on the Achenbach Young Adult Self-
Report. Empirical findings supported this hypothesis. As shown in Fig. 12.3, parental
supervision had a strong effect on externalizing problem scores for the group of
individuals with two copies of the minor (risk) allele. Among this group of children,
of those who had experienced poor parental supervision at age 11years, 28% ended
up in the elevated externalizing problems group, in contrast with just 16% of children
who had experienced strong parental supervision. Children who had no copies of the
Fig. 12.3 The interaction between GABRA2 and parental supervision on externalizing psychopa-
thology (Dick etal., 2009, Archives of General Psychiatry)
risk allele were buffered from the adverse effect of poor parental supervision: both
high-supervision and low-supervision groups were unlikely to end up with persistent
elevated externalizing problems outcomes.
Gene-by-Intervention (GxI) Effects
The gene-by-environment interaction findings reported above suggest that some

children are relatively impervious to the effects of environmental experiences on
externalizing outcomes. They are buffered from the adverse effects of harsh physi-
cal discipline or poor parental supervision and, reciprocally, do not receive any extra
benefit from parenting characterized by non-harsh discipline or strong supervision.
These findings suggest that some children might benefit more than others from sys-
tematic interventions that are administered as part of a therapeutic setting. Belsky
and Pluess (2009) termed this pattern differential susceptibility. Brody, Chen, and
Beach (2013) have provided the strongest evidence to date of differential suscepti-
bility to preventive intervention among children.
The Fast Track Prevention Experiment
The Fast Track experiment represents a unique opportunity to test the hypothesis that
some children, identified by genetic markers, might be more susceptible to the effects
of preventive intervention than other children. Numerous preventive interventions
have grown from the assertion that training in social-cognitive skills can interrupt the
adverse effect of early environmental experiences and help a child adopt more socially
competent, nonaggressive behaviors. Guerra and Slaby (1990) successfully taught
242 K.A. Dodge
decision-making skills to adolescent offenders. Hudley and Graham (1993) trained

aggressive boys to make nonhostile attributions. Lochman and Wells (2004) taught
coping skills to young aggressive boys. Jones, Brown, and Aber (2011) reported posi-
tive impact of the 4 Rs program to promote social-cognitive processes (specifically, to
reduce hostile attributional biases, decrease aggressive response strategies, and alter
normative beliefs about aggression) on elementary school childrens teacher-rated
aggressive behavior.
Durlak, Weissberg, Dymnicki, Taylor, and Schellingers (2011) meta-analysis of
213 prevention studies that aimed to enhance social emotional learning yielded mean
intervention effect sizes (ES) of 0.56 on these skills and 0.22 on conduct problem
behavior measured immediately post-intervention. Six months later, intervention
impact remained significant although reduced in effect size (ES=0.26 and 0.14 for
social emotional learning and conduct problems, respectively). Wilson and Lipsey
(2007) yielded a mean intervention effect size of 0.21 for the effect of school-based
cognitive-behavioral programs on aggressive and disruptive behavior.
Although these findings encourage the implementation of social-cognitive skill
training interventions with aggressive children, the relatively modest effect sizes
suggest that perhaps some aggressive children respond more favorably to this inter-
vention than do other children.
The Fast Track program exposed high-risk kindergarten children to a 10-years
intervention designed to improve their social-cognitive skills through small group
activities, classroom curricula, parent training, peer coaching, and tutoring.
Kindergarten children from each of four geographic sites were identified as high
risk for adolescent antisocial behavior through a multiple-gating screening proce-
dure (see Lochman & CPPRG, 1995, for details) that combined teacher and parent
ratings of disruptive behavior to all 9594 kindergarteners across three cohorts
(19911993) in 55 schools at four geographic sites. Children were selected based on
a within-site standardized screen score, moving from the highest score downward
until desired sample sizes were reached within sites, cohorts, and conditions.
Ultimately, 891 children (n=445 for intervention and n=446 for control) partici-
pated. The mean externalizing scale T score on the Kindergarten Teachers Report
Form of the Child Behavior Checklist (Achenbach, 1991b) was 66.4 (national
mean=50, sd=10). The mean age of participants at the time of identification was
6.5years (SD=0.48). The sample included 51% African American, 47% European
American, 2% other ethnicity, and 69% boys.
Intent-to-treat analyses of a randomized controlled trial yielded significant main-
effect impacts of intervention on childrens aggressive behavior after Grades 1
(CPPRG, 1999), 3 (CPPRG, 2002), 45 (CPPRG, 2004), and 9 (CPPRG, 2007).
Not all analyses at all ages yielded significant effects (e.g., no effects were found in
grades 78, CPPRG, 2010).
Analyses at age 25 indicated that 69% of the control-group members displayed
at least one externalizing, internalizing, or substance-abuse psychiatric problem
(based on self or peer interview), in contrast with 59% of those assigned to interven-
tion. This pattern also held for separate tests for externalizing problems, internaliz-
ing problems, and substance-abuse problems, and for each of three cohorts, four
sites, male, female, African American, white, moderate-risk, and high-risk sub-
groups. Intervention participants also received lower severity-weighted violent and
drug crime conviction scores, lower risky sexual behavior scores, and higher well-
being scores.
Importantly, some analyses of the impact of random assignment to Fast Track
yielded significant effects only for the highest risk subgroup (e.g., CPPRG, 2011),
suggesting that the impact of exposure to Fast Track might vary across subgroups of
children.
Glucocorticoid Receptor Gene asaModerator
Belsky has advanced the provocative hypothesis that those children who are at
highest risk for adverse developmental outcomes may be the ones who benefit
most from intervention services. This phenomenon has been termed biological
sensitivity to context (Boyce & Ellis, 2005) or differential susceptibility (Belsky,
Bakermans-Kranenburg, & van IJzendoorn, 2007). Highly susceptible children
are hypothesized to be especially responsive to environmental stimuli both for
better and for worse. Under conditions of positive intervention, these children
achieve better outcomes relative to less susceptible peers, but under conditions of
adverse environments, these children experience worse outcomes than less sus-
ceptible peers.
Although initial formulations of this model focused on temperament as a marker
of susceptibility, current research is focused on genetic differences between indi-
viduals (Belsky etal., 2007, 2009; Belsky & Pluess, 2009). The best evidence for
differential susceptibility comes from studies that use randomized controlled trials
to study how behavioral interventions have different consequences for individuals
who carry different variants of certain genes (Bakermans-Kranenburg & van
Ijzendoorn, 2011; Brody et al., 2013; van IJzendoorn et al., 2011). Brody et al.
(2009) tested and supported this hypothesis with the Strong African American
Families intervention, a 7-week, family-based youth risk-behavior prevention pro-
gram for rural African Americans.
As a way to test the hypothesis of differential susceptibility, the Fast Track inves-
tigators focused on the glucocorticoid receptor gene NR3C1, which encodes a pro-
tein critical to human stress physiology. The hypothalamus-pituitary-adrenal (HPA)
axis mediates an individuals response to stress. In the face of a perceived threat,
hormones released from the hypothalamus result in the release of the glucocorticoid
cortisol, the paramount stress hormone involved in energy storage and expenditure,
digestion, and immune function. Importantly, the HPA axis regulates itself by down-
regulating HPA axis activity and allowing the body to return to homeostasis once
the perceived threat is resolved. This inhibitory signaling is mediated by glucocor-
ticoid receptor proteins encoded by the gene NRC31 (DeRijk, van Leeuwen, Klok,
& Zitman, 2008; McEwen, 2012; Meaney, 2001; Sapolsky, Romero, & Munck,
2000). Dysregulated glucocorticoid signaling disrupts HPA axis response to and
244 K.A. Dodge
recovery from a wide range of stressors and has been linked to externalizing psy-
chopathology (Fardet, Petersen, & Nazareth, 2012; Stadler, Poustka, & Sterzer,
2010; van Zuiden etal., 2011). Children exhibiting low cortisol reactivity to experi-
mental challenge respond less favorably than high cortisol-reactive children to an
intervention designed to reduce disruptive behavior (van de Wiel, van Goozen,
Matthys, Snoek, & van Engeland, 2004).
Based on this evidence, Albert etal. (2015a) the Fast Track investigators hypoth-
esized that glucocorticoid receptor variants would differentiate responsiveness to
the Fast Track intervention. Prior research has not clarified which variants in the
gene NR3C1 should be studied, so a gene-wide association study approach (Dick,
2011) was used, which tests variation throughout a gene, with findings then being
subjected to a statistical correction to account for false positives arising from mul-
tiple testing. Ten single-nucleotide polymorphisms (SNPs) identified through a tag-
ging procedure were at least partially independent in the current samples (maximum
linkage disequilibrium as measured by R2 equal to 0.56 (European-American) and
0.53 (African-American).
The outcome of interest was any externalizing psychopathology at age 25.
Analyses were conducted separately for African American and European American
participants, but no evidence was found that any measured SNP moderated inter-
vention effectiveness for African Americans. As shown in Fig. 12.4, among
European Americans, for whom the main effect of intervention was pronounced
(i.e., prevalence of any externalizing psychopathology was 46% in the interven-
tion group and 61% in the control group), the effect of intervention was moder-
ated by variation in the glucocorticoid receptor gene NR3C1. The Fast Track
intervention was more efficacious in preventing externalizing psychopathology
among carriers of the rs10482672 A allele (=0.520, p=0.00006) than among
noncarriers. Among carriers of the A allele, 18% of treated children manifested
any externalizing psychopathology at age 25 follow-up, in contrast with 75% of
control children. In contrast, for noncarriers of the A allele, 56% of treated chil-
dren and 57% of control children manifested externalizing psychopathology at
follow-up. The remaining nine SNPs did not show evidence of moderating Fast
Track intervention response. The pattern of interaction effect held not only for the
Fig. 12.4 The interaction 0.9

between the rs10482672 0.8 Control Group
A allele and the fast track Intervention Group
intervention on 0.7
externalizing 0.6
psychopathology at age 25 0.5
(Albert etal., 2015, 0.4
Journal of Policy Analysis
and Management) 0.3
0.2
0.1
0
Non-Carriers Carriers
broad variable of any externalizing psychopathology but also for specific measures
of antisocial personality disorder, alcohol abuse, marijuana abuse, hard drug use,
and attention deficit/hyperactivity disorder.
ediation oftheInteraction BetweentheGlucocorticoid

M
Receptor Gene andIntervention
The developmental cascade model described earlier suggests that intervention

effects (and gene-by-intervention effects) might begin with earlier stages of inter-
vention in the form of effects on aggressive behavior and conduct disorder in ele-
mentary and middle school and grow into larger effects on externalizing
psychopathology in high school and then even larger effects in adulthood. This
model is depicted in Fig. 12.5. Led by Albert etal. (2015b), the next research steps
in this program were to test the hypotheses that a similar NR3C1 by intervention
effect would hold for aggressive behavior outcomes in childhood and adolescence
and that these moderation effects would mediate the NR3C1 by intervention effect
on adult externalizing psychopathology.
Analyses proceeded in three steps. In Step 1, genetic main effects were tested on
the social adjustment of Fast Track participants in kindergarten, prior to their enroll-
ment in the intervention trial. Children who carried more copies of the rs10482672
A allele did not differ from their peers who carried fewer copies on eight of the ten
Achenbach Child Behavior Checklist scales, although in all cases, the children who
carried two copies of the A allele had the higher scores. Children who carried more
copies of the A allele exhibited significantly more Anxious/Depressed behavior and
more Thought Problems than peers who carried fewer copies.
In Step 2, the interaction between the rs10482672 genotype and intervention was
tested on proximal outcomes of externalizing psychopathology in elementary school
Fig. 12.5 Model of mediation of the gene-by-intervention interaction effect on adult externalizing
psychopathology
246 K.A. Dodge
(grades 36) and high school (grade 7 through 2years following high school). Both
analyses revealed a significant interaction indicating that higher risk children,
genetically, were most responsive to intervention. For each additional copy of the
susceptibility allele a child carried, the Fast Track intervention decreased childhood
externalizing psychopathology by 0.88 standard deviations. There was no effect of
intervention on childhood externalizing psychopathology for children who carried
no copies of the susceptibility allele. For children who carried one susceptibility
allele, intervention decreased externalizing psychopathology by 0.71 standard devi-
ations, and for children who carried two susceptibility alleles, intervention decreased
childhood externalizing psychopathology by 1.59 standard deviations.
Similar findings held for externalizing psychopathology in adolescence. For each
additional copy of the susceptibility allele a child carried, intervention decreased
adolescent externalizing scores by 1.33 standard deviations. There was no interven-
tion effect for children who carried no copies of the susceptibility allele. For chil-
dren who carried one susceptibility allele, intervention decreased adolescent
problem behavior by 1.14 standard deviations, and for children who carried two
susceptibility alleles, intervention decreased adolescent problem behavior by 2.47
standard deviations.
Finally, the cascade hypothesis was tested that gene-by-intervention effects on
adult outcomes would be mediated by effects on childhood and adolescent external-
izing problems. Both childhood and adolescent scores were statistically significant
mediators of the gene-by-intervention effect on age-25 externalizing psychopathol-
ogy, accounting accounted for 16% and 49% of the total effect, respectively.
Furthermore, a portion of the gene-by-intervention effect on adolescent externalizing
scores was independent of any effect on the childhood scores, and, in turn, the ado-
lescent externalizing scores accounted for 40% of the total gene-by-intervention
effect on age 25 externalizing psychopathology even after controlling for effects on
childhood externalizing scores.
Summary, Conclusions, andPossible Next Steps
The findings from the Fast Track experiment shed light on the developmental behav-
ioral mechanisms through which an important interaction between a polymorphism
in the NR3C1 gene and responsiveness to intervention might operate. Children who
carried risk alleles were highly responsive to intervention. That is, in the absence of
intervention they displayed high levels of externalizing psychology, whereas when
exposed to an intervention designed to teach them social-cognitive skills they dis-
played relatively low levels of externalizing psychopathology. Children who were
not carriers did not respond to intervention. This interaction effect begins with the
initial response to intervention in elementary, and it cascades across development
into a larger effect in adolescence and ultimately a very strong effect by adulthood.
The large impact of an intervention that focuses explicitly on social-cognitive
skills on children who are carriers of a risk allele in the NR3C1 gene suggests that
this gene operates on brain mechanisms related to social-cognitive functioning,

such as cognitive responses to stress or threat. Furthermore, the impact of the
NR3C1 gene on intervention impact suggests that externalizing psychopathology
develops at least partly as a function of processes that are influenced by this gene.
The findings reported here offer support for the general hypothesis that genetic
factors will moderate intervention effects in theoretically sensible ways. Although
the findings could suggest using genetic information to identify individual children
who are most likely to benefit from an intervention, this step is premature. Several
research steps are needed first, as follows.
First, the empirical finding of an interaction between NR3C1 and intervention
needs replication. The field of gene-by-environment interaction findings has led to
exciting new theories, but some of the excitement has been dashed by failures to
replicate. Although it is unlikely that the two-decade-long Fast Track trial will be
duplicated, similar interventions that target childrens social-cognitive skills could
test whether the intervention effect is moderated by the NR3C1 gene. In fact, noth-
ing about the theory of gene-by-intervention effects is peculiar to the Fast Track
intervention per se; replication with a different intervention that also targets social-
cognitive processes would provide broad support for the general hypothesis. Such
cross-study replications could be fostered by a network of intervention studies that
have genetic information available.
Second, the generality and specificity of the impact of the NR3C1 gene needs to
be identified by examining the interaction between this gene and other outcomes of
the Fast Track intervention. In elementary school, this intervention had a positive
impact on academic skills, social-cognitive skills, and peer relations success. In
young adulthood, the main effects of this intervention spread beyond externalizing
disorders to positive outcomes in health and well-being, reductions in anxiety and
substance use, and even improvements in civic participation through voting behav-
ior (CPPRG, 2015). It is not clear whether the NR3C1 gene will moderate all of
these intervention effects. Identifying the limits of this interaction effect will con-
tribute to understanding how this gene operates.
Third, it is important to identify the mediating factors through which the NR3C1-
by-intervention effect operates. A hypothesis of a social-cognitive skills pathway is
offered in this chapter, but this mediation hypothesis must be tested directly.
Numerous social-cognitive processes could be implicated, including emotion rec-
ognition skills, executive function, self-regulation, problem-solving, and decision-
making. Identification of the specific pathway will inform an understanding of the
more direct action of this gene. Other mediating hypotheses are plausible and could
be tested as well, such as a general stress reduction pathway.
The research described here could open the examination of other genes that
might also moderate intervention effects, such as those described earlier in this
chapter (monoamine oxidase A and gamma-aminobutyric acid receptor subunit
alpha-2. As the field of molecular behavioral genetics gets more sophisticated, it is
being recognized that single genes are unlikely to account for large portions of the
variance in any behavior, either as a main effect or in interaction with intervention.
A modified approach is to use genome-wide assessments to identify a polygenic
248 K.A. Dodge
risk score that accounts for larger portions of variance. Such an approach has not
been attempted with a gene-by-intervention interaction but is ripe for doing so.
These new research areas await the next decade. Perhaps the most important
contribution of this work might well be to advance the concept of using an interven-
tion experiment to learn about processes through which genes exert impact on psy-
chopathology outcomes.
References
Achenbach, T.M. (1991b). Manual for the Teachers Report Form and 1991 Profile. Burlington,
VT: Department of Psychiatry, University of Vermont.
Albert, D., Belsky, D.W., Crowley, D.M., Conduct Problems Prevention Research Group, Bates,
J.E., Pettit, G.S., Dodge, K.A. (2015a). Developmental mediation of genetic variation in
response to the Fast Track Prevention Program. Development and Psychopathology, 27(1),
8195.
Albert, D., Belsky, D.W., Crowley, D.M., Latendresse, S.J., Aliev, F., Riley, B., Dodge, K.A.
(2015b). Can genetics predict response to complex behavioral interventions? Evidence from a
genetic analysis of the Fast Track randomized control trial. Journal of Policy Analysis and
Management, 34(3), 497518.
Bakermans-Kranenburg, M.J., & van Ijzendoorn, M.H. (2011). Differential susceptibility to rear-
ing environment depending on dopamine-related genes: new evidence and a meta-analysis.
Belsky, J., & Pluess, M. (2009). Beyond diathesis stress: Differential susceptibility to environmen-
tal influences. Psychological Bulletin, 135, 885908.
Belsky, J., Bakermans-Kranenburg, M. J., & van IJzendoorn, M. H. (2007). For better and for
Belsky, J., Jonassaint, C., Pluess, M., Stanton, M., Brummett, B., & Williams, R. (2009).
Vulnerability genes or plasticity genes? Molecular Psychiatry, 14, 746754.
Boyce, W. T., & Ellis, B. J. (2005). Biological sensitivity to context: I. An evolutionary-
developmental theory of the origins and functions of stress reactivity. Development and
Brody, G.H., Beach, S.R., Philibert, R.A., Chen, Y.F., & Murry, V.M. (2009). Prevention effects
moderate the association of 5-HTTLPR and youth risk behavior initiation: gene x environment
hypotheses tested via a randomized prevention design. Child Development, 80, 645661.
Brody, G. H., Chen, Y. F., & Beach, S. R. (2013). Differential susceptibility to prevention:
54, 863871.
Conduct Problems Prevention Research Group (1999). Initial impact of the Fast Track Prevention
Trial for Conduct Problems: I. The high-risk sample. Journal of Consulting and Clinical
Conduct Problems Prevention Research Group (2002). Using the Fast Track Randomized
Prevention Trial to test the early-starter model of the development of serious conduct problems.
Conduct Problems Prevention Research Group (2004). The effects of the Fast Track Program on
serious problem outcomes at the end of elementary school. Journal of Clinical Child and
Adolescent Psychology, 33, 650661.
Conduct Problems Prevention Research Group (2007). Fast Track randomized controlled trial to
prevent externalizing psychiatric disorders: Findings from grades 3 to 9. Journal of the
American Academy of Child and Adolescent Psychiatry, 46, 12501262.
Conduct Problems Prevention Research Group (2010). The difficulty of maintaining positive inter-
vention effects: A look at disruptive behavior, deviant peer relations, and social skills during the
middle school years. Journal of Early Adolescence, 30, 593624.
Conduct Problems Prevention Research Group (2011). The effects of the Fast Track preventive
intervention on the development of conduct disorder across childhood. Child Development, 82,
331345.
Conduct Problems Prevention Research Group (2015). Impact of early intervention on psychopa-
thology, crime, and well-being at age 25. American Journal of Psychiatry, 172, 5970.
DeRijk, R.H., van Leeuwen, N., Klok, M.D., & Zitman, F.G. (2008). Corticosteroid receptor-
gene variants: Modulators of the stress-response and implications for mental health. European
Journal of Pharmacology, 585, 492501.
Dick, D.M. (2011). Gene-environment interaction in psychological traits and disorders. Annual
review of clinical psychology, 7, 383.
Dick, D.M., Bierut, L., Hinrichs, A.L., Fox, L., Bucholz, K.K., Kramer, J., Foroud, T. (2006).
The role of GABRA2 in risk for conduct disorder and alcohol and drug dependence across
developmental stages. Behavior Genetics., 36(4), 577590.
Dodge, K.A. (2006). Translational science in action: Hostile attributional style and the develop-
ment of aggressive behavior problems. Development and Psychopathology, 18, 791814.
Dodge, K. A., & Pettit, G. S. (2003). A biopsychosocial model of the development of chronic
conduct problems in adolescence. Developmental Psychology, 39, 349371.
Dodge, K.A., Bates, J.E., & Pettit, G.S. (1990). Mechanisms in the cycle of violence. Science,
250, 16781683.
Dodge, K.A., Greenberg, M.T., Malone, P.S., & Conduct Problems Prevention Research Group
(2008). Testing an idealized dynamic cascade model of the development of serious violence in
adolescence. Child Development, 79, 19071927.
Dodge, K.A., Lansford, J.E., Burks, V.S., Bates, J.E., Pettit, G.S., Fontaine, R., & Price, J.M.
(2003). Peer rejection and social information-processing factors in the development of aggres-
sive behavior problems in children. Child Development, 74, 374393.
Dodge, K.A., Malone, P.S., Lansford, J.E., Miller, S., Pettit, G.S., & Bates, J.E. (2009). A dynamic
cascade model of the development of substance-use onset. Monographs of the Society for
Research in Child Development., 74, Serial No. 294.
Dodge, K.A., Pettit, G.S., Bates, J.E., & Valente, E. (1995). Social information-processing pat-
terns partially mediate the effect of early physical abuse on later conduct problems. Journal of
Abnormal Psychology, 104, 632643.
Durlak, J.A., Weissberg, R.P., Dymnicki, A.P., Taylor, R.D., & Schellinger, K.B. (2011). The
impact of enhancing students social and emotional learning: A meta-analysis of school-based
universal interventions. Child Development, 82, 405432.
Edenberg, H.J., Dick, D.M., Xuei, X., Tian, H., Almasy, L., Bauer, L.O., Begleiter, H. (2004).
Variations in GABRA2, encoding the alpha 2 subunit of the GABA(A) receptor, are associated
with alcohol dependence and with brain oscillations. American Journal of Human Genetics,
74(4), 705714.
Edwards, A.C., Dodge, K.A., Latendresse, S.J., Lansford, J.E., Bates, J.E., Pettit, G.S., Dick,
D.M. (2010). MAOA uVNTR and early physical discipline interact to influence delinquent
behavior. Journal of Child Psychology and Psychiatry, 51, 679687.
Fardet, L., Petersen, I., & Nazareth, I. (2012). Suicidal behavior and severe neuropsychiatric dis-
orders following glucocorticoid therapy in primary care. American Journal of Psychiatry, 169,
491497.
Guerra, N.G., & Slaby, R.G. (1990). Cognitive mediators of aggression in adolescent offenders:
2. Intervention. Developmental Psychology, 26, 269277.
Hudley, C., & Graham, S. (1993). An attributional intervention to reduce peer-directed aggression
among African American boys. Child Development, 64, 124138.
250 K.A. Dodge
Jones, S.M., Brown, J.L., & Aber, J.L. (2011). Two-year impacts of a universal school-based
social-emotional and literacy intervention: An experiment in translational developmental
research. Child Development, 82, 533554.
Lochman, J.E., & the Conduct Problems Prevention Research Group (1995). Screening of child
behavior problems for prevention programs at school entry. Journal of Consulting and Clinical
Lochman, J.E., & Wells, K.C. (2004). The Coping Power Program for preadolescent boys and
their parents: Outcome effects at the 1-year follow-up. Journal of Consulting and Clinical
Psychology, 72(4), 571578.
McEwen, B. S. (2012). Brain on stress: How the social environment gets under the skin.
Proceedings of the National Academy of Sciences of the United States of America, 109,
1718017185.
ences in stress reactivity across generations. Annual Review of Neuroscience, 24, 11611192.
Pettit, G.S., Lansford, J.E., Malone, P.S., Dodge, K.A., & Bates, J.E. (2010). Domain specificity
in relationship history, social-information processing, and violent behavior in early adulthood.
Journal of Personality and Social Psychology, 98, 190200.
Sapolsky, R.M., Romero, L.M., & Munck, A.U. (2000). How do glucocorticoids influence stress
responses? Integrating permissive, suppressive, stimulatory, and preparative actions. Endocrine
Reviews, 21, 5589.
Stadler, C., Poustka, F., & Sterzer, P. (2010). The heterogeneity of disruptive behavior disorders
implications for neurobiological research and treatment. Frontiers in Psychiatry, 1, 114.
Van IJzendoorn, M.H., Bakermans-Kranenburg, M.J., Belsky, J., Beach, S., Brody, G., Dodge,
K.A., Scott, S. (2011). Gene-by-environment experiments: a new approach to finding the
missing heritability. Nature Reviews Genetics, 12, 881.
van de Wiel, N. M., van Goozen, S. H., Matthys, W., Snoek, H., & van Engeland, H. (2004).
Cortisol and treatment effect in children with disruptive behavior disorders: a preliminary
study. Journal of the American Academy of Child and Adolescent Psychiatry, 43, 10111018.
Wilson, S.J., & Lipsey, M.W. (2007). School-based interventions for aggressive and disruptive
behavior: Update of a meta-analysis. American Journal of Preventive Medicine, 33(Suppl. 2S),
130143.
van Zuiden, M., Geuze, E., Willemen, H.L. D.M., Vermetten, E., Maas, M., Heijnen, C.J., &
Kavelaars, A. (2011). Pre-existing high glucocorticoid receptor number predicting develop-
ment of posttraumatic stress symptoms after military deployment. American Journal of
Chapter 13
Applying Behavioral Genetics Research
toInform thePrevention ofDevelopmental
Psychopathology: Drawing fromthePrinciples
ofPrevention Science
LeslieD.Leve
Beginning with the successful sequencing of the human genome in 2003 (Collins,
Morgan, & Patrinos, 2003), the idea that we could leverage personalized approaches
to medicine has sparked interest from researchers, medical providers, and patients
alike. More than a decade later, excitement about personalized medicine approaches
remains high, prompted recently by President Obamas 2015 State of the Union
address in which he discussed the promise of tailored medical treatments and a
precision medicine approach (https://www.whitehouse.gov/the-press-office/2015/
01/20/remarks-president-state-union-address-january-20-2015). Launched with a
$215 million investment, Obamas Precision Medicine Initiative aims to develop a
new model of research and intervention that will enable clinicians to apply individu-
alized treatments based on a patients biology. Historically, most western treatments
and interventions have been designed using a one-size-fits-all approach. However,
differences in a persons genetic makeup may lead some individuals to benefit more
than others from the standard intervention approach for a given condition. The long-
term goal of a precision medicine approach is to help clinicians better understand the
complex mechanisms underlying a patients health, disease, or condition and recom-
mend treatments that will be most effective through knowledge about that individu-
als genetic makeup. One aim of the Precision Medicine Initiative is the development
of a national cohort study of one million or more people to propel the understanding
of health and disease and further the development of tailored interventions (Ashley,
2015). Thus, time is of the essence to understand the relevance of genetics for the
development of interventions that target physical health conditions and also to
examine whether the field is ready to apply a personalized medicine approach to
behavioral health outcomes that play a role in developmental psychopathology. This
L.D. Leve, Ph.D. (*)

University of Oregon, 5251 College of Education, Eugene, OR 97403, USA
e-mail: leve@uoregon.edu

DOI10.1007/978-3-319-49227-8_13
252 L.D. Leve
Fig. 13.1. Ten standards Standards of Evidence in Prevention Science Research

drawn from the discipline 1. Guiding Conceptual Theory
of prevention science that
2. Goal of Health Promotion
can help guide effective
translation of knowledge 3. Attention to Mean-Level Improvements
from behavioral genetics to 4. Identification of Mediators
prevention to facilitate a
5. A Process Orientation
precision medicine
approach 6. Attention to the Risk Distribution of the Population
7. Goal of Sustainability and Dissemination
8. Inclusion of Long-Term Follow-Up Data
9. Findings Reported for All Hypothesized Outcomes
10. Replication of Effects
chapter examines the role that findings from behavioral genetics research can play in
facilitating a personalized medicine approach in the prevention of behavioral health
problems. Ten core standards in prevention science are used as the framework for
evaluating the readiness of the field to proceed with a personalized medicine approach
to the prevention of psychopathology (see Fig. 13.1).
The promise of the precision medicine approach stems from successful examples
in the medical field for which preventive interventions are already available for
individuals with specific genetic variants. For example, BRCA mutations predispose
individuals to breast and ovarian cancer; other examples include lynch syndrome,
which is associated with increased risk of hereditary nonpolyposis colorectal can-
cer, and familial hypercholesterolemia (Khoury & Evans, 2015). If genetic varia-
tions such as these are identified in a person through genetic testing, evidence-based
interventions can be delivered to help reduce the risk for associated adverse health
outcomes. The vast majority of dialog about precision medicine to date has focused
on physical health conditions, such as cancer or cardiovascular disease, and has
used molecular genetics approaches to identify the relevant gene variants and muta-
tions. Less often discussed are psychosocial problems that plague society, such as
depression, drug addiction, criminality, or psychosis.
A behavioral genetics approach has yet to be part of the precision medicine dia-
log. Can we improve the success rates of evidence-based psychosocial interventions
by identifying a persons inherited behaviors, inferred through behavioral genetics
approaches that quantify behavioral similarities among family members? A struc-
ture and process for addressing this question is the focus of this chapter.
If we want to use precision medicine to prevent developmental psychopathology,
we first need to understand the core foundations of prevention science. In the sec-
tions that follow, ten core standards that guide prevention science research are
described (see Fig. 13.1). With each standard, an example from prevention science
is provided to illustrate the standard and how it is applied in prevention research.
Then, the translation of each standard to behavioral genetics research is discussed.
Findings from behavioral genetics research that are most readily translatable to each
prevention science standard and therefore are most relevant to pursuit of a precision
13 Applying Behavioral Genetics Research toInform thePrevention 253
medicine approach are described. In addition, gaps in the knowledge base are high-
lighted, and each section describes necessary next steps to facilitate the translation
of behavioral genetics research findings to successfully inform a precision medicine
approach to the prevention of developmental psychopathology.
The ten standards described in this chapter are drawn from the Society for
Prevention Researchs standards of evidence for efficacy, effectiveness, and scale-up
research in prevention science (Gottfredson etal., 2015). Although the Gottfredson
etal. document describes additional prevention science standards, the ten standards
discussed here are the most immediately relevant to the translational approach pro-
posed and represent design attributes that are necessary to attend to when consider-
ing genetic research in terms of preventive interventions to reduce psychopathology.
The genetics research findings discussed in this chapter come solely from behavioral
genetics studies (e.g., research using twin, extended family, or adoption samples; see
Reiss chapter and Neiderhiser chapter in this volume for an overview of these meth-
ods). Other chapters provide examples from molecular genetics studies and their
application to prevention (see Dodge chapter and Brody chapter, this volume). As is
detailed in the sections that follow, relative to other genetic approaches, a behavioral
genetics translational approach may align most closely with the aims of prevention
science and a focus on developmental psychopathology and may be more readily
adapted to community service settings where practitioners often use family-based,
process-oriented interventions in the absence of medical (DNA) data about their cli-
ents. A brief discussion of the benefits and limitations of molecular genetics transla-
tional approaches follows the review of the ten prevention science principles. First,
we turn our attention to the ten prevention science principles and provide examples
from prevention science and from behavioral genetics that pertain to each principle.
Prevention Science Principle 1: Guiding Conceptual Theory
Grounding an intervention in a theoretical model is a highly relevant, integrative

principle of prevention science when one is considering a precision medicine
approach to developmental psychopathology. If we do not base precision medicine
approaches on a genetics- or biology-guided theory of action, we are not likely to
identify heritable mechanisms that are associated with the success of an interven-
tion. In accordance with prevention science standards, a causal theory of the inter-
vention should be tested (Gottfredson etal., 2015, SPR Standard for Efficacy, 4)
and the core components of the intervention and the theory relating these com-
ponents to the outcomes must be identified and described (Gottfredson etal., 2015,
SPR Standards for Efficacy, 2.d). The theory should include both an action theory
describing how the intervention will affect the mechanisms and a conceptual theory
describing how the mechanisms are related to the hypothesized outcomes.
Numerous examples from the field of prevention science posit a mechanistic
theory and then test it within the context of an intervention. One example, Pattersons
coercion theory (Patterson, 1982; Patterson, Reid, & Dishion, 1992), describes a
254 L.D. Leve
dynamic process of mutual reinforcement whereby parents inadvertently reinforce

their childrens negative behaviors, which in turn elicits parent negativity, which
then exacerbates childrens negative behaviors, and so on, escalating the negative
interactions until one member wins and the other gives up or loses. When coer-
cive interactions are the predominant interaction style within the family, children
develop conduct problems that sustain throughout development (Granic & Patterson,
2006). Further, the coercive interaction patterns that children learn at home general-
ize to the peer setting, and children affiliate with peers who reinforce one another
for their behavior (Dishion, Patterson, Stoolmiller, & Skinner, 1991). In preventive
interventions informed by coercion theory, the goal of the intervention is thus to
reduce coercive parenting practices and deviant peer affiliation (the causal mecha-
nisms) to prevent conduct problems and related behavior problems (e.g., substance
use, sexual risk taking, or school failure).
Coercion theory has served as the basis for a large number of intervention stud-
ies. For example, in a study of families undergoing divorce, Forgatch and colleagues
found that a coercion theory-based intervention produced positive effects on coer-
cive discipline and positive parenting, which then led to decreased child noncompli-
ance (Martinez & Forgatch, 2001). In a 9-year follow-up study, reductions in child
delinquency (teacher-reported delinquency and police arrests) were observed for
these youths. Consistent with coercion theory, improved parenting practices and
reduced affiliation with deviant peers were causal mechanisms that were affected by
the intervention and were subsequently associated with the long-term delinquency
reductions (Forgatch, Patterson, DeGarmo, & Beldavs, 2009). These empirical
studies illustrate a core principle in prevention science: testing a theory of causal
mechanisms through the use of a randomized intervention study.
Although the examples provided here describe positive intervention effects, we
know that even the most successful psychosocial interventions do not improve out-
comes for 100% of the recipients. One reason for the variable success rates might
be that individuals have different genetic predispositions that affect their receptivity
to environmental interventions, such as positive reinforcement or positive peer affil-
iations. Turning to behavioral genetics research, we therefore need to search for
studies that are grounded in a theory-based approach and that might provide insight
into why environmentally based psychosocial interventions work for some but not
for all. Several theories of geneenvironment interaction (GE) have been pro-
posed and tested and may prove useful in translating behavioral genetics work into
preventive intervention contexts related to this prevention science standard (see
Reiss, Leve, & Neiderhiser, 2013, for a description of four theories of GE). In
recent years, two of the more widely described genetically relevant theories are the
differential susceptibility hypothesis (Belsky & Pluess, 2009) and the biological
sensitivity to context theory (Ellis & Boyce, 2008). These theories are related, and
according to both, the same behavior or trait can be a liability in some environments
but an asset in other environments. That is, individuals with a certain genetic profile
might be more responsive to the social environment than are those with a different
genetic profile and may draw strength from beneficial social contexts but also
become more afflicted by adverse contexts. According to these theories, from an
evolutionary perspective, variation in environmental response reactions may be an

adaptive mechanism intended to increase the capacity of children to match their
response profiles to the environments to which they are exposed.
Empirical evidence from behavioral genetics studies and from developmental
studies supports the theories of differential susceptibility and biological sensitivity to
context. For example, relations between the level of childhood support and adversity
and the magnitude of stress reactivity were found to be curvilinear in one set of stud-
ies, with childrens stress reactivity being highest in high support and in high adver-
sity contexts, and lowest in more neutral contexts (Ellis, Essex, & Boyce, 2005).
Similarly, Belsky and Pluess (2009) reviewed a range of temperament studies and
genetic studies and suggest that, overall, the pattern of findings indicated that some
individuals appeared to be more susceptible to the adverse effects of unsupportive
environments yet also received more benefit from supportive environments. A behav-
ioral genetics example from a study examining posttraumatic stress disorder (PTSD)
and trauma exposure in twins found that genetic factors can influence the risk of
exposure to some forms of trauma, possibly through individual differences in person-
ality that influence choices (Stein, Jang, Taylor, Vernon, & Livesley, 2014). As such,
differential susceptibility processes may make individuals with a certain genetic pro-
file more susceptible to the effects of trauma (and potentially, more susceptible to the
effects of an environmental intervention, although this was not directly tested). The
findings from this type of theory-guided behavioral genetics study could be trans-
lated to intervention studies aimed at preventing PTSD by helping identify geneti-
cally influenced personality traits that might make a person more (or less) susceptible
to maladaptive and adaptive environments. Clinicians could then identify individuals
who might be most at risk following trauma and might also benefit most from a
therapeutic environmental intervention.
Additional theory-guided behavioral genetics work is needed to postulate which
genetically influenced characteristics are most likely to affect which specific out-
comes for a given individual as a function of exposure to enriched and to impover-
ished environments. This theory-guided work would describe why a genetically
influenced characteristic is hypothesized to lead to differential susceptibility for a
given outcome and would specify the kinds of environmental qualities necessary to
be considered enriched or impoverished. Then, a sampling approach is needed
that comprises equal numbers of individuals in each of the target categories (high or
low level of the genetically influenced trait; enriched or adverse environment).
Finally, the behavioral genetics study would include a test of whether individuals
with the hypothesized genetically influenced susceptibility trait show a greater
range of reactions in both improvised and enriched environments for the hypothe-
sized outcome. Further, a rigorous theory-guided mechanistic test would also show
that such differential susceptibility effects are specific to the hypothesized outcome
and not present for genetically influenced traits that are not part of the theoretical
model. In other words, a strong test of the theory would examine the specificity of
environmental effects on the outcome guided by the theoretical model. These
theory-guided steps are readily accomplished within existing twin, adoption, and
extended family designs, but they have not yet been attempted. Generating addi-
256 L.D. Leve
tional knowledge in this area by analyzing existing twin and adoption data could
facilitate the translation of this work to inform the adaption of, or selective applica-
tion of, a specific environmental intervention for a given genetically influenced out-
come and move us closer to implementing a precision medicine treatment approach
to psychopathology.
Prevention Science Principle 2: Goal ofHealth Promotion
The primary goal of prevention science is to conduct work to improve public

health, including the identification of protective factors and processes that lead to
positive health behaviors and outcomes (Society for Prevention Research, 2011).
Specifically, it is desirable to collect data outcomes that have clear public health
impact (Gottfredson et al., 2015, SPR Standard for Efficacy, 3.c [D]). This core
principle can be seen in the component parts of most evidence-based interventions
that rely on a strengths-based approach to behavior change, with the goal of increas-
ing an individuals adaptive functioning. One example of the health promotion focus
in prevention science is the Promoting Alternative Thinking Strategies (PATHS)
intervention, which was designed to facilitate the development of self-control, emo-
tional understanding, positive self-esteem, relationships, and interpersonal problem-
solving skills (Greenberg, Kusche, Cook, & Quamma, 1995). In PATHS, these goals
are accomplished through lessons taught in the classroom two to three times per
week that facilitate the dynamic relationship between cognitive-affective understand-
ing and real-life situations. The lessons include daily activities that promote the gen-
eralization of skills learned and support for ongoing positive behavior. Studies during
the past 30 years have documented how the use of PATHS has successfully increased
protective factors and positive outcomes for children, including increased self-con-
trol, understanding and recognition of emotions, effective conflict-resolution strat-
egies, reading skills, executive function, cognitive flexibility, ability to develop
effective alternative solutions, ability to tolerate frustration, and ability to plan ahead
to solve complex tasks (Conduct Problems Prevention Research Group, 2010; http://
www.prevention.psu.edu/projects/PATHSFindings.html). Although PATHS specifi-
cally emphasizes child skill building, another core health promotion focus is the
improvement of effective parenting practices, similar to that described earlier in the
coercion theory example in Prevention Science Principle 1. The prevention science
goal of health promotion is achieved in prevention research by applying theories
about developmental processes that promote well-being and targeting these pro-
cesses in the context of a randomized intervention trial.
While considering the core prevention science value of health promotion in
terms of genetic research, it is helpful to identify behavioral genetics studies that
focus on protective processes and positive adjustment outcomes, just as the PATHS
prevention study did. Across disciplines, a significant challenge is the tendency of
researchers to focus on negative processes that lead to maladaptive outcomes, for
example, child maltreatment leading to behavioral problems. This same emphasis
on maladjustment has permeated genetic research on psychosocial processes and

outcomes, and the behavioral genetics field has primarily conducted research on
negative environments and maladaptive outcomes (Plomin, DeFries, Knopik, &
Neiderhiser, 2013). For example, the first adoption studies that examined G E
processes focused on the development of alcoholism as a function of alcoholism in
the biological parents (Cadoret, Cain, & Grove, 1980) and psychiatric disturbance
and divorce in the adoptive parents (Cadoret, Troughton, OGorman, & Heywood,
1986). Very recent adoption studies of adult adoptees also focus on alcohol use
disorder outcomes as a function of alcohol use disorder in biological and adoptive
family members (Kendler etal., 2015). Thus, psychopathology, rather than adaptive
behaviors, has been the core concern. The behavioral genetics literature is also
replete with examples from twin studies about alcohol use and negative correlates,
such as gambling (Slutske, Ellingson, Richmond-Rakerd, Zhu, & Martin, 2013) or
maltreatment (Young-Wolff, Kendler, Ericson, & Prescott, 2011). These studies are
indeed crucial to further understand the etiology of alcoholism and other problems
so findings can be translated to prevention research. Nevertheless, it is equally
important to locate genetic studies that have identified positive characteristics so
that we can target mechanisms of health promotion that can be taught or enhanced
within the context of an intervention.
A recent example of the focus on health promotion mechanisms in the adoption
literature is a study that examined the development of social competence in young
children. Positive attributes of the birth parent and the adoptive-family rearing envi-
ronment were examined as predictors of childrens social competence (Van Ryzin
et al., 2015). The putative mechanisms (environmental and genetic) and the out-
come (child social competence) all involved a positive characteristic. A significant
interaction was found between birth parent sociability and responsive adoptive par-
enting when predicting child social competence at school entry. An analysis of the
interaction revealed the buffering role of genetic strengths in the development of
childrens social competence. Twin studies have also examined geneenvironment
processes related to social competence in young children. For example, Roisman
and Fraley (2012) examined the relationship between early parental support and
childrens social competence in kindergarten. They found that the shared environ-
ment (environmental influences that lead to similarities between children in a fam-
ily) accounted for approximately half of the association between parenting and
social competence, suggesting the potential buffering role of early parental support
in increasing childrens social competence. Together, these and related behavioral
genetics studies that focus on positive environmental mechanisms, protective
genetic influences, and positive outcomes can be more readily translated into behav-
iors that preventionists can identify and promote during the course of an interven-
tion, with an eye toward health promotion as opposed to a focus on psychopathology.
For example, the Roisman and Fraley (2012) study suggests that preventive
interventions could have a positive impact on childrens competence by targeting
parental support. Additional genetically informed studies are needed that include an
environmental mechanism that is health promotion oriented and known to be asso-
ciated with outcomes, even when genetic influences are controlled. These studies
258 L.D. Leve
could then suggest genetically relevant mediating mechanisms that are associated
with well-being. Accumulating knowledge in this area will provide a wider menu of
genetically influenced positive characteristics and positive contextual environments
that could be targeted in prevention interventions to advance a precision medicine
approach.
revention Science Principle 3: Attention toMean-Level

P
Improvements
A primary goal of preventive interventions is to improve outcomes for those exposed

to a particular intervention. Thus, by design, prevention studies typically focus on
mean-level improvements in behavior as a function of the intervention. Further, any
existing pretest differences between the intervention and control group must be
measured and statistically controlled (Gottfredson etal., 2015, SPR Standard for
Efficacy, 6.d). Given this goal, at a basic level, intervention studies examine whether
there are mean-level differences between the intervention and control groups on an
outcome of interest, once baseline differences are controlled. In contrast, behavioral
genetics studies typically use a correlational approach to examine associations
between individuals and draw conclusions about the relative genetic and environ-
mental contributions to a behavior. The differences in the statistical approaches used
by the two disciplines (mean-level differences between groups vs. correlations
between variables within and between groups) have hampered the translation of
findings from one field to the other. For example, it is not uncommon to hear profes-
sionals in the field confused about how a behavior can be highly heritable, and yet
psychosocial intervention studies can target and improve this same behavior. A case
in point is that executive functioning is known to be heritable (Engelhardt, Briley,
Mann, Harden, & Tucker-Drob, 2015), but it is also clear that it can be improved by
means of environmental interventions (e.g., Tang, Yang, Leve, & Harold, 2012). The
use of different analytical approaches is the root cause of the perceived discrepancy
between the heritability of a behavior and its malleability. Behavioral genetics stud-
ies examine correlations between family members who are related to one another in
varying degrees (e.g., identical twins, fraternal twins, cousins, or adoptees and their
adoptive and biological parents) and infer the proportion of variance in a behavior
that is associated with shared environmental or heritable sources by examining
between-group differences in the correlations. In contrast, intervention studies com-
pare mean-level changes in behavior between two or more groups of participants. A
behavior or trait can simultaneously be highly heritable and yet modifiable at the
mean level through an environmental intervention.
Examination of young childrens literacy skills in an intervention study and in a
behavioral genetics study provides a concrete example of this phenomenon. A study
by Pears etal. (2014) examined effects on young childrens letter naming and sound
fluency skills before and after they participated in an intervention that targets literacy
skills in high-risk children about to enter kindergarten. For both literacy outcomes,
scores for children in the intervention condition improved more than scores for chil-
dren in the control condition (control childrens scores did not improve at all, whereas
intervention childrens scores did). Thus, this mean-level analysis indicated that
young childrens literacy skills can be improved through the well-timed delivery of a
theory-based intervention.
An adoption study that also focused on literacy skills in young children revealed
that correlations between adopted children and their biological mothers and fathers
ranged from .24 to .29 on a standardized test of basic reading skills, which suggests
genetic influences on reading skills. In contrast, the correlations between these same
adopted children and their adoptive mothers and fathers ranged from .03 to .02,
which suggests a lack of shared environmental influences (Leve etal., 2014). At first
glance, these results may seem incongruent with those from the Pears etal. (2014)
intervention study that found that an environmental intervention could improve
childrens literacy scores. When a traditional correlational approach common to
behavioral genetics studies (see Neiderhiser chapter, this volume) was used, adopted
childrens reading scores were more highly correlated with their biological parents
scores than with their adoptive parents scores, suggesting genetic influences on
individual variation in literacy skills.
Examining these same data using the mean-level analysis approach common to
prevention research yields a different picture, however. In the same Leve etal. study
(2014), adopted childrens basic reading scores (means) were significantly different
than their biological parents reading scores but not significantly different than their
adoptive parents scores. In other words, using a mean-level approach, the mean-level
reading scores of the sample of adopted children resembled the sample mean-level
scores for adoptive parents more than they resembled the sample mean-level score
for biological parents. This behavioral genetics finding suggests that the family
environment can influence childrens reading proficiency and that childrens literacy
scores can improve as a function of being reared by parents with higher reading
levels, just as the Pears etal. (2014) study found that a family-based intervention
influenced childrens literacy scores.
Given the different types of hypotheses tested and conclusions drawn when
mean-level versus correlational approaches are used, it would be extremely valuable
for translation to prevention research if data from behavioral genetics studies
emphasized mean-level scores as a function of environmental inputs. In adoption
studies, this would entail examining the extent to which childrens mean-level
scores are similar to adoptive parents mean-level scores (versus their biological
parents scores). In twin and extended-family studies, this includes testing whether
positive environments increase twins mean-level scores across all levels of genetic
relatedness. Although all this can be done within behavioral genetics methodolo-
gies, mean-level changes are seldom highlighted in the published behavioral
genetics literature, and emphasis is instead placed on the percent of variance attrib-
utable to genetic versus environmental factors. Giving additional weight to mean-
level differences in published reports would help bridge the gap between genetic
research and prevention and would help prevention researchers more clearly see the
relevance of behavioral genetics findings to intervention development. Cases in
260 L.D. Leve
which behavioral genetics studies identify both mean-level changes and a geneti-
cally influenced behavior may provide good targets for a precision medicine
approach. In other words, this approach can help identify traits that are both heri-
table and that can be increased or reduced through an environmental intervention
and therefore yield a genetically informed intervention approach.
Prevention Science Principle 4: Identification ofMediators
In conjunction with a guiding conceptual theory (see Prevention Science Principle

1), a related standard in prevention science research is having a clear theory of
causal mechanisms (including identification of mediators) (Gottfredson etal., 2015,
SPR Standard for Efficacy, 2.b). In other words, what proximal behavior is the inter-
vention targeting and attempting to change that will ultimately lead to improved
outcomes? One example of the focus on mediating mechanisms in prevention sci-
ence comes from a foster parent intervention program, Keeping Foster Parents
Trained and Supported (KEEP). The goal of this intervention is to reduce child prob-
lem behaviors through strengthening foster parents skills. In one evaluation of
KEEP, foster parents in the intervention group received 16 weeks of training in
supervision and support in behavior management methods, with a focus on increas-
ing the use of positive reinforcement relative to the amount of discipline they used. A
four-to-one rule was taught and practiced in the foster parent group sessions and was
recommended as part of regular home practice assignments (attempt to reinforce the
child at least four times for every one correction/discipline). The proportion of posi-
tive reinforcement that foster parents used was the hypothesized mediating mecha-
nism. Analyses examined the extent to which the KEEP intervention improved the
proportion of positive reinforcement foster parents used with their children and
assessed whether improvements in the proportion of positive reinforcement were
associated with reductions in child behavior problems. Results supported the role of
positive reinforcement as a mediator of intervention-related improvements in child
behavior problems (Chamberlain etal., 2008).
Translating the mediating mechanism concept to behavioral genetics research,
studies that examine geneenvironment correlation (rGE) may be more readily
translatable to prevention and move us closer to being able to apply a precision
medicine approach to intervention, particularly when parenting-based interventions
are the focus. In behavioral genetics research, evocative rGE refers to a mechanism
that explains the pathway whereby genes affect a childs behavior, which in turn
affects a specific parenting behavior. In other words, it informs us about a mecha-
nism (a child trait or behavior) that links genetic influences with parenting behaviors
(see Neiderhiser chapter, this volume, for additional explanation of evocative rGE).
Studies that examine evocative rGE provide a potential synergy between behavioral
genetics research and prevention studies because they identify a child-based, geneti-
cally influenced mechanism that could be targeted to advance a precision medicine
approach to intervention. The adoption design is well suited to test evocative rGE
by examining whether associations between biological parent characteristics (G)

and adoptive parent characteristics (E) are linked by a specific genetically influ-
enced child behavior.
An example of this approach is an adoption study that examined evocative rGE
in a sample of adopted children followed longitudinally from age 7 to age 12
(OConnor, Deater-Deckard, Fulker, Rutter, & Plomin, 1998). In this study, the
authors examined whether adopted children with a biological family history of anti-
social behavior received more negative parenting from their adoptive parents than
did children without a biological family history of antisocial behavior, which would
suggest an evocative rGE mechanism from biological parent to adopted child to
adoptive parent. Repeated measures analysis of variance indicated that children at
genetic risk were consistently more likely to receive negative parenting from their
adoptive parents than were children not at genetic risk, indicating an evocative rGE
mechanism (OConnor etal., 1998). That is, children who were more genetically at
risk elicited more negative parenting from their adoptive parents than did children
who were not genetically at risk. This approach to examining GE interplay revealed the
role of the child in eliciting negative parenting and provides information about a
potential mechanism that could be targeted in future prevention research. In this
example, two genetically informed intervention opportunities are revealed, one
aimed at reducing child behavior problems (so that negative parenting is not evoked),
and a second aimed at helping parents react less negatively to their childs genetically
influenced negative behaviors.
In a second example from a different adoption study, Harold and colleagues
examined the association between attention deficit hyperactivity disorder (ADHD)
symptoms in biological parents and harsh parenting in adoptive parents when chil-
dren were entering formal schooling (Harold etal., 2013). They found that biologi-
cal mothers ADHD symptoms were associated with adoptive mothers hostile
parenting through disrupted behavior (impulsivity/activation) in the adopted child;
hostile parenting was subsequently associated with an increase in adopted child
ADHD symptoms. As in the OConnor etal. (1998) study, the identification of a
specific mediating mechanism (in this instance, early disrupted child behavior) can
provide prevention researchers with identifiable genetically influenced behaviors
in a causal chain. First, the association between rearing mothers hostility and child
ADHD symptoms suggests that early interventions that target a mothers hostility
toward her child, especially a child with temperamental attributes of impulsivity/
activation that appear to evoke such maternal responses, may strengthen the effi-
cacy of intervention programs aimed at reducing child ADHD symptom levels in
the general population. Second, the evocative pathway identified from child impul-
sivity/behavioral activation to adoptive mother hostility suggests that parents may
also be trained to better identify specific behavioral (and genetically influenced)
attributes in their child early in development that evoke hostile maternal responses
which, in turn, promote continuity of ADHD problems. These approaches might
ultimately help reduce the expression of ADHD symptoms in children in the gen-
eral population at a period in their development (school entry) when such symp-
toms can have long-term effects on multiple developmental domains.
262 L.D. Leve
The work described in this section highlights the benefits of using findings from
evocative rGE behavioral genetics studies to learn about genetically influenced
child behavioral mediators that could be targeted in future interventions to advance
a precision medicine approach. In contrast, the more common approach to studying
geneenvironment interplay is to test GE rather than rGE.GE studies tell us
about moderators of associations, which can provide information about subgroups
who are most affected by a specific environment or by an intervention. Although
such information is important, the genetically influenced mediator approach facili-
tated by evocative rGE studies provides a closer parallel to the mechanism/mediator
approach in prevention research and is an underexplored linkage between preven-
tion science and genetic research.
Prevention Science Principle 5: AProcess Orientation
As is evident from the examples provided in the earlier sections of this chapter, pre-
ventive interventions operate through process mechanisms, whether it be coaching a
mother to be more effective at disciplining and monitoring her child, helping a teen-
ager develop refusal skills to avoid illicit drug use despite pressure from peers to use,
or improving the studentteacher relationship. In each of these examples, dynamic
behavioral interaction processes are targeted in the intervention. These behavioral
processes occur through interactions, such as parentchild, studentteacher, child
peer, husbandwife, or childsibling. One example of the process orientation used in
prevention science is Doziers Attachment and Biobehavioral Catch-up (ABC) inter-
vention. This ten-session intervention was designed to help caregivers facilitate
healthy regulation of their childs behavior and stress responses by teaching caregiv-
ers to be highly responsive to their childs emotions and increase nurturing care and
by promoting attachment security (Dozier, Peloso, Lewis, Laurenceau, & Levine,
2008). Parent coaches provide parent training in the home during weekly 1-h ses-
sions with the caregiver and child. The parent coachs primary role is to provide in
the moment feedback about the parents interactions with the child while observing
the parents behavior and making comments about behaviors that relate to the inter-
vention targets. This dyadic, process-oriented focus has contributed to the ABC
interventions success in strengthening and improving the parentchild attachment
relationship. For example, ABC children participating in a study were more often
secure and less often disorganized in their attachments to caregivers than were the
control children; 32% of the ABC children (vs. 57% of the control children) had a
disorganized attachment to their caregiver and 52% of the ABC children (vs. 33% of
the control children) had a secure attachment approximately 1month after the inter-
vention (Bernard etal., 2012). This example illustrates how prevention researchers
intervene at the dyadic process level to understand the intricate nature of social inter-
actions and work to coach individuals (in this case, caregivers) to modify the nature
of their interactions to ultimately improve adjustment outcomes. Using a process
approach is consistent with the prevention science standard of selecting measures
that are psychometrically sound and including multiple measures and/or sources to
avoid demand characteristics (Gottfredson etal., 2015, SPR Standard for Efficacy,
3.e.[D]), and the standard that measures and outcomes be of practical importance
(Gottfredson etal., 2015, SPR Standard for Effectiveness, 6.a.).
While considering this process-oriented approach in prevention science, it is help-
ful to identify behavioral genetics studies that have also used a process-oriented
approach to apply findings from genetic research to prevention research and increase
the precision of existing interventions. Genetic research has typically lagged behind
developmental and prevention research in this area; the majority of published genetic
research has measured environment at a global level rather than through the use of
observational studies of dyadic interactions. Further, in twin and extended-family
studies, behavioral genetics methodologies typically infer environmental effects by
using twin and family statistical modeling techniques to estimate environmental
effects rather than focus on the measured environment (with notable exceptions, see
Reiss, Neiderhiser, Hetherington, & Plomin, 2000). Thus, it is important to examine
findings from the more rare genetic studies that include process-oriented, observa-
tional paradigms.
The Early Growth and Development Study (EGDS; Leve etal., 2013) offers an
example of a process-oriented approach within an adoption study. In a study of 290
linked sets of adoptive families and birth mothers and 95 linked birth fathers, adop-
tive mothers interacted with their adopted toddler during a cleanup task (Leve etal.,
2009). Mothers were asked to make sure that only the child cleaned up and to remind
him/her as necessary. The videotaped cleanup task interactions were later coded for
structured parenting behaviors, which comprised all types of statements in which a
behavior change was suggested, including questions (e.g., Where does this ring
go?), statements (e.g., Lets put the duck in this box), and directives (e.g., Put all
of the cups in here). A defining feature was that there was an explicit or implicit
behavior change or a specific action desired of the child. Genetic risk was assessed
in terms of birth mother and birth father psychopathology (anxiety, depression, anti-
sociality, and drug use). Results indicated a G E effect such that high levels of
structured parenting were beneficial for children who were at high genetic risk (high
birth parent psychopathology) and predicted fewer child behavior problems.
Conversely, the same type of parenting was harmful for children at low genetic risk
and was associated with more behavior problems. However, parental positive rein-
forcement during the cleanup task was found to be beneficial for all children, regard-
less of their genetic risk level (Leve, Harold, Ge, Neiderhiser, & Patterson, 2010).
Examining GE by using process-oriented measures of the environment pro-
vides readily translatable information about potential environmental processes that
might be important when genetic risk is present, which moves us closer to a precision
medicine approach. For example, an interventionist could show a parent of a child at
high genetic risk specific segments from their videotaped interaction to illustrate
examples of how they provided effective structure for their child (and examples of
when they could have offered additional structure). This connects video coaching
commonly used in prevention scienceto the context within which the genetic find-
ings were obtained. The ability of the EGDS study to examine multiple domains of
264 L.D. Leve
parenting at the behavioral level (structured parenting and positive reinforcement)

provides more specificity to the environmental target. These dyad-level data are sig-
nificantly more specific than a score on a questionnaire or checklist because they
consist of observable dyad-level behaviors that a therapist can target. Further, this
type of process information about GE can be incorporated into existing behavioral
interventions more readily than can, for example, information about a GE interac-
tion when E is inferred from genetic modeling techniques and not measured at all.
Identifying additional process-oriented behavioral genetics findings will help move
the field one step closer to a precision medicine approach to interventions intended
to reduce developmental psychopathology in that the role of genetically influenced
characteristics on harmful and beneficial dyadic processes will be clarified.
revention Science Principle 6: Attention totheRisk

P
Distribution ofthePopulation
Another principle in prevention science that has direct relevance to the translation
of genetic findings to prevention research is prevention sciences focus on tailoring
interventions to the risk level of the target population. Preventive interventions can
be classified as universal (addressing an entire population, such as an entire school,
with programs aimed at preventing problem behaviors from developing), selective
(targeting a subset of the population on the basis of membership in a specific at-risk
group, for example, low-income families, with the goal of preventing the develop-
ment of serious problems), or indicated (aimed at individuals who exhibit early
signs of problems, such as youths who have been in the juvenile justice system, with
the aim of implementing programs to prevent further onset of difficulties) (Mrazek
& Haggerty, 1994; Society for Prevention Research, 2011). One standard in preven-
tion science is that the intervention specifies the population (and time and setting)
that is targeted in the study (Gottfredson etal., 2015, SPR Standard for Efficacy, 1.).
Prevention programs have different goals and intervention strategies as a function of
the type of intervention and population served (universal, selective, indicated), with
more intensive approaches typically used as one moves up the spectrum of risk. For
example, Treatment Foster Care Oregon (an indicated intervention) is an intensive
intervention that includes comprehensive foster care services, such as weekly foster
parent groups, weekly individual therapy, weekly family therapy, and daily phone
check-ins, for approximately 6months (Chamberlain, 2003). In contrast, Positive
Action is a school-based intervention (universal) that includes schoolwide climate
change and a detailed curriculum with lessons two to four times a week, for a total
of approximately 35h of curriculum (Flay & Allred, 2003). Both interventions have
been shown to be effective in reducing problem behavior and appear on many
evidence-based practice lists, even though they differ substantially in terms of the
number of contact hours with members of their respective client populations. This
difference is largely a function of the risk level of the population served, given the
indicated versus universal prevention sample used.
Examination of the Blueprints for Healthy Youth Development website, which

includes a list of evidence-based programs that have been shown to be effective in
reducing problem behaviors in youths and families, reveals that of more than 1300
youth promotion programs reviewed, 14 interventions met the highest standard of
model program as of the writing of this report (http://www.blueprintsprograms.
com/programCriteria.php). Of the 14 model programs, four have been classified as
universal, seven as selective, and seven as indicated (four programs were effective
across two risk levels, and thus the numbers here sum to 18). A similar ratio of univer-
sal/selective/indicated interventions is noted on other best practices websites for top-
tier programs (http://www.nrepp.samhsa.gov/; http://coalition4evidence.org/). Two
important implications can be drawn from these evidence-based practice lists: First,
an array of programs exists across each risk level, and the intervention strategies and
intensity typically vary according to whether the intervention is universal, selective, or
indicated. Second, the majority of the highest tier evidence-based interventions focus
on populations that have some level of risk (selective or indicated samples).
Thus, when applying findings from genetic studies to prevention research, it is
important to match the sample type (universal, selective, indicated) between the
prevention program and the genetic study. If a mismatch occurs, and, for example,
a GE finding from a population-based (universal) sample is applied to an indi-
cated intervention, the E studied in the genetic design may not be well aligned
with the E for the individual in the indicated prevention study. In addition, because
the majority of the highest tier interventions focus on populations that have some
level of risk (selective or indicated), genetic studies that are drawn from risk sam-
ples will be more directly relevant to the majority of prevention studies. To date,
however, most twin studies have focused on population-based samples, which tend
to have lower risk characteristics. Their results might therefore be most applicable
to universal preventive interventions. Similarly, in most adoption studies, the rear-
ing parents are from higher sociodemographic backgrounds and consequently often
represent a lower risk segment of the population. These sample characteristics do
not preclude translation; rather, it is crucial to recognize the risk level of the sample
and match it across the genetic and the prevention study.
Behavioral genetics research on GE findings related to socioeconomic status
illustrates why it is so important to match sample types across genetic and preven-
tion studies when considering applying genetic findings to interventions to further a
precision medicine approach. Turkheimer, Haley, Waldron, DOnofrio, and
Gottesman (2003) examined childrens IQ in a sample of 7-year-old twins from the
National Collaborative Perinatal Project, which included a significant proportion of
families living near or below the poverty level. The modeling results indicated that
the proportion of variance in childrens IQ that was attributable to genes and to the
environment varied with socioeconomic status. In families living in poverty, the
majority of variance in IQ was attributable to the shared environment; genes played
very little role. In contrast, in families who were affluent, genes accounted for the
majority of the variance in childrens IQ, and shared environment played little to no
role. This pattern of effects was replicated in a separate sample of young twins
(Tucker-Drob, Rhemtulla, Harden, Turkheimer, & Fask, 2011). At age 2 years,
266 L.D. Leve
genes accounted for close to half of the variation in mental ability of children raised
in high-SES homes but accounted for negligible variation in mental ability of chil-
dren raised in low-SES homes. A study of adults also found a similar disparity
between the influence of genes on IQ in adults of high-SES backgrounds versus
those in low-SES backgrounds and an amplification of genetic differences in adults
from higher SES backgrounds but not in those from lower SES backgrounds (Bates,
Lewis, & Weiss, 2013).
Considering this genetic work and the role that SES plays in magnifying or
reducing genetic and environmental influences, the design of a genetically informed
preventive intervention study would be very different as a function of whether chil-
dren from impoverished backgrounds were the focus of the intervention or whether
children living in enriched environments were the target population, because the
factors that influence variance in their outcomes differ. In terms of the former, which
might be characteristic of a selective intervention sample, we know that the shared
environment (e.g., specific parenting strategies used family-wide) accounts for a
large portion of the individual variation in IQ, and therefore a focus on parenting
skills and global family environment qualities might be an important avenue for
intervention. Conversely, among children in an enriched environment who might be
enrolled in a universal preventive intervention, targeting specific family-wide char-
acteristics may not help those individuals who are on the lower end of the IQ distri-
bution improve their IQ relative to that of their peers. Knowledge of the extent to
which the environment contributes to variance in a given outcome (e.g., IQ), given
a specific context (e.g., high versus low SES), can help prevention researchers better
align their environmental intervention target to their sample composition, be it a
community sample (universal), low-income sample (selective), or criminal sample
(indicated). Similarly, consideration of effects related to specific sample character-
istics can help genetic researchers frame their results in ways that can be more read-
ily applied to prevention studies.
revention Science Principle 7: Goal ofSustainability

P
andDissemination
Another standard in prevention science focuses on the goal of ensuring that the
intervention is delivered under the same types of conditions as one would expect in
the community institutions where such interventions are most likely to be situated
during scale-up (e.g., by teachers rather than by research staff) (Gottfredson etal.,
2015, SPR Standard for Effectiveness, 5.a). In other words, the research aspects of
the study should be developed in a way that maximizes the ability of the interven-
tion to be readily incorporated into the service sector with the same fidelity and
effects as in the original trial, once the research study has concluded. This preven-
tion science standard is particularly important to consider if we are to move forward
with a precision medicine approach that would be applied in everyday practice set-
tings rather than in controlled research settings.
An example of the importance of this standard in prevention science comes from

one of the most widely disseminated, family-focused behavioral health interven-
tions to date, the NurseFamily Partnership (Olds, 2006). In the original interven-
tion, trained nurses conducted home visits with low-income pregnant mothers from
pregnancy through child age 2 years, with positive long-term effects found on
maternal and child health and behavioral health outcomes (Olds, Henderson, &
Kitzman, 1994; Olds etal., 2010). In the process of dissemination activities, ques-
tion arose about whether similarly positive results could be achieved if paraprofes-
sionals, rather than nurses, were used as the interventionists. If so, this approach
might reduce the cost of implementing the program. To address this question, Olds
and colleagues conducted a new study in which they compared the outcomes of the
prenatal and infant/toddler home visits in a randomized design in which mothers
were either home visited by a paraprofessional or by a nurse or were in a services as
usual control condition (Olds et al., 2014). The results revealed no significant
improvements when paraprofessionals were used (as compared with the control
condition) on childrens emotional/behavioral problems. Conversely, relative to
children in the control condition, nurse-visited mothers and children had fewer total
emotional/behavioral problems at age six, fewer internalizing problems at age nine,
and less dysfunctional attention at age nine. These findings suggest that retaining
the use of a nurse home visitor, as was intended in the original program develop-
ment, is likely to yield similar results to those found in the original studies in terms
of reductions in child behavior problems (and thus is recommended). However, if
the program is modified such that paraprofessional home visitors are used, reduc-
tions in child behavior problems may not be apparent (although other effects may be
present). Understanding the importance of the context in which an intervention will
ultimately be implemented ensures that it is feasible and sustainable, that the key
findings are translatable, and that programs remain effective when implemented in
the future with less research oversight, as would be the case in a precision medicine
application of services in a community health setting.
Building on the principles of sustainability and dissemination established in pre-
vention science, if we use genetic data to inform personalized medicine approaches
to the prevention of psychopathology, we likewise need to draw from studies and
approaches that can be readily applied in real-world intervention service settings,
where such interventions would be likely to be implemented. In the field of devel-
opmental psychopathology, once a program is disseminated outside of the research
context, the typical provider is a community mental health agency or other public or
social service agency. Twin and adoption studies offer many strengths relative to
dissemination and sustainability because they involve the use of genetic data that
many community health practitioners already have about their patient: information
about biological family members. Most clients complete a personal health history
interview and share information with their therapist or community health provider
about the behavioral health of other family members. Thus, the therapist has access
to the type of information used in behavioral genetics studies and could make treat-
ment decisions as a function of this information, while attending to the extant scien-
tific literature on geneenvironment transaction related to this behavior.
268 L.D. Leve
Translating this model to GE findings from an adoption study provides one

example of how such a translation from behavioral genetics research findings to
intervention-ready knowledge could inform a precision medicine approach.
Natsuaki etal. (2010) identified a significant GE interaction between birth moth-
ers major depressive disorder diagnosis (MDD) and adoptive mothers responsive-
ness. In this study, children of birth mothers with MDD showed higher levels of
fussiness in toddlerhood when their adoptive mothers had been less responsive to
them as infants, as compared to infants whose adoptive mothers had been more
responsive. Conversely, when adoptive mothers were highly responsive, toddlers
with birth mothers with MDD did not show elevated fussiness; these babies were
protected from expressing their inherited risk for depression. This finding suggests
that if a practitioner knows the family history of the biological parents, and one of
those parents presents with MDD, promoting a responsive adoptive caregiving envi-
ronment can help buffer their child from the genetic risk for MDD and decrease the
childs distress. In this example of a genetically informed intervention extension of
the research findings, no specific genetic data are needed (unlike genetic studies that
involve collection of DNA samples). Rather, the aggregate effects of genes are
inferred via the adoption design methodology, and that information can then be
translated to guide specific therapeutic treatments that are informed by the thera-
pists knowledge of the psychopathology of biologically related extended family
members. As such, the translation from GE findings in adoption research to pre-
ventive intervention research can proceed more immediately than in other types of
genetic studies, in cases in which findings are robust and prevention experts are
informed about the findings from this literature. A systematic examination of GE
findings that uses adoption and twin methodologies for a specific behavioral out-
come could help identify patterns of findings in the field, such as the Natsuaki etal.
(2010) finding described previously, and move us closer to testing an intervention
approach for specific target behaviors that is informed by an understanding of the
family history of psychopathology for each individual client.
revention Science Principle 8: Inclusion ofLong-term

P
Follow-up Data
Prevention science places a high value on documenting the long-term and sustained
outcomes of an intervention. There must be at least one long-term follow-up at an
appropriate interval beyond the end of the intervention (Gottfredson etal., 2015,
SPR Standard for Efficacy, 3.d). The rationale for this standard is that positive inter-
vention effects may either diminish or increase over time, and longer term follow-up
data help establish the full spectrum of outcomes for a particular intervention. There
is no specific recommended duration of follow-up (although 6months is thought to
be a minimum) because duration should be decided based on the guiding principles
in the theory (see Prevention Science Principle 1). This principle has direct rele-
vance to the goals of precision medicine in terms of seeing sustained positive effects
of an intervention. An excellent example of the utility of long-term follow-up data

comes from the Fast Track project. Fast Track is a preventive intervention program
that includes academic tutoring and lessons in developing social skills and regulat-
ing behaviors, beginning in kindergarten and continuing through high school
(McMahon et al., 1999). The developmental theory that guided the intervention
posits that children who have experienced family and contextual stressors may be ill
prepared to succeed in school, which can lead to behavior problems and school dif-
ficulties in the short term and peer problems and academic failure in the longer term.
Participants in the Fast Track study were followed over time, and at the end of first
grade, children assigned to the intervention condition had higher social, emotional,
and academic skills; better peer relations; fewer conduct problems; less use of spe-
cial education; and were exposed to better parenting (McMahon et al., 1999). A
long-term follow-up of Fast Track participants occurred at age 25; 69% of the par-
ticipants in the control condition displayed at least one externalizing, internalizing,
or substance abuse psychiatric problem, whereas 59% of the participants assigned
to the intervention condition displayed such problems (Dodge et al., 2014). This
pattern of outcomes was hypothesized based on the original guiding theory and
illustrates the benefit that long-term follow-up in prevention studies can have in
documenting developmental pathways and the sustained intervention effects that
may accrue increasing cost savings over time.
With respect to the value of conducting long-term follow-up to genetic studies, it
is important that these studies not only examine concurrent processes and outcomes,
but also test whether GE patterns that are found at one time point also show the
hypothesized associations later in development. For example, if stress and inherited
traits interact to increase the likelihood of depression at a specific time point in
adulthood, theory would suggest that this effect would also be found 2 years later,
because depression is a recurring disorder and the effects of stress can be cumula-
tive. Unfortunately, it is rare to see GE studies that test longitudinal patterns of
GE; however, some studies have done so and report interesting longitudinal pat-
terns. For example, using data from the Minnesota Twin Family Study, Samek etal.
(2015) tested the developmental changes in GE interplay in the concurrent and
prospective associations between parentchild relationship problems and external-
izing behavior at ages 18 and 25. They identified a GE effect at age 18, such that
genetic influences on externalizing behavior were greater when there were more
parentchild relationship problems. However, this GE effect was not present at
age 25 when examined cross-sectionally, nor did parentchild relationship prob-
lems at age 18 moderate genetic influences on externalizing problems at age 25. The
authors concluded that the GE identified in the association between adolescent
parentchild relationship problems and externalizing problems was both proximal
and developmentally limited.
When one is considering whether the field is ready to implement interventions
based on GE findings, this type of result gives pause for concern because some
GEs are fairly transient and not predictive of long-term behavior. The core focus
of preventive interventions is to prevent problems from occurring in the future;
thus, if G E effects disappear naturally over time, how would we know if a
270 L.D. Leve
GE-informed preventive intervention had a sustained impact on an outcome of

interest? In the context of a genetically informed intervention trial, if the GE
interaction effect dissipated over time, even in the control group, it would be dif-
ficult to draw conclusions about the utility of such a precision medicine approach.
Using the example described earlier in this section (Samek et al., 2015), on the
basis of the GE identified in the age 18 data alone, one might theorize that a
reasonable genetically informed intervention approach would be to target parent
child relationship problems in children at genetic risk because they magnified child
externalizing problems in the context of genetic influences. However, because this
effect was neither sustained longitudinally nor found cross-sectionally at age 25, a
parentchild, relationship-based intervention at age 18in which individuals were
selected for the intervention because of genetic risk might not show any long-term
effects. If we are to continue to advance intervention science by incorporating
knowledge from G E research, we need to search the available literature for
GE studies that show not only concurrent effects, but also show lasting develop-
mental effects across the hypothesized follow-up period. This strategy will help
ensure that when we design GE-informed interventions, we are basing the inter-
vention not on an effect that will naturally dissipate over time, but rather on one
that is preventative of later problems. Failure to do so may yield genetically
informed interventions that ultimately do not show appreciable improvement over
interventions that are not genetically informed.
revention Science Principle 9: Findings Reported forAll

P
Hypothesized Outcomes
Most prevention studies assess multiple primary and secondary outcomes. One
guiding principle in prevention science is that results must be reported for every
targeted outcome that has been measured in the efficacy study, regardless of
whether they are positive, nonsignificant, or negative (Gottfredson etal., 2015,
SPR Standard for Efficacy, 7.a). The purpose of this principle is to help prevent
publication bias, that is, publishing only positive intervention results. Adhering to
this principle can be difficult even for those who strive to do so, because most jour-
nals are less inclined to publish articles with null effects. One effective strategy has
been to include all outcomes (positive, nonsignificant, or negative) for a given
construct in a single paper. An example of this approach can be seen in the 12-month
delinquency outcomes of Treatment Foster Care Oregon for girls (Leve,
Chamberlain, & Reid, 2005). In this intervention, delinquency was the primary
targeted outcome. The sample, recruited from the juvenile justice system as adoles-
cents, had an average of nearly 12 prior criminal offenses, and the goal of the
intervention was to prevent ongoing delinquent behavior. The study measured
delinquency in four ways: by means of self-report, caregiver report, official arrest
records, and number of days inlocked settings. Results for all four delinquency
outcomes were presented in an article about the study. Days inlocked settings and
caregiver-reported delinquency showed statistically significant intervention effects

(p<.05). Official arrests trended toward significance (p=.10), and self-reported
delinquency was not significant. Including all measures of the hypothesized out-
come makes possible a more complete understanding of the potency and breadth of
effects of an intervention. It also highlights the limits of an intervention and cir-
cumstances in which effects might not be present even though they had been
hypothesized. Many journals now require that prevention studies be registered in a
clinical trials database, such as clinicaltrials.gov, before they will publish the trial
results. All hypothesized primary and second outcomes must be recorded in the
clinical trials registries, along with the measures for each, before the trial is com-
pleted and data are analyzed.
Difficulty finding a strong publication outlet for nonsignificant effects on some
measures presents a similar set of challenges in the genetic literature. Two excep-
tions are the Nonshared Environment and Adolescent Development Study (NEAD;
Reiss etal., 2000) and the Colorado Adoption Project (DeFries, Plomin, & Fulker,
1994), both of which have produced books that sequentially present findings (sig-
nificant and nonsignificant) across a range of outcomes. The authors of each study
were diligent in presenting findings that were expected, findings that were not
expected, and null findings across multiple measures of each core construct. As
such, readers can gain a more accurate picture of where findings are robust across
measures and where findings may be more idiosyncratic and perhaps not ripe for
translation to prevention without additional research.
For example, the NEAD book describes the outcomes of the longitudinal study
of 720 pairs of same-sex adolescent sibling pairs and their parents. Reiss et al.
(2000) summarize how the investigators initially hypothesized that siblings differ-
ential experiences in their families might explain sibling differences. The NEAD
study design and measures originated in this concept, and the authors sought to
discover important nonshared environmental influences within the family in the
course of the research. However, as the book details, the original hypotheses were
not confirmed, and the authors concluded that the household family environment
was not an important source of nonshared environment for adolescents. Instead,
they identified mechanisms of gene expression and the potential role of family pro-
cesses in these mechanisms. In the course of their hypothesis testing and book
development, the authors redid all primary analyses using the simplest analytic
methods possible, and they present all these findings for all measures in their book.
This is an excellent example of how reporting all findings for a given hypothesis
helps prevent against false positives and gives readers a more definitive sense of the
state of the findings. When considering whether a GE finding is ready for transla-
tion into a preventive intervention, it is important to attend to the prevention science
principle of reporting findings for all hypothesized outcomes and measures and to
carefully assess the genetic studies that support the translation to see if all measures
of the hypothesized outcomes have been described. Failure to do so may lead to
pursuit of a precision medicine idea that is very specific to one measure or one time
point and is not generalizable.
272 L.D. Leve
Prevention Science Principle 10: Replication ofEffects
A final and related prevention science principle discussed in this chapter is replication,
and in particular, replication by an independent researcher/research team. Specifically,
a researcher who is neither a current nor past member of the program developers
team should conduct data collection and analysis (Gottfredson et al., 2015, SPR
Standard for Effectiveness, 5.f, D). Many prevention scientists assert that before an
intervention is ready to be widely disseminated across individuals, groups, or com-
munities, control over the program delivery, data collection, and analysis should be
conducted by an individual who is not affiliated with the original investigative team.
The field of prevention science has yielded a number of positive results from
independent replications. For example, Menting et al. (2013) conducted a meta-
analysis of the effectiveness of the Incredible Years parent-training program on the
prevention of disruptive behavior and promotion of prosocial behavior. They exam-
ined the results from 50 studies that had collected data immediately after the inter-
vention. Overall, the results showed a high degree of replication, which demonstrated
evidence for the effectiveness of the intervention. The mean effect size for disrup-
tive child behavior across informants was d=.27. Of note, some differences were
found by sample risk type for the parent-report data (see Prevention Science
Principle 6), with treatment studies associated with larger effects (d = .50) than
indicated sample studies (d=.20) and selective sample studies (d=.13). In addition,
there was no effect of developer involvement on the effect sizes. At the same time,
an increasing number of independent evaluations for prevention programs have
failed to replicate the original findings, even though they reportedly had met all
criteria for effectiveness (e.g., see Malti, Ribeaud, & Eisner, 2011, for a failed rep-
lication in a Swiss trial of the Triple P universal intervention program, or Lfholm,
Olsson, Sundell, & Hansson, 2009, for a failed replication in a Swedish trial of an
indicated sample receiving Multisystemic Therapy). The Cochrane database of
systematic reviews (http://community.cochrane.org/) and the Campbell systematic
reviews (http://www.campbellcollaboration.org/lib/?go=monograph) are two
resources that provide comprehensive reviews and meta-analyses in many areas of
prevention science related to behavioral health and are a resource for examining the
replicability of a specific intervention or type of intervention.
With the explosion of candidate gene studies during the past 10 years, the field
of genetics has also turned its attention to the standard of replicability, although the
topic of replication has not been as prominent in the field of behavioral genetics. A
critical review of the first 10 years of candidate gene studies highlights the replica-
tion challenges in the field in data extracted from all published studies (N=103
studies) from the first decade (through 2009) of G E research using candidate
genes in psychiatry (Duncan & Keller, 2011). Their review indicated that 96% of
the first reported GE findings in a candidate gene study were significant, versus
only 27% of the replication attempts. The authors conclude that these findings sug-
gest the existence of publication bias among GE studies in the candidate gene
field, making hypotheses related to candidate GE effects appear falsely robust.
Further, the Duncan and Keller (2011) review identified an apparent publication
bias among replication attempts, because positive replication attempts had smaller
average sample sizes than did negative ones. Their power calculations using
observed sample sizes suggested that most candidate gene GE studies are under-
powered, and they suggest that low power, along with the likely low prior probabil-
ity of a given GE hypothesis being true, suggests a high prevalence of Type I
errors in this area.
This critique, as well as meta-analyses showing failures to replicate (e.g., Eaves,
2006; Munafo & Flint 2009; Risch et al., 2009), led many leading journals to
develop policies that make replication in candidate gene GE studies a required
criterion for publication. For example, Journal of Abnormal Child Psychologys
publication policy about candidate gene studies states that when a manuscript pres-
ents the first test of a given candidate gene association, it is necessary to provide a
successful replication in a second, independent sample in the same manuscript
(Johnston, Lahey, & Matthys, 2013). The Behavior Genetics and Psychological
Science journals have a similar policy of required replication prior to publication
(Hewitt, 2012). The issue of replication has not been as publicized in the behavioral
genetics literature as it has in the candidate gene literature, but it is of equal impor-
tance. Findings from one behavioral genetics study require replication in order to
meet the standards of prevention science and move us closer to being able to lever-
age the findings to develop a precision medicine approach. This can be difficult
because a limited number of twin studies are available and even fewer adoption
studies. Nonetheless, resources such as the special double-volume issue of Twin
Research and Human Genetics in 2013 include a summary of dozens of twin studies
and adoption studies and their findings, and they are an excellent resource for exam-
ining the replication of findings across multiple behavioral genetics studies.
Adherence to replication of genetic findings is crucial prior to developing a geneti-
cally informed intervention based on those findings; failure to do so may lead to
failure of precision medicine practices with respect to the prevention of develop-
mental psychopathology. Significantly more replication efforts are needed in the
genetics field at large, and specifically in the behavior genetics field, before we can
proceed with greater clarity and confidence in implementing genetically informed
interventions for the prevention of psychopathology.
elevance ofMolecular Genetic Studies toPrecision

R
Medicine Initiatives
This chapter emphasizes the importance of attending to ten standards in prevention

science research while we move closer to being able to conduct genetically informed
interventions for the prevention of psychopathology. The field of behavioral genetics
was selected as the focus of the chapter because it has received significantly less atten-
tion than have molecular genetics approaches to genetically informed interventions.
In addition, as illustrated by the genetic design options discussed in this chapter,
274 L.D. Leve
behavioral genetics may be the most readily translatable to a precision medicine

approach. It focuses on genetically influenced behaviors inferred at the whole genome
level rather than on specific measured gene variants that typically account for only a
small portion of the variance in psychopathology outcomes and require biological
sampling methods.
However, it is important to note the growing number of molecular genetics stud-
ies that have examined whether behavioral interventions are more or less effective
for individuals with or without a hypothesized genetic marker of susceptibility (for
two examples see Brody chapter and Dodge chapter, this volume). At least 22
GE analyses have been conducted to test differential susceptibility theory using
intervention studies with a total of 3257 participants. Preliminary support has been
provided for some aspects of differential susceptibility theory (Bakermans-
Kranenburg & Van IJzendoorn, 2015; Van IJzendoorn & Bakermans-Kranenburg,
2015). In the majority of these studies, the GE findings related to the interven-
tion were hypothesized post hoc, after the intervention had been completed. In
addition, the 22 studies found that eight different susceptibility genes were involved
in the differential susceptibility effects, and the outcomes included a wide range of
behaviors, such as alcohol use, anxiety, externalizing behavior, literacy, IQ, and
depression.
Thus, although the molecular genetics line of work related to precision medicine
is innovative and holds promise, significant work must be done to connect a priori
theories and explanatory mechanisms to specific outcomes of interest (see Prevention
Science Principles 1 and 4). And, molecular genetic studies typically require sample
sizes in the thousands to provide sufficient power, which limits the ability to collect
observational data that would contribute to understanding processes and mecha-
nisms (see Prevention Science Principle 5). Further, there are additional challenges
related to sustainability and dissemination (see Prevention Science Principal 7) in
that an intervention provider would need information about the same aspects of the
clients DNA that were the focus of the original molecular genetic studies. This
information would enable the interventionist or therapist to decide whether the
intervention would be beneficial, given the clients genetic makeup. Having access
to genetic data is not commonplace among mental health service providers at this
time, however, particularly those who serve high-risk populations in community
health settings. For the findings from molecular genetics GE studies to be ready
for dissemination, significant groundwork must be laid for service providers to col-
lect and/or have access to the appropriate genetic data needed to implement a preci-
sion medicine approach. Effects sizes in molecular genetics studies also tend to be
quite small, likely because such studies typically assess only one class of DNA
marker or variation (Thapar & Harold, 2014). In comparison, behavioral genetics
approaches examine the aggregate, anonymous effects of the whole genome.
President Obamas Precision Medicine Initiative may help pave the way for preci-
sion medicine approaches that are guided by molecular genetics studies in medical
settings. In future years, this step may provide a road map that could increase the
uptake of a GE-informed intervention strategy that targets developmental psycho-
pathology. Until Obamas initiative has yielded additional knowledge about the
application of effective precision medicine approaches in medical settings, however,
a behavioral genetics approach may move us closer to understanding how genetically

based characteristics of an individual can affect the extent to which a specific therapy
or treatment may be beneficial or not, and help guide a precision medicine approach
to the prevention of developmental psychopathology.
ext Steps forAdvancing aBehavioral Genetics Informed

N
Approach tothePrevention ofDevelopmental
Psychopathology
This chapter describes ten of the core standards in the field of prevention science
and illustrates how each standard has been applied in preventive intervention
research during the past several decades. These standards have helped the field of
prevention science develop an expansive set of evidence-based programs known to
reduce psychopathology in children, individuals, and families. The implications of
these standards to the translation of behavioral genetics research to inform preven-
tion are described as they apply to the pursuit of a precision medicine approach. To
maximize the relevance of behavioral genetics studies to inform this goal, the fol-
lowing next steps are recommended.
1. Conduct a literature review to identify and link behavioral genetics studies that
test a causal, mechanistic theory with prevention studies that test a similar
causal theory. This first step will help identify the specific theory-driven predic-
tor and outcome behaviors that align across behavioral genetics and prevention,
and can serve as a starting point for narrowing the field to domains of psycho-
pathology where there is strong theory-guided evidence across both prevention
and behavioral genetics studies.
2. Within the subset of linked studies identified in recommendation 1, search for
studies that include a focus on health promotion and have measured positive
predictor and outcome variables. In some cases, these data may exist but might
not yet be analyzed or published. This step is important because the ultimate
outcome in a successful precision medicine approach to developmental psycho-
pathology is healthy adjustment, as indicated not only by the absence of psy-
chopathology, but the presence of indicators of well-being and adaptive
adjustment. Further, because interventions are typically skill building, there is
a great need to identify behavioral genetics studies that measure parallel
strengths and skills that can be targeted and improved via intervention.
3. Reanalyze behavioral genetics studies based on the studies selected in the first
two recommendations, to examine whether the hypothesized environmental
predictor(s) lead to mean-level improvements in the targeted outcome(s). As
reviewed earlier, most published behavioral genetics studies have focused on
correlations within family members; the examination of mean-level improve-
ments as a function of a specific environmental predictor within behavioral
genetics studies will bring us one step closer to identifying behaviors that are
not only heritable, but are modifiable via environmental intervention.
276 L.D. Leve
4. Search the behavioral genetics literature for studies in which environmental

mediators can be identified that could be targeted in a future intervention. A
core premise of prevention research is to target environmental mediators, or
mechanisms, of change. In the case of behavioral genetics research, such medi-
ators would likely be found in the environmental components of variance and
could be either nonshared or shared environmental components. A challenge
here is identifying environmental mediators that are malleable (rather than ones
that are more difficult to change, such as socioeconomic status). Further, exam-
ining instances of geneenvironment correlation can help with the search for
mediating mechanisms that are associated with genetic influences, but that can
be modified via intervention.
5. Conduct behavioral genetics studies that incorporate observational data so that
the specific processes underlying dyadic and family interactions can be exam-
ined using a genetically informed approach. Because preventive interventions
often aim to modify family interaction patterns, the addition of knowledge
about family interaction processes from behavioral genetics studies can help
provide new information about aspects of family interactions that are geneti-
cally influenced, which will ultimately help refine a genetically informed
approach to intervention.
6. Conduct new behavioral genetics studies that comprise individuals who share
risk characteristics with the groups of individuals targeted in an efficacious
prevention study. Because historically, most behavioral genetics studies have
sampled community-based or normative populations, and because we know
that the effects of genes can vary across risk level, it is imperative that new
genetically informed studies be conducted with samples of individuals who
have risk levels that are comparable to those of individuals in most selected
and indicated intervention studies. If this next step is not achieved, behav-
ioral genetic findings will remain relevant to the development of universal
prevention studies, but relevance to selected or indicated studies may be
reduced.
7. Train community-based practitioners to collect a detailed family history of psy-
chopathology from clients. Although collecting a family history of physical
health conditions is commonplace, particularly in medical settings, it is much
rarer for a family history of mental health to be assessed. Community mental
health practitioners do not often ask for detailed extended family histories of
psychopathology. Further, when the goal is to use these data to consider a
genetically informed approach to practice, it is critical that both practitioner
and client understand that the goal of this assessment is to report about biologi-
cal relatives only.
8. Collect long-term follow-up data from behavioral genetic studies and/or
conduct new studies that contain a longitudinal examination of geneenvironment
interplay. Before we can implement effective genetically informed preventive
interventions, we need to ensure that the genetics foundation that the interven-
tion is based upon has sustainable effects. If genes interact with environments
at a single time point, but not in the future, there may be little benefit from
incorporating knowledge from genetic studies into prevention practice.
9. Search for behavioral genetics studies that have measured the core outcome of
interest across multiple indicators. This action will help ensure that genetically
informed prevention approaches are based on a highly reliable outcome. If the
extant behavioral genetics literature suggests that geneenvironment effects are
present for only one type of measure (e.g., for criminal arrest data but not for self-
reported delinquency), it can help inform intervention development so that a
matched outcome is examined across the behavioral genetics and prevention study.
10. Replication, replication, replication. The literature is replete with failed replica-
tion attempts in the area of geneenvironment interplay, and there are countless
unpublished null effect findings. Before we can take strong strides in develop-
ing a genetically informed approach to prevention science, we need to examine
whether the geneenvironment effect that the intervention is based upon has
been replicated. Although it takes time to replicate, failure to base a precision
medicine approach to psychopathology on replicated effects is likely to be
costly, and ultimately harmful to the field and to the populations served.
Conclusion
It would be a grand challenge for a single genetic study to meet all ten standards
described in this chapter. This is in part because these standards were originally
developed for prevention science and not for the field of genetics. In fact, many
published prevention studies have not successfully met the majority of these stan-
dards. However, if the goal of a precision medicine approach is to use knowledge
from genetic studies to improve existing interventions or to develop new interven-
tions, it is important to adhere to as many of the prevention standards as possible.
Without consistency across the two disciplines in their core standards, we may pur-
sue avenues that are unlikely to yield improvements in the effectiveness of existing
preventive interventions, or we may develop new GE-informed prevention pro-
grams that are no more effective than existing prevention programs. Nonetheless,
there is a great need to pursue this line of work, to better bridge the disciplines, and
to look for and conduct behavioral genetics studies that map onto the standards of
prevention science to facilitate translation across the two fields. The ten steps
described here will help move us closer to an integrated precision medicine approach
to developmental psychopathology. Even the most successful prevention studies in
the area of psychopathology tend to have moderate effect sizes, and a genetically
informed approach to prevention may help improve those sizes.
Moreover, every intervention has a group of nonresponders who do not show
improvement even after receiving the full dose of a rigorous evidence-based preven-
tion program. If successful, applying a genetically informed intervention approach
might help improve effect sizes in prevention studies (see Leve etal., 2010, for a
review of this concept). A greater number of individuals might show improved out-
comes and better behavioral health if we can successfully map information about an
individuals genetic liabilities and strengths onto a given prevention approach.
Prevention researchers need to be ready to incorporate findings and knowledge from
278 L.D. Leve
genetic studies by focusing on understanding patterns of nonresponsiveness to

interventions. For whom is a specific preventive intervention not working? Are there
systematic patterns of nonresponsiveness that might be linked to genetic etiology?
Could we modify our intervention approaches for specific subgroups if we had this
knowledge? Gaining additional understanding of why and for whom specific pre-
vention programs are and are not effective will help ready the field of prevention to
incorporate knowledge from genetic studies and facilitate the implementation of
prevention strategies that are guided by precision medicine. Given the rapidly evolv-
ing science in this area and the call for action from the President of the United
States, it is crucial that prevention scientists and geneticists alike closely attend to
the Precision Medicine Initiative, watch the medical field for advances and chal-
lenges as precision medicine approaches are attempted, and position our science
and our intervention programs related to developmental psychopathology to incor-
porate knowledge from behavioral genetics studies in the coming decade, once suf-
ficient knowledge is present to suggest the benefit of a precision medicine approach
to interventions targeting psychopathology.
References
Ashley, E.A. (2015). The precision medicine initiative: A new national effort. JAMA, 313(21),
21192120. doi:10.1001/JAMA.2015.3595.
Bakermans-Kranenburg, M.J., & Van IJzendoorn, M.H. (2015). The hidden efficacy of interven-
tions: Gene environment experiments from a differential susceptibility perspective. Annual
Review of Psychology, 66, 381409. doi:10.1146/annurev-psych-010814-015407.
Bates, T. C., Lewis, G. J., & Weiss, A. (2013). Childhood socioeconomic status amplifies
genetic effects on adult intelligence. Psychological Science, 24, 21112116.
doi:10.1177/0956797613488394.
Belsky, J., & Pluess, M. (2009). Beyond diathesis stress: Differential susceptibility to environmen-
tal influences. Psychological Bulletin, 135(6), 885908. doi:10.1037/a0017376.
Bernard, K., Dozier, M., Bick, J., Lewis-Morrarty, E., Lindhiem, O., & Carlson, E. (2012).
Enhancing attachment organization among maltreated children: Results of a randomized clini-
cal trial. Child Development, 83, 623636. doi:10.1111/j.1467-8624.2011.01712.x.
Cadoret, R.J., Cain, C.A., & Grove, W.M. (1980). Development of alcoholism in adoptees raised
apart from alcoholic biologic relatives. Archives of General Psychiatry, 37(5), 561563.
doi:10.1001/archpsyc.1980.01780180075008.
Cadoret, R.J., Troughton, E., OGorman, T.W., & Heywood, E. (1986). An adoption study of
genetic and environmental factors in drug abuse. Archives of General Psychiatry, 43(12),
11311136. doi:10.1001/archpsyc.1986.01800120017004.
Chamberlain, P. (2003). Treating chronic juvenile offenders: Advances made through the Oregon
Multidimensional Treatment Foster Care model. Washington, DC: American Psychological
Association.
Chamberlain, P., Price, J., Leve, L. D., Laurent, H., Landsverk, J. A., & Reid, J. B. (2008).
Prevention of behavior problems for children in foster care: Outcomes and mediation effects.
Prevention Science, 9, 1727. doi:10.1007/s11121-007-0080-7.
Collins, F. S., Morgan, M., & Patrinos, A. (2003). The Human Genome Project: Lessons from
large-scale biology. Science, 300, 286290. doi:10.1126/science.1084564.
Conduct Problems Prevention Research Group. (2010). The effects of a multiyear universal social-
emotional learning program: The role of student and school characteristics. Journal of Consulting
and Clinical Psychology, 78(2):156168.
DeFries, J.C., Plomin, R.E., & Fulker, D.W. (1994). Nature and nurture during middle childhood.
London: Blackwell.
Dishion, T. J., Patterson, G. R., Stoolmiller, M., & Skinner, M. L. (1991). Family, school, and
behavioral antecedents to early adolescent involvement with antisocial peers. Developmental
Psychology, 27(1), 172180. doi:10.1037/0012-1649.27.1.172.
Dodge, K.A., Bierman, K.L., Coie, J.D., Greenberg, M.T., Lochman, J.E., McMahon, R.J.,
Conduct Problems Prevention Research Group. (2014). Impact of early intervention on psy-
chopathology, crime, and well-being at age 25. The American Journal of Psychiatry, 172(1),
5970. doi:10.1176/appi.ajp.2014.13060786.
Dozier, M., Peloso, E., Lewis, E., Laurenceau, J.-P., & Levine, S. (2008). Effects of an attachment-
based intervention on the cortisol production of infants and toddlers in foster care. Development
and Psychopathology, 20(3), 845859. doi:10.1017/S0954579408000400.
Duncan, L.E., & Keller, M.C. (2011). A critical review of the first 10 years of candidate gene-by-
environment interaction research in psychiatry. The American Journal of Psychiatry, 168(10),
10411049. doi:10.1176/appi.ajp.2011.11020191.
Eaves, L.J. (2006). Genotypeenvironment interaction in psychopathology: Fact or artifact? Twin
Research and Human Genetics, 9, 18. doi:10.1375/twin.9.1.1.
Ellis, B. J., & Boyce, W. T. (2008). Biological sensitivity to context. Current Directions in
Psychological Science, 17(3), 183187. doi:10.1111/j.1467-8721.2008.00571.x.
Ellis, B.J., Essex, M.J., & Boyce, W.T. (2005). Biological sensitivity to context: II.Empirical
explorations of an evolutionary-developmental theory. Development and Psychopathology, 17,
303328. doi:10.1017/S0954579405050157.
Engelhardt, L.E., Briley, D.A., Mann, F.D., Harden, K.P., & Tucker-Drob, E.M. (2015). Genes
unite executive functions in childhood. Psychological Science, 26(8), 11511163.
doi:10.1177/0956797615577209.
Flay, B.R., & Allred, C.G. (2003). Long-term effects of the Positive Action program. American
Journal of Health Behavior, 27, 621. doi:10.5993/AJHB.27.1.s1.2.
Forgatch, M.S., Patterson, G.R., DeGarmo, D.S., & Beldavs, Z.G. (2009). Testing the Oregon
delinquency model with 9-year follow-up of the Oregon Divorce Study. Development and
Psychopathology, 21(02), 637660. doi:10.1017/S0954579409000340.
Gottfredson, D.C., Cook, T.D., Gardner, F.E., Gorman-Smith, D., Howe, G.W., Sandler, I.N., &
Zafft, K.M. (2015). Standards of evidence for efficacy, effectiveness, and scale-up research in
prevention science: Next generation. Prevention Science, 16, 893926. doi:10.1007/
s11121-015-0555-x.
Granic, I., & Patterson, G.R. (2006). Toward a comprehensive model of antisocial development:
A dynamic systems approach. Psychological Review, 113(1), 101131. doi:10.1037/0033-
295X.113.1.101.
Greenberg, M.T., Kusche, C.A., Cook, E.T., & Quamma, J.P. (1995). Promoting emotional com-
petence in school-aged children: The effects of the PATHS curriculum. Development and
Psychopathology, 7, 117136. doi:10.1017/S0954579400006374.
A. (2013). Biological and rearing mother influences on child ADHD symptoms: Revisiting the
Psychiatry, 54, 10381046. doi:10.1111/jcpp.12100.
Hewitt, J. K. (2012). Editorial policy on candidate gene association and candidate gene-by-
environment interaction studies of complex traits. Behavior Genetics, 42(1), 12. doi:10.1007/
s10519-011-9504-z.
Johnston, C., Lahey, B. B., & Matthys, W. (2013). Editorial policy for candidate gene studies.
Journal of Abnormal Child Psychology, 41(4), 511514. doi:10.1007/s10802-013-9741-0.
Kendler, K.S., Ji, J., Edwards, C., Ohlsson, H., Sundquist, J., & Sundquist, K. (2015). An extended
Swedish national adoption study of alcohol use disorder. JAMA Psychiatry, 72(3), 211218.
doi:10.1001/jamapsychiatry.2014.2138.
Khoury, M.J., & Evans, J.P. (2015). A public health perspective on a national precision medicine
cohort: Balancing long-term knowledge generation with early health benefit. JAMA, 313,
21172118. doi:10.1001/jama.2015.3382.
280 L.D. Leve
Leve, L.D., Chamberlain, P., & Reid, J.B. (2005). Intervention outcomes for girls referred from
juvenile justice: Effects on delinquency. Journal of Consulting and Clinical Psychology, 73(6),
11811185. doi:10.1037/0022-006X.73.6.1181.
Leve, L.D., Harold, G.T., Ge, X., Neiderhiser, J.M., & Patterson, G. (2010). Refining intervention
targets in family-based research: Lessons from quantitative behavioral genetics. Perspectives
on Psychological Science, 5, 516526. doi:10.1177/1745691610383506.
Leve, L.D., Harold, G.T., Ge, X., Neiderhiser, J.M., Shaw, D., Scaramella, L.V., & Reiss, D.
(2009). Structured parenting of toddlers at high versus low genetic risk: Two pathways to child
problems. Journal of the American Academy of Child and Adolescent Psychiatry, 48, 1102
1109. doi:10.1097/CHI.0b013e3181b8bfc0.
Leve, L.D., Harold, G.T., Shaw, D.S., Natsuaki, M.N., Reiss, D., & Neiderhiser, J.M. (2014).
Heritable yet malleable: Environmental influences on adopted childrens achievement aca-
demic. Paper presented at the 2014 Behavior Genetics Association annual meeting,
Charlottesville, VA, 1821 June 2014.
Leve, L.D., Neiderhiser, J.M., Shaw, D.S., Ganiban, J., Natsuaki, M.N., & Reiss, D. (2013). The
Early Growth and Development Study: A prospective adoption study of child behavior from
birth through middle childhood. Twin Research and Human Genetics, 16, 412423. doi:10.1017/
thg.2012.126.
Lfholm, C. A., Olsson, T., Sundell, K., & Hansson, K. (2009). Multisystemic therapy with
conduct-disordered young people: Stability of treatment outcomes two years after intake.
Evidence & Policy: A Journal of Research, Debate and Practice, 5(4), 373397.
Malti, T., Ribeaud, D., & Eisner, M.P. (2011). The effectiveness of two universal preventive inter-
ventions in reducing childrens externalizing behavior: A cluster randomized controlled trial.
Journal of Clinical Child and Adolescent Psychology, 40(5), 677692. doi:10.1080/15374416
.2011.597084.
Martinez, C. R., & Forgatch, M. S. (2001). Preventing problems with boys noncompliance:
Effects of a parent training intervention for divorcing mothers. Journal of Consulting and
Clinical Psychology, 69(3), 416428. doi:10.1037/0022-006X.69.3.416.
McMahon, R. J., Bierman, K. L., Coie, J. D., Dodge, K. A., Greenberg, M. T., Lochman, J. E., &
Pinderhughes, E. E. (1999). Initial impact of the Fast Track prevention trial for conduct prob-
lems: I. The high-risk sample. Journal of Consulting and Clinical Psychology, 67(5),
631647.
Menting, A.T., de Castro, B.O., & Matthys, W. (2013). Effectiveness of the Incredible Years par-
ent training to modify disruptive and prosocial child behavior: A meta-analytic review. Clinical
Psychology Review, 33(8), 901913. doi:10.1016/j.cpr.2013.07.006.
Mrazek, P.J., & Haggerty, R.J. (Eds.) (1994). Reducing risk for mental disorders: Frontiers for
preventive intervention research. Washington, DC: National Academy Press.
Munafo, M.R., & Flint, J.(2009). Replication and heterogeneity in geneenvironment interaction
studies. The International Journal of Neuropsychopharmacology, 12, 727729. doi:10.1017/
S1461145709000479.
Natsuaki, M.N., Ge, X., Leve, L.D., Neiderhiser, J.M., Shaw, D., Conger, R.D., Reiss, D.
(2010). Genetic liability, environment, and the development of fussiness in toddlers: The roles
of maternal depression and parental responsiveness. Developmental Psychology, 46, 1147
1158. doi:10.1037/a0019659.
OConnor, T. G., Deater-Deckard, K., Fulker, D., Rutter, M., & Plomin, R. (1998). Genotype-
environment correlations in late childhood and early adolescence: Antisocial behavioral prob-
lems and coercive parenting. Developmental Psychology, 34(5), 970981.
doi:10.1037/0012-1649.34.5.970.
Olds, D. L. (2006). The nursefamily partnership: An evidence-based preventive intervention.
Infant Mental Health Journal, 27(1), 525. doi:10.1002/imhj.20077.
Olds, D.L., Henderson, C.R., & Kitzman, H. (1994). Does prenatal and infancy nurse home visi-
tation have enduring effects on qualities of prenatal caregiving and child health at 25 to 50
months of life? Pediatrics, 93(1), 8998.
Olds, D.L., Holmberg, J.R., Donelan-McCall, N., Luckey, D.W., Knudtson, M.D., & Robinson,
J.(2014). Effects of home visits by paraprofessionals and by nurses on children: Follow-up of
a randomized trial at ages 6 and 9 years. JAMA Pediatrics, 168(2), 114121. doi:10.1001/
jamapediatrics.2013.3817.
Olds, D. L., Kitzman, H. J., Cole, R. E., Hanks, C. A., Arcoleo, K. J., Anson, E. A., ... Stevenson,
A. J. (2010). Enduring effects of prenatal and infancy home visiting by nurses on maternal life
course and government spending. Archives of Pediatrics & Adolescent Medicine, 164(3), 419
424. doi:10.1001/archpediatrics.2010.49.
Patterson, G.R. (1982). Coercive family process. Eugene, OR: Castalia.
Patterson, G.R., Reid, J.B., & Dishion, T.J. (1992). Antisocial boys. Eugene, OR: Castalia.
Pears, K.C., Healey, C.V., Fisher, P.A., Braun, D., Gill, C., Conte, H.M., Ticer, S. (2014).
Immediate effects of a program to promote school readiness in low-income children: Results of
a pilot study. Education and Treatment of Children, 37(3), 431460. doi:10.1353/etc.2014.0021.
Plomin, R., DeFries, J. C., Knopik, V. S., & Neiderhiser, J. M. (2013). Behavioral genetics.
NewYork: Palgrave Macmillan.
Reiss, D., Leve, L.D., & Neiderhiser, J.M. (2013). How genes and the social environment moder-
ate each other. American Journal of Public Health, 103, S111S121. doi:10.2105/
AJPH.2013.301408.
Reiss, D., Neiderhiser, J.M., Hetherington, E.M., & Plomin, R. (2000). The relationship code.
Cambridge: Harvard University Press.
Risch, N., Herrell, R., Lehner, T., Liang, K.-Y., Eaves, L., Hoh, J., Merikangas, K.R. (2009).
of depression: A meta-analysis. JAMA, 301, 24622471. doi:10.1001/jama.2009.878.
Roisman, G.I., & Fraley, R.C. (2012). A behavior-genetic study of the legacy of early caregiving
experiences: Academic skills, social competence, and externalizing behavior in kindergarten.
Child Development, 83(2), 728742. doi:10.1111/j.1467-8624.2011.01709.x.
Samek, D.R., Hicks, B.M., Keyes, M.A., Bailey, J., McGue, M., & Lacono, W.G. (2015). Gene
environment interplay between parentchild relationship problems and externalizing disorders
in adolescence and young adulthood. Psychological Medicine, 45(02), 333344. doi:10.1017/
S0033291714001445.
Slutske, W.S., Ellingson, J.M., Richmond-Rakerd, L.S., Zhu, G., & Martin, N.G. (2013). Shared
genetic vulnerability for disordered gambling and alcohol use disorder in men and women:
Evidence from a national community-based Australian Twin Study. Twin Research and Human
Genetics, 16(2), 525534. doi:10.1017/thg.2013.11.
Society for Prevention Research. (2011). Standards of knowledge for the science of preven-
tion. Retrieved from http://www.preventionresearch.org/conferences/publications/standards-
of-knowledge/
Stein, M. B., Jang, K. L., Taylor, S., Vernon, P. A., & Livesley, W. J. (2002). Genetic and environ-
mental influences on trauma exposure and posttraumatic stress disorder symptoms: A twin
study. American Journal of Psychiatry, 159(10), 16751681.
Tang, Y.Y., Yang, L., Leve, L.D., & Harold, G.T. (2012). Improving executive function and its
neurobiological mechanisms through a mindfulness-based intervention: Advances within the
field of developmental neuroscience. Child Development Perspectives, 6, 361366.
doi:10.1111/j.1750-8606.2012.00250.x.
Thapar, A., & Harold, G. (2014). Editorial perspective: Why is there such a mismatch between
traditional heritability estimates and molecular genetic findings for behavioural traits? Journal
of Child Psychology and Psychiatry, 55, 10881091. doi:10.1111/jcpp.12294.
Tucker-Drob, E.M., Rhemtulla, M., Harden, K.P., Turkheimer, E., & Fask, D. (2011). Emergence
of a gene socioeconomic status interaction on infant mental ability between 10 months and 2
years. Psychological Science, 22(1), 125133. doi:10.1177/0956797610392926.
Turkheimer, E., Haley, A., Waldron, M., DOnofrio, B., & Gottesman, I. I. (2003).
Socioeconomic status modifies heritability of IQ in young children. Psychological Science,
14(6), 623628.
Van IJzendoorn, M.H., & Bakermans-Kranenburg, M.J. (2015). Genetic differential susceptibility
on trial: Meta-analytic support from randomized controlled experiments. Development and
Psychopathology, 27(1), 151162. doi:10.1017/S0954579414001369.
282 L.D. Leve
Van Ryzin, M.J., Leve, L.D., Neiderhiser, J.M., Shaw, D.S., Natsuaki, M.N., & Reiss, D. (2015).
Genetic influences can protect against unresponsive parenting in the prediction of child social
competence. Child Development, 86(3), 667680. doi:10.1111/cdev.12335.
Young-Wolff, K.C., Kendler, K.S., Ericson, M.L., & Prescott, C.A. (2011). Accounting for the
association between childhood maltreatment and alcohol-use disorders in males: A twin study.
Psychological Medicine, 41(1), 5970. doi:10.1017/S0033291710000425.
Chapter 14
Challenges forIntervention Research Within
theGEX Framework
PatrickH.Tolan andBennettL.Leventhal
hallenges forIntervention Research WithintheGEX

C
Framework
In the past 20years, the geneenvironment transaction (GEX) approach to develop-

mental psychopathology has rapidly emerged, providing many clues about potential
causal pathways, while also leading to evolving models of increasing elaboration
about the variety of ways in which genetic and environmental contributors might
relate to developmental psychopathology. Similarly, there is increasing recognition
that initial findings may serve less as direct indicators of causal mechanisms and more
as signposts about the limitations of simple models and the complexity of develop-
mental psychopathology gene environment transactions (see McGue and Carey,
Meany, and Reiss chapters in this volume as examples of the evolving understanding
and related increasing complexity, specificity, and diversity of how geneenvironment
transactions can be characterized). Accompanying the developments has been a grow-
ing understanding of gene expression as well as the dependency of this expression on
complex transactions. It is increasingly apparent that environmental variations offer as
much complexity as do genetic contributions to the development of psychopathology.
As a result, recently, there has been more attention directed at methodological consid-
erations in conducting this very crucial but complex research.
P.H. Tolan (*)

Youth-Nex Center, University of Virginia, 405 Emmet Street South,
Charlottesville, VA 22904-4281, USA
e-mail: pht6t@virginia.edu
B.L. Leventhal
Department of Psychiatry, University of California at San Francisco, 405 Emmet Street SouthBox
0984-CAS, Room LP-152, 401 Parnassus Avenue San Francisco, CA 94143-0984, USA

DOI10.1007/978-3-319-49227-8_14
Identifying the notion of transactions within the course of human development

has caused the rethinking of initial formulations in order to offer greater consideration
to the diversity of outcomes among individuals sharing similar genetic and environ-
mental features (multifinality), and the extent to which endophenotypes and cer-
tainly syndromes may be traceable to multiple sources that differ across persons and
groups (equifinality; Cannon & Keller, 2006). Similarly, concepts such as critical
periods, cascade of risk and resilience, and attention to geneenvironment influ-
ences on developmental influences (e.g., parenting), not just on pathology/symp-
toms, are now essential elements of basic assumptions about geneenvironment
transactions. Overall, the pattern of progress has moved rapidly from simplified and
unitary models with limited reliability and robustness to ones with more subtlety,
complexity, specificity. The overall trend is to greater elaborations in understanding
the role of geneenvironment variation on developmental psychopathology.
It is now important to attend to the role of experimental or intervention studies,
those that attempt to understand induced change to test causal role(s) and processes.
This symposium and the resulting volume are among the first attempts to focus on this
topic for developmental psychopathology. The related proceedings are oriented
toward digesting the implications of the advancing geneenvironment transactions
framework for intervention studies and, in turn, to report on efforts and issues in appli-
cation of intervention studies to understand geneenvironment-based hypotheses and
models. This discussion has yielded a more informed and informing understanding of
the multiple considerations associated with undertaking intervention studies. These
considerations include debates about what caution is counseled in designing interven-
tion studies in the context of the current knowledge base and how to properly incorpo-
rate ethical guidelines in the application and use of such studies. The issues include
many that are not specific to GEX research but rather to developmental analyses and
to intervention analyses in particular (mediation and moderation). There are even
more complexities related to conceptualizations of the GEX processes and its compo-
nent relationships and to the inferences that can be drawn from current efforts in
which genetic variation in response to psychosocial interventions is studied. Most
fundamentally, the collected contributions show that it is no longer premature to con-
sider intervention research as a fundamental tool in GEX research but that it is done
within a tension of multiple ethical and methodological concerns (Radder, 2009).
I ssues inConceptualizing GeneEnvironment Dependencies

andTransactions Differentiating GeneEnvironment
Relations
As noted by Moffitt, Caspi, and Rutter (2005), the GEX approach brings several
advantages over models that ignore either gene or environment or do not consider the
interaction of the two, including improving ability to detect effects that are not visible
in simpler models. This helps to understand variations in effects of developmental
influences and interventions. There are several notable distinctions in the types of
14 Challenges forIntervention Research Within theGEX Framework 285
gene and environment relations which are identified and enumerated in this volume.,
Most grossly, these focus on whether the relation is one of interaction, meaning the
genetic and environmental contribution are dependent on the other for an effect or one
of correlation, which is the gene and environmental conditions co-occur but not as
interdependent in effect (Tolan, Rutter, & Dodge, 2013 provide a discussion of these
as variations in how parenting might affect psychopathology). The geneenvironment
correlation can be evocative, that is, the genetic variation may relate to eliciting envi-
ronmental conditions or can be passive, meaning the co-occurrence is incidental and
environmental conditions are not traceable to the genetic influence on behavior.
Further elaborations beyond these basic distinction are being theorized, discov-
ered, and refined at a rapid pace about geneenvironment correlations, interactions,
and complex dependencies over time and developmental processes (see Reiss, this
volume; Dodge, 2017). These conceptualizations and tests of possible interactions
(or sets of interactions of varying type and extent of dependency) suggest that test-
ing for interaction effects, whether to isolate reactivity to intervention or empirically
test environmental shifts to show dependency effects, must consider chains of inter-
actions or multiple coinciding processes. Even if methodological considerations
such as sample size needed for multiple interaction tests or for low incident modera-
tion (affecting few in a population) are not practically limiting of these studies,
statistical models of probability theory must be inadequate to validly test multiple
potential dependencies simultaneously.
Differentiating Gene Variant Basis andFeatures ofNetworks
Forays into understanding the pathogenesis of major psychiatric illnesses such as

Autism and Depression demonstrate the need to differentiate de novo mutations
from common variants (Bagot etal., 2016; Meaney, this volume). At this point, it
seems that multiple and complex variations of GEX transactions will be challeng-
ing. There is also the need to consider genegene relations as well and to character-
ize network features and processes of the sets of dependencies and to do so in
relation to developmental timing and life-course variations (Willsey etal., 2013).
Influences on genetic expression through processes such as methylation and telo-
mere alteration are additional specifications of network models of GEXs (Radder,
2009). At this point, it seems that elaborations on the basic concept of some form
of dependency and summative effects of gene and environmental contributions to
psychopathology are advantageous and perhaps should be preferred over tests of
strict interactions. The explanations of risk and related intervention tests need to be
cast with a recognition of potentially independent and multiple gene and environ-
mental influences, along with dependencies that may have complex effects over
time. It is likely that multiple gene and environment intersections can occur, includ-
ing multiple forms of geneenvironment correlations and multiple forms of strict
interactions (McGue & Carey, this volume; Meaney, this volume; Tsuang, Bar,
Stone, & Faraone, 2004).
ifferentiating Susceptibility toEnvironmental Influence

D
fromModels ofSpecific Pathology Influence
In addition to differentiating geneenvironment correlations from transactions in

which the impact of one influence is constrained by the presence of the other, dif-
ferentiation can be identified as the nature of the geneenvironment relation for
susceptibility to psychopathology. Some have interpreted such effects as reflecting
differences in genetic sensitivity to environmental influence (Belsky & Pluess,
2009; Tsuang etal., 2004), while others have cast the differential outcome as a par-
ticular effect of genetic variation on how environment impacts behavior, not just an
overall susceptibility to influence (Moffitt etal., 2005). The former characterization
of the interaction effect is that environmental factor impact is because of individual
variation in genetic sensitivity or reactivity to environmental conditions generally.
For example, from this perspective the same persons might benefit more from par-
ticularly nurturing environments as well as have more detrimental outcomes from
more impoverished or harmful environment. The latter approach looks for a specific
outcome effect and a specific variation relation; that a specific gene variation and
environmental risk factor combine to produce psychopathology symptoms.
How one theorizes the interdependency processes fundamentally informs what
gene variations are of interest and what type of testing of effects must be undertaken.
Intervention studies may be particularly useful for helping to clarify the extent of a
given interdependency and how it reflects a singular direction or specific effect versus
a more general environmental sensitivity. Moreover, if the assumption is that the
geneenvironment effect is singular in direction and specific in outcome then an
experimental test that altered the environment, from enriched to impoverished will not
be of interest. If the view is that the genetic contribution is about overall sensitivity to
environment this will be a particularly valuable experimental paradigm. Depending
on how the specific interactions are thought to cause psychopathology or increase its
probability, the role of the environmental manipulation to test the interaction will also
be different in that it will be aimed at a specific susceptibility rather than a generally
greater environmental impact (Keller & Cannon, 2006). One model leads to analyses
that focus on specific moderation relationshipsa genetically susceptible subgroup
shows greater problems, while the other will demonstrate or test that with intervention
that allows those at-risk to show greater than typical capabilities.
ifferentiating Endophenotypes fromSymptoms

D
andSyndromes
Those interested in tracing neurophysiological substrates of psychopathology have

noted the challenge of great heterogeneity in symptoms and intensity of symptoms
among those with the same diagnosis with current organization of syndromes (Moffitt
et al., 2005). This suggests a need to more specifically differentiate symptoms or
components of symptoms into termed endophenotypes (Gottesman & Gould, 2003).

An example of this is focusing on facial emotion recognition deficits to track genetic
variations among bipolar patients rather than more molar symptoms, such as elation/
mania (Soeiro-de-Souza etal., 2012). Moreover, as explanatory models move toward
complex multiple interactions and multiple potential causal pathways for a given
symptom, the intervention models and effects of intervention will become more com-
plex, as well. Also, it may be that the same symptom or endophenotype arises from
different pathways or causes when occurring as part of different types of psychopathol-
ogy. Deficits in understanding emotional facial expression, due to lack of sensitivity
and inaccuracy, may mark multiple psychopathologies but may be dependent on differ-
ent gene and environment interactions and networks of effects. As most interventions
focus on reduction of disorder/syndromes or relatively molar components (e.g., aggres-
sive behavior in conduct disorders), this issue warrants careful consideration for inter-
preting intervention variation in impact on those with a given diagnosis (Beauchaine,
2009). For example, Frick, Blair, and Castellanos (2013) reviewed the differentiation
of disruptive behavior disorders characterized by callous-unemotional traits from those
without this feature, noting differential intervention reactivity and related genetic varia-
tions. Wakschlag, Tolan, and Leventhal (2010) suggest this differentiation can be fur-
ther distinguished into five components of disruptive behavior disorder that can be the
focus in more specific interventions. They theorize differential genetic and environ-
mental risk interactions underlying these basic elements, suggesting differential inter-
vention response, as well.
Conceptualizing Environmental Influences
While much of the excitement for the GEX model is to track genetic variants or dif-
ferences in genetic heritage as differentiating psychopathology risk, due consider-
ation of how environmental influences are conceptualized and measured is also
critical for sound research. Just as outcome behaviors and potential genetic con-
tributors need specific definitions with good discrimination from similar constructs/
potential explanations, useful results depend on such clarity about environment
influences. It is often difficult to isolate a specific environmental influence from co-
occurring influences and from potential contending explanation; there is a tendency
to simply consider such influences in a general way such as simple differentiation
of poor parenting or exposure to life stress. Complex and dynamic phenomena can
vary along multiple dimensions (see Tolan etal., 2013 for a discussion of this topic
in relation to parenting influences on disruptive behavior disorders). As with out-
comes, reliable and specific definitions and measurement are critical for sensitivity
to detect effects, reliable and robust findings, and valid understanding of the impli-
cations of a given set of findings. Moreover, to be useful for developmental psycho-
pathology, the variations in multiple levels of environmental influences (e.g.,
microsystems from mesosystems, Bronfenbrenner, 1979), it may be necessary to
capture the variation in a given environmental influence within measurement of and
appropriate modeling of nesting relationships with other levels (Tolan, Guerra, &
Kendall, 1995). In addition, as with any moderation analysis, the joint distribution
of the coincidence of the risk factors of interest can greatly affect sensitivity to
detect effects (Farrell, Henry, & Bettencourt, 2013). Therefore, careful consider-
ation of the natural patterns of variations in environmental conditions is an impor-
tant influence on validity of studies of GEX.For example, when an extreme level of
an environmental variation is needed to detect effects, it is more likely that this
effect will be difficult to replicate, either because the benchmark or adverse environ-
mental influence was found rather than prehypothesized or because measurement
reliability of extreme scores is less reliable. In addition, such findings may have
limited translatability for sound developmental psychopathology theory because
results are more likely to be sample/population specific.
I ssues inConceptualizing andMeasuring Intervention

Effects WithinaGEX Model
Many of the challenges for the GEX framework intervention studies are not specific
to that framework, but accompany any attempt to show effects dependent on interac-
tion. In most current research, the interest is in testing for differential reaction to an
intervention, based on genetic variation. These studies apply moderation analyses,
focusing on differential effects as represented by interaction terms and then tracing a
statistically significant interaction to different intervention slopes for subgroups
defined by crossing the theorized interventions. The program of research by Gene
Brody and colleagues (Brody, 2017) represents one of the most developed and care-
fully formulated efforts of this kind. Brody refers to the strict moderation studies as
first generation. In his work, he has followed up initial moderation findings with tests
of genetic variation relating to differential cognitive processes that mediate stress-
related interactions with genetic risk for alcohol abuse (Brody, this volume).
Measuring Moderation andMediation
Moderation tests of interventions have many challenges and some limitations that
are pertinent to studies within the GEX framework. Perhaps most pertinent is the
low sensitivity of most interaction tests; this makes large samples necessary to
detect interactions and to accurately determine the portion of the sample for which
intervention effects are significantly different (Farrell et al., 2013; Preacher,
Curran, & Bauer, 2006). While it is still a prevailing practice in GEX modeling to
attempt identification of a benchmark to categorize environmental risk (e.g., a
certain level of violence exposure is the threshold for genetic variation in risk to
occur). However, this approach seems fraught with unreliability and underestima-
tion of genetic variation and environmental variation interdependencies (see
Preacher etal., 2006 for a discussion of the limitations). Even if analyses were
designed to consider at what level of environmental variation there is genetic
variation interaction in the likelihood of a symptom the results are quite vulnera-
ble to false positives; capitalizing on multiple statistical tests and post hoc identi-
fication of interactions (e.g., that at a certain level of adverse event exposure, one
can find genetic variation related to the outcome). The ability to move forward
with GEX models, particularly regarding detecting differential intervention
effects, will depend on more careful consideration of how moderation is con-
ceived and modeled. Similarly, the sample size and sample appropriateness for the
theorized interdependency are also critical for valid moderation analyses (see
Kim, this volume for a discussion of sampling issues including those related to
moderation detection).
As intervention research moves from moderation tests to identifying potential
genetic variations related to variability in environmental susceptibility (specific or
as a general tendency) to focus on geneticenvironmental interactions as mediators
of differential effects, the methodological issues affecting sensitivity and reliability
of mediation analyses become quite important (MacKinnon & Pirlott, 2015).
Incumbent considerations of mediation analyses are a clear theory of causal pro-
cesses and how the mediation occurs between a cause and the outcome. For inter-
vention research, this means specifying the processes affected by the intervention
and to statistically test that the mediator accounts for much if not all of the signifi-
cant relation between the predictor and the outcome. As MacKinnon and Pirlott
(2015) and others have noted, mediation analyses are highly susceptible to non-
specificity because the mediator tested is a general process and is likely to be influ-
ential because the theory is correct, and no other competing mediators (counterfactual
mediators) are tested. Moreover, mediation analyses are susceptible to confounds
being mistakenly interpreted as the causal link. For example, it may be that in a test
of genetic variation of risk for conduct disorder, parenting training for consistency
of positive reinforcement has effects related to genetic variation. However, it may
be that parental learning and consistent use of such parenting skills is dependent on
the responsiveness of the child, which might be linked to genetic susceptibility for
a chronic disorder. While changing parenting skills may appear to be a mediator, it
may well be that this is a confound of responsiveness to parenting.
A challenge in tracking mediation across development which occurs in interven-
tion studies is clarity about how mediation occurs over time and development. For
example, how might geneenvironment interactions early in development that pro-
mote susceptibility to schizophrenia mediate outcomes of symptom expression that
begin in early adulthood? How is that to be modeled to show a specific point in
development has the effect even if expression may be much later? In addition, if
multiple mediators are at work, as is likely for most types of psychopathology, stud-
ies must consider how to model multiple mediators. The latter can be practically
challenging, given the samples sizes and the measurement precision needed. In
addition, it can be difficult to specify the relationships between multiple mediators
to reliably and validly model these statistically.
Improving Measurement Reliability andValidity
As noted earlier, conceptualizing the genetic markers and processes of interest is a

major consideration that applies to intervention studies. As models include multi-
gene, genegene interactions, networks of genetic interdependencies and other
elaborations and inasmuch as they account for gene structure and processes respon-
sible for risk influence, the importance of reliability of measurement is increasingly
critical. Small decreases in reliability can have large impact on ability to detect
effects (Luan, Wong, Day, & Wareham, 2001).
As specificity of process, differentiation of variation, and specificity of outcomes
become identified more, replicated, and are represented by validated measures that
measure environmental influences at the appropriate unit, opportunities to capture
the meaning of biological processes (e.g., telomere variations) will improve. In par-
ticular, there is important work to be done in developing better measures of behav-
ioral tendencies (e.g., self-control) thought to represent elemental components of
social and emotional functioning across the life span so that developmental tracking
can be improved (e.g., how does self-control as typically measured in infancy by
distraction resistance relate to self-control measured in adolescents as thrill seeking
and in young adulthood as agency and persistence for goal seeking?).
Further, there has been limited study of how theoretically linked component
social and emotional skills and behavioral competencies and symptoms are actually
linked. For example, there is limited understanding about how executive function-
ing components are distinct, interdependent, and ordered in terms of primacy and
developmentally criticality. Related to this is the need for more attention to how
developmental changes in expression of behavior, symptoms, character traits, and
sensitivity to environmental variation could affect what geneenvironment depen-
dencies are identified and how intervention effects are expected to affect these.
Improving Attention toSample Implications
An often overlooked but important impact on any study is the appropriateness of the
sample. How well does it represent the population of interest, provide diversity in
background, gender, ethnic heritage, and other indicators of important social mean-
ing. Also, is the size adequate to have sufficient sensitivity to effects (see Kim this
volume, Kim & Leventhal, 2015). Sample size has received the most attention
because of the interaction focus and its attending inflation of sample sizes from
those commonly obtained in developmental research. Also as genome-wide analy-
ses have been undertaken, large samples are needed to detect multiple effects on a
small portion of the sample (Kim, 2017). However, an additional consideration is
sample validity. For example, clinical samples have different base rates of disorder
than community samples and are often chosen with an emphasis on homogeneity of
at least disorder presentation. The implications of findings among a population with
100% rates of disorder are quite different from those chosen to represent a defined
community population where the rates are likely to be 10% or below for a given
disorder and even smaller for more specifically defined symptoms, behaviors, or
problematic outcomes.
For intervention studies, a critical consideration is the comparability of intervention
and comparison condition samples with respect to demographics, risk, motivation, and
exposure to developmental conditions in which interventions occur and can affect out-
comes over time (Farrell etal., 2013). This is the cornerstone for valid intervention
research and the reason the randomized control trial is so valuable (Tolan et al., 2013).
Also of importance is determining the appropriateness of the sample for the population
for which generalization is sought. Not only is comparability critical, the meaning of
intervention research rests on how samples are selected and included. For example,
many prevention studies target whole populations with the intent to affect the modest
portions likely to eventuate the problems. However, this is not the same as targeting
high-risk samples (Brody, this volume). A preventive effort with only high-risk partici-
pants is going to have different components, different base rates, and different expecta-
tions about how intervention relates to likely effect sizes (Farrell etal., 2013). When
cast within a GEX framework the role of genetic risk and of how that relates to the
identified population can require careful analysis and thoughtful specification.
Ethical considerations of GEX approaches to intervention. In addition to the sampling
issues outlined, attention to sampling helps highlight the importance of very thought-
ful attention to how GEX studies are undertaken and to be used (Resnik, 1998). The
history of genetic terms and asserted differences to justify prejudice, oppression, and
attempted genocide calls for careful consideration of the potential misunderstanding
and misuse of intervention studies (Moffitt etal., 2005; Radder, 2009). While inter-
vention or experimental studies are among the most powerful methods of advancing
scientific knowledge, a connotation of attempted social manipulation or of selection
to oppress or exclude can be easily attached to this work. As Brody notes (this vol-
ume) there is great need to differentiate the potential benefits of models that incorpo-
rate sophisticated geneticenvironmental interdependencies but not for providing
opportunities to suggest one group or persons with a given characteristic are who
should be targeted. However, simplified rhetorical prohibitions do not serve well
either. As models of causality incorporate genetic contributors, more precise interven-
tions are one potential outcome that could be very valuable. Careful consideration of
the social and historical context of such work needs to be constant considerations in
research design and interpretation, with strong adherence to developed ethical prin-
ciples and operating requirements. This is true with regard to general understanding
and in light of historical precedents of misuse (Radder, 2009).
ooking Forward: TheRole ofIntervention Research

L
intheGEX Framework
Some of the most striking aspects of considering the role of intervention research
from and within a GEX framework are the dynamism of the empirical knowledge,
the theoretical models for organizing that knowledge, and the interplay of these with
methodological issues of developmental study and experimental tests. Intervention

research is one of the most powerful methods of scientific study available, and as
shown in this volume, there are many examples of how this method can contribute
to advancing knowledge about GeneEnvironment interplay and to advancing
healthier development.
Two major types of contributions are presented here and constitute how interven-
tions might help with the scientific advance of the GEX framework and understand-
ing of developmental psychopathology. First, is the use of experimental manipulation
to test theorized processes or components of psychopathology influence. In animal
studies this can include study of how purposeful manipulation of risk characteristics
and environmental manipulations can provide insights into GEX modeling overall,
but also specific mechanisms, developmental timing effects, and epigenetic pro-
cesses. Ethical considerations preclude such manipulations in human studies, but
interventions can be provided that are thought to be palliative, remedying, or pre-
ventive in relation to particular psychopathologies, comprising symptoms, or com-
ponent endophenotypes and psychological and social processes of influence. Such
interventions can be studied for support of impact as theorized within a GEX model
and also for differential effects as theorized based on genetic variations. This
approach is exemplified in the reports provided here and elsewhere regarding testing
variation in reaction to an intervention based on gene allele variation. More recently,
examination of differences in gene expression processes (e.g., telomere alterations)
has been incorporated, providing glimpses about epigenetic phenomena.
Both types of intervention studies have great potential not only in experimentally
testing advances gained from genome-wide scans, candidate gene variation, and
behavioral genetics pattern tracking, but also in offering insights that can spur on
the next generation of theory about geneenvironment transactions. Another
advance could be better methods for determining for which subgroups a given inter-
vention is effective, and for whom it is not useful. If the goal of precision medicine
is to provide the correct intervention for persons based on their particular physiol-
ogy and psychology, then interventions are critical for differentiating such specifics
and for determining how precise geneenvironment interdependencies are. These
studies also provide stronger guidance about distribution of resources for support-
ing health development and treating maladies, if operating with a very strong sensi-
tivity about potential discrimination and misuseeither exclusion for those in need
and for those who do not meet a specific profile, or isolating and stigmatizing those
with such a profile that differentiates response to intervention.
This volume suggests that utilizing interventions as a major part of GEX fram-
ing of developmental psychopathology can be vital for facilitating advancement
of understanding and applying findings to treating and preventing developmental
psychopathology. Yet, the examination brings into relief that what is known at
present is like a temporary and still to be refined organization of knowledge. It is
likely there are important gaps, and rapid and perhaps transformative understand-
ing of elements and processes in GEX transactions are imminent. These suggest
judicious use of intervention research at this point. In addition, the need for greater
articulation of guidelines and suggestions for methodological, substantive, and
ethical considerations are needed, even though several of the chapters in this vol-
ume provide important steps toward those. Many considerations must be inte-
grated for intervention research to best serve this approach to developmental
psychopathology. But, this volume contains many valuable examples of how such
work can be done well and that the way forward for GEX understanding of devel-
opmental psychopathology is to undertake additional, more refined, useful inter-
vention research.
References
Bagot, R.C., Cates, H.M., Purushothaman, I., Lorsch, Z.S., Walker, D.M., Wang, J., Nestler,
E.J. (2016). Circuit-wide transcriptional profiling reveals brain region-specific gene networks
regulating depression susceptibility. Neuron, 90, 969983. doi:10.1016/j.neuron.2016.04.015.
Beauchaine, T.P. (2009). The role of biomarkers and endophenotypes in prevention and treatment
of psychopathological disorders. Biomarkers in Medicine, 3,13. doi:10.1111/j.1540-5834.
2009.00509.
Belsky, J., & Pluess, M. (2009). Beyond diathesis-stress: Differential susceptibility to environmen-
tal influences. Psychological Bulletin, 135(6), 885908.
Brody, G. (2017). Using genetically informed prevention trials to test gene x environment hypoth-
eses. In P. H. Tolan& B. L. Leventhal (Eds.), Advances in development and psychopathology.
Brain Research Foundation Symposium Series, Volume II: Gene-environment transactions in
Developmental Psychopathology: Role of intervention research. New York: Springer.
Bronfenbrenner, U. (1979). The ecology of human development: Experiments by nature and
design. Cambridge, MA: Harvard University Press.
Cannon, T.D., & Keller, M.C. (2006). Endophenotypes in the genetic analyses of mental disor-
ders. Annual Review of Clinical Psychology, 2, 267290. doi:10.1146/annurev.clinpsy.
2.022305.095232
Dodge, K. (2017). Developmental mechanisms in gene by intervention (GxI) effects in the Fast
Track trial. In P. H. Tolan & B. L. Leventhal (Eds.), Advances in development and psychopa-
thology. Brain Research Foundation Symposium Series, Volume II: Gene environment transac-
tions in developmental psychopathology: Role of intervention research. New York: Springer.
Farrell, A.D., Henry, D.B., & Bettencourt, A. (2013). Methodological challenges examining sub-
group differences: Examples from universal school-based youth violence prevention trials.
Prevention Science, 14, 121133.
Frick, P.J., Blair, R.J., & Castellanos, F.X. (2013). Callous-unemotional traits and developmental
pathways to the disruptive behavior disorders. In P. H. Tolan, & B. L. Leventhal (Eds.),
Advances in development and psychopathology. Brain Research Foundation Symposium Series,
Volume I: Disruptive behavior problems (pp.69102). NewYork: Springer.
Gottesman, I., & Gould, T. (2003). The endophenotype concept in psychiatry: Etymology and
strategic intentions. The American Journal of Psychiatry, 160, 636645. doi:10.1176/appi.
ajp.160.4.636.
Kim, Y.S., & Leventhal, B.L. (2015). Genetic epidemiology and insights into interactive genetic
and environmental effects in autism spectrum disorders. Biological Psychiatry, 77, 6674.
Kim, Y. S. (2017). Sampling in the examination of gene-environmental interactions within a neu-
rodevelopmental disorder framework. In P. H. Tolan & B. L. Leventhal (Eds.), Advances in
development and psychopathology. Brain Research Foundation Symposium Series, Volume II:
Gene-environment transactions in Developmental Psychopathology: Role of intervention
research. New York: Springer.
Luan, J.A., Wong, M.Y., Day, N.E., & Wareham, N.J. (2001). Sample size determination for
studies of gene-environment interaction. International Journal of Epidemiology, 30,
10351040.
MacKinnon, D.P., & Pirlott, A.G. (2015). Statistical approaches for enhancing causal interpreta-
tion of the M to Y relation in mediation analysis. Personality and Social Psychology Review,
19, 3043. doi:10.1177/1088868314542878.
McGue, M. & Carey, B. E. (2017). Gene-environment interaction in the behavioral sciences:
Findings, challenges, and prospects. In P.H. Tolan & B.L. Leventhal (Eds.), Advances in devel-
opment and psychopathology. Brain Research Foundation Symposium Series, Volume II: Gene-
environment transactions in developmental psychopathology: Role of intervention research.
Meaney, M.J. (2017). Epigenetics and the biology of gene x environment interactions. In P.H.
Tolan & B.L. Leventhal (Eds.), Advances in development and psychopathology. Brain Research
Foundation Symposium Series, Volume II: Gene-environment transactions in developmental
psychopathology: Role of intervention research.
Moffitt, T. E., Caspi, A., & Rutter, M. (2005). Strategies for investigating interactions between
measured genes and measured environments. Archives of General Psychiatry, 62, 473481.
PMCID: PMC1414673.
Preacher, K.J., Curran, P.J., & Bauer, D.J. (2006). Computational tools for probing interactions
in multiple linear regression, multilevel modeling, and latent curve analysis. Journal of
Educational and Behavioral Statistics, 31, 437448.
Radder, H. (2009). The philosophy of scientific experimentation: a review. Automated
Experimentation, 1, 12. Available and downloaded from http://www.aejournal.net/con-
tent/1/1/2. doi:10.1186/17594499-1-2
Reiss, D. (2017). Genetics, behavioral intervention and human development. In P.H. Tolan & B.L.
Leventhal (Eds.), Advances in development and psychopathology. Brain Research Foundation
Symposium Series, Volume II: Gene-environment transactions in Developmental
Psychopathology: Role of intervention research.
Resnik, D.B. (1998). The ethics of science. London: Routledge.
Soeiro-de-Souza, M., Otaduy, M.C. G., Dias, C.Z., Bio, D.S., Machado-Vieira, R., & Moreno,
R.A. (2012). The impact of the CACNA1C risk allele on limbic structures and facial emotions
recognition in bipolar disorder subjects and healthy controls. Journal of Affective Disorder,
141, 94101. doi:10.1016/j.jad.2012.03.014.
Tolan, P.H., Guerra, N.G., & Kendall, P.C. (1995). A developmental-ecological perspective on antiso-
cial behavior in children and adolescents: Toward a unified risk and intervention framework. Journal
of Consulting and Clinical Psychology, 63(4), 579584. doi:10.1037/0022-006X.63.4.579.
Tolan, P.H., Rutter, M., & Dodge, D. (2013). The multiple pathways of family influence on disrup-
tive behavior problems. In P.Tolan, & B.Leventhal (Eds.), Advances in development and psy-
chopathology. Brain Research Foundation Symposium Series, Volume I: Disruptive behavior
problems (pp.161192). NewYork: Springer.
Tsuang, M.T., Bar, J.L., Stone, W.S., & Faraone, S.V. (2004). Gene-environment interactions in
mental disorders. World Psychiatry., 3, 7383.
Wakschlag, L.S., Tolan, P.H., & Leventhal, B.L. (2010). Research review: Aint Misbehavin:
Towards a developmentally-specified nosology for preschool disruptive behavior. Journal of
Child Psychology and Psychiatry, 51(1), 322. doi:10.1111/j.1469-7610.2009.02184.x.
Willsey, A.J., Sanders, S.J., Li, M., Dong, S., Tebbenkamp, A.T., Muhle, R.A., State, M.W.
(2013). Coexpression networks implicate human Midfetal deep cortical projection neurons in
the pathogenesis of autism. Cell, 155, 9971007.
Index
A Arginine vasopressin gene (AVP), 72, 187, 189

Active rGE, 113, 218 Assortative mating, 98, 101
Adrenocorticotropic hormone (ACTH), 73, 187 Attachment and biobehavioral catch-up (ABC)
Adults in the making (AIM), 211, 212, intervention, 262
216217, 219221, 223, 225, 226 Attention deficit hyperactivity disorder
AIM Randomized Trial, 216217 (ADHD), 41, 59, 131, 184,
Aldehyde dehydrogenase (ALHD2) 190193, 195, 245, 261
polymorphisms, 133 Autism Genome Project (AGP), 162, 163
The American Psychiatric Association Autism spectrum disorder (ASD)
Diagnostic Statistical Manuel, 177 behavioral disorders, 158, 159
Antisocial behavior comorbidities, and biomarkers, 158
correlations with aggressive behavior, 238 dizygotic (DZ), 159
defensive processing patterns, 239 gene-environment interactions, 157
deviancy training, 237 medical disorders, 158
MAOA, high-activity alleles, 227 monozygotic (MZ), 159
nonrandom mating, 101 myriad, 160
parenting aspects, 40 neurodevelopmental disorder, 157
SLC6A4 gene promoter, 185186 non-shared environmental risk, 160
smoking exposure, restricted fetal growth, 20 risk factors, 159
violence and, 236 utero, 160
Anxiety disordersBDNF, 150151 AVP receptor 1A (AVPR1A), 189
candidate gene studies, 148
COMT, 151
differential susceptibility hypothesis, 151 B
genetic and environmental factors, 152 Bacillus subtilus, 10
GxE effects, 147 Behavioral interventions
GWAS, 149 childrens intellectual capacities, 18
hypothesis tests, 148 childs intellectual abilities, 18
neural circuitry, 147 environmental variance, 18
pharmacological treatments, 147, 152 gene-driven positive amplification
psychiatric disorders, 149, 152 process, 18
schizophrenia, 152 gene expression, 1517
SERT, 149, 150 gene methylation/expression, 1718

DOI10.1007/978-3-319-49227-8
296 Index
Behavioral interventions (cont.) mental and physical health professions, 177

groups of Individuals, 1415 nurse home visitation, 191
heritable factors, 18 participation, quality daycare, 191
influence of environments, 18 serotonergic system, 185186
marital process, 19 social and behavioral, 190
mechanisms, 18 Coercion theory, 253254, 256
parallel explanation, 18 Conduct disorder (CD), 177, 183, 188
pleiotropic effects, brain mechanisms, 13 Conduct Problems Prevention Research
quantitative and molecular genetics, 17, 18 Group, 235
selection effects, 19 Continuum of indirectness, 119
sibling-specific environments, 19 Copy number variants (CNVs), 162
timing of interventions, 18 Corticotrophin-releasing factor (CRF), 68,
Behavior Genetics and Psychological Science 72, 73
(Journal), 273 Corticotrophin releasing hormone (CRH), 72,
Birmingham University (UK) childrens 74, 187189
hospital, 10 Cytokines
Brain derived neurotrophic factor (BDNF), aggressive behavior, 187
7883, 150151 pro-inflammatory proteins, 188
Breast-feeding, 194 signalling pathway, 189
Bucharest Early Intervention Project (BEIP), 75 transcription factor genes, 186
B vitamin-dependent folate, 168
D
C Dex/CRH test, 74
Catechol-O-methyltransferase (COMT) gene, Diathesisstress model, 40
82, 151, 168 Dizygotic (DZ) twins, 11, 39, 96, 120, 159
Categorical diagnostic systems, 139 DNA methylation, 168, 186, 187, 189
Child behavioral outcomes, 97, 103, 105 gene transcription, 64, 70
Child Development Project, 235, 238 genomic silencing, 65
Childhood adversity, 7377, 82 HDACs, 64
Childhood mood disorder, 187 5hmC, 64
Child Study Center, 11 inter-individual differences, 74
Chronic physical aggression (CPA) Dopamine receptor 4 (DRD4) gene, 219,
animal and human physical 220, 222
aggression, 178 Dopaminergic and GABAergic candidate
behavior/psychiatric disorder, 190 genes, 222
DSM classification, 178
early childhood to adolescence
age-crime curve, 179 E
antisocial behavior, 178 The Early Growth and Development Study
Canada and United States, 179, 180 (EGDS), 15, 263
chronic behavior, 181 Early Start Denver Model (ESDM), 170
developmental trajectories, 181 Ecological fallacy concept, 135
international data, 182 EEA. See Equal environments assumption
low socioeconomic environments, 180 (EEA)
social learning, aggression Endophenotypes
hypothesis, 178 anxiety-related, 80
environmental risk factors, 183184 gene-environment interactions, 4
epigenetic mechanisms, 185 GE research, 49
experiment, 190 symptoms and syndromes, 286287
genetic and environmental mechanisms, 178 Environmental risk factors
genetic risk factors, 184185 adverse birth events or neonatal distress, 164
GxE framework, 192196 chronic physical aggression, 183184
immune system, 186189 doseresponse curves, 139
Index 297
early behavioral genetic studies, 4 development of psychopathology, 235

genetics, 167169 elementary school, 247
intellectual disability, 164 The Fast Track prevention experiment,
individual genetic and, 159 241246
paternal and maternal age, 164 gene-by-environment effects, 236
VPA, 163 gene-by-intervention effects, 236
Epigenetics genome-wide assessments, 247
cell typing, 65 glucocorticoid receptor gene, 243245
DNA methylation, 6366 growth prediction, 238
environmental regulation, human monoamine oxidase A and gamma-
methylomes aminobutyric acid receptor, 247
allele-specific methylation, 7577 NR3C1 gene, 247
BEIP, 75 social-cognitive process, 247
maternal stress, 75 social information processing mechanisms,
epigenetic marks, 61 237239
functional genomes, 65 Fetal alcohol syndrome (FAS), 134
GE interactions (see Geneenvironment Folate metabolism, 168
(GE) interaction research) Fragile X syndrome (FXS), 161, 165, 166, 169
gene expression, 65
genomic silencing, 65
GR expression and function, humans, 73, 74 G
histone modifications, 6163 Gamma-aminobutyric acid receptor subunit
monozygotic twins, 65 alpha-2 (GABRA2), 240241
nucleosome structure, chromatin, 61, 62 Gene-by-intervention (GxI) effects, 241
Equal environments assumption (EEA), 99, Geneenvironment correlation (rGE), 218, 260
102, 103 active, 113
Evocative rGE, 104, 113, 260262 behavior genetics, 111
childhood intervention, 126127
Commonwealth of Virginia, 111
F developmental canalization hypothesis, 119
False-positive findings, 49, 132, 135 environment transmission, 123126
Family-based behavioral genetics evocative, 113
child and parent heritable influences, 103 family environmental variables and
combination designs, 96 childhood outcomes, 122
conceptual issues, 9798 genetic processes, 119
degree of bias, 101 genetic traits, expression, 112
EEA, 102 genotypes, 121
heritable and environmental effects, 97 heritability increases with age, 114117
heritable influences, role in, 103 heritable characteristics, 113
intervention research, 104, 105 human consciousness, 118
MZ twin pairs, genomes of, 102 innovation and amplification, 122
parentoffspring adoption, 96, 101 interactionism, 113
prenatal sharing, 102 (see also Gene life span, 118
environment interplay) meta-analysis, 121
physical and psychological nonrandom environmental exposures, 112
characteristics, 101 passive, 113
shared and nonshared environmental, 96 phenotypes, 120
technical issues, 98100 physical facilities, indifferent coaching,
twin/sibling studies, 96 and long record of losing, 112
Family skills training programs, 212 proximal processes, 121
Fast track (FT) trial quantitative genetics, 112
aggressive behavior, 238 Scarr and McCartneys model, 120
chronic externalizing disorder, 236237 solenoid starters, 119
defensive processing patterns, 239 training programs, 113
298 Index
Geneenvironment (GE) interaction RNA expression/gene methylation, 169

research, 133136 scale transformation, 46
acetylation, 71 SLC6A4 gene, 77
acrimonious NatureNurture debate, 35 social control model, 36
additive vs. multiplicative model, 46 statistical model, 45
allele-specific methylation, 7577 schizophrenia, 170
animal models, 170 tailoring prevention efforts, 38
approach, 4041 TBI, 39
assessment, 4849 Thomson-Reuters Web of Science, 37
AVP expression, 72 TSA treatment, 72
behavioral and endocrine responses, Geneenvironment interplay
stress, 67 attention to mean-level improvements,
behavioral genetic literature, 36 258260
behavioral sciences, 37 behavioral genetics approach, 252,
biological pathways, 38 275277
biometricians seek, 36 behavioral health outcomes, 251
candidate-gene research, 51, 169 conceptual theory, 253256
childhood maltreatment, 38 description, 100
complex behavioral phenotypes, 35 epigenetic approach, 100
CRF expression, 68, 72, 73 evidence-based psychosocial
cross-over/qualitative interaction, 46 interventions, 252
diathesisstress model, 36 gene expression, 100
environment conditions, early life, 67 genetic variations, 252
ESDM, 170 health promotion, 256258
folate and vitamin B12, 170 in human populations, 100
genetic and environmental factors, 35, 170 hypothesized outcomes, 270271
genomics research, 38 long-term follow-up data, 268270
glucocorticoid receptor (GR), 68 mediators identification, 260262
GWAS, 170 medicine approach, 251
high-support environment, 46 molecular genetic studies, 273275
H3K9ac and H3K4me3, 71 personalized medicine approach, 251, 252
HPA response, 67, 68 precision medicine approach, 252
increasing sample size, 4748 process orientation, 262264
individualized/genomic medicine, 50 replication of effects, 272273
influence of environment, 165167 risk distribution, 264266
interpregnancy interval/maternal sustainability and dissemination, 266268
diabetes, 169 systematic variations, 100
intervention context, 51 translational approach, 253
IQ heritability, 67 variance and modeling GxEs, 100
latent-variable approach, 3940 Geneenvironment transaction (GEX)
leveraging GWAS findings, 4950 clinical interventions, 3
measurement error, 45 complex transactions, 283
methylenetetrahydrofolate reductase, 170 developmental psychopathology, 1, 2,
multiplicative scale, 45 283, 284
NGFI-A, 70 endophenotypes, 286287
norm of reaction, 66 environmental influences, 287288
Nr3c1 gene, 68 experimental/intervention studies, 3, 284
pharmacologic treatment of depression, gene variant basis and networks, 285
51, 171 G-E transaction frame, 3
plasticity genes, 7882 human development, 284
power, 4245 intervention research, 2, 45, 291293
prevention/intervention strategies, 171 measurement reliability and validity, 290
psychology and psychiatry, 36 mediational models, 3
quantitative phenotype, 46 methodological considerations, 285
Index 299
psychosocial interventions, 284 ferric chloride, 10

sample implications, 290291 follow-up studies, 12
susceptibility to environmental genetic influence, 11
influence, 286 genetic inquiry, 12
Gene expression groups of individuals
behavioral therapeutic and preventive adolescent smokers, 14
interventions, 16 child genotype, 15
child psychopathology, 17 family-oriented prevention, 14
cognitive behavior, 16 genetic literature, 15
early family interventions, 17 infants and toddlers, 15
environmental process, 16 polygenic risk score, 14
fMRI, 16 psychosocial interventions, 15
genetic allelic variation, 16 public health oriented interventions, 14
illegal substance use, 17 quantitative genetics, 14
longitudinal studies, 16 self-regulation, 14
marital dissatisfaction, 17 serotonin transporter gene
marriage-oriented intervention, 17 polymorphism, 14
mediation procedures, 16 smoking, 14
mental disorders, 16 social and emotional development, 15
methylation and expression, mRNA, 15 substantial influences, 15
molecular, 15 heterozygote parents, 12
outside skin mechanisms, 17 human development, 9
peripheral blood, 16 low phenylalanine diet, 10
stress-induced depression, 16 mental abilities, 11
white cells and brain, 15 MZ and DZ twins, 11, 13
Gene intervention nicotine withdrawal symptoms, 12
AIMs efficacy, 226 pharmacological and behavioral
alcohol and drugs, 211 therapeutics, 13
discrimination and stigmatization, 228 phenyl pyruvic acid, 10
environmental risk mechanisms, 227 phenylalanine challenge test, 10, 13
epidemiological data, 211 post-World War II revulsion, 11
etiological models, 217 rate of learning, 12
family interventions, 212 scientific genetics, 9
family skills training programs, 212 selective breeding, 9
first-generation GI research, 219, 220, slopes of performance, 12
222, 229 worldwide revulsion, 11
genomic and environmental effects, 227 Genetic risk factors
genotypic targeting, 228229 accelerating array, 161
5-HTTLPR and antisocial behavior, 227 ascertainment bias, 169
SAAF, 211 CNVs, 162
second-generation GI research, 224, 225 co-occurring medical conditions, 161
targeted interventions and prevention DNA methylation and chromatin
paradox, 228 modification, 167
universal and selective preventive epigenetics, 167
interventions, 229 FXS, 161
vulnerability cognitions, 226 LGD, 162
Genetic models MET gene promoter, 168
ancients, 9 pregnancy, 168
Bacillus subtilus, 10 vitamins and one-carbon metabolism, 168
behavioral interventions, 12 Genome-wide approach, 189
complex syndrome, 11 Genome-wide association studies (GWAS),
effective dietary treatment, 10 42, 59, 60, 149
environmental influences, 13 Genome-wide GxE studies (GWIS), 152
experiments, 12 German dermatologist Herman Siemens, 11
300 Index
Glucocorticoid receptor (GR) gene, 187, Monoamine oxidase A (MAOA), 60, 184186,
243246 188, 189, 227, 239240, 247
Glucocorticoids, 6770, 77, 188, 243 Monozygotic (MZ) twins, 159
GE. See Geneenvironment (GE) Montreal longitudinal-experimental study, 179
interaction research
GEX. See Geneenvironment transaction
(GEX) N
National Research Council, 2013, 11
Nerve-growth factor-induced factor A
H (NGFI-A), 68, 70
Histones acetylation, 189 Neurodevelopmental disorders (NDDs)
5-HTTLPR polymorphism, 185, 219 ALHD2 polymorphisms and alcohol
5-Hydroxymethylcytosine (5hmC), 64 exposure, 133
Hypothalamic-pituitary-adrenal (HPA) axis, 67, ASD, ID and ADHD, 131
68, 7174, 80, 187189, 194, 243 candidate gene approach, 135
case-control studies, 132, 133
description, 131
I developmental psychopathology,
Intraclass correlations (ICCs), 115 137140
ecological fallacy concept, 135
FAS, 134
J intervention research, 133
Journal of Abnormal Child Psychologys methodological challenges, 135
publication policy, 273 population stratification, 134
RCT, 136137
rGE and GxE, 131
K sampling appropriateness, 132
Keeping Foster Parents Trained and statistical tests, 134, 135
Supported (KEEP) intervention Neurofibromatosis, 161
program, 260 Nonrandom mating, 98, 101
L P
Latent growth modeling (LGM), 217 Parenting
Likely gene-disrupting (LGD), 162 child behavioral outcomes, 17, 97
Louisville Twin Study (LTS), 116 child and parent heritable influences, 103
early risk factors, 183
family-based behavioral genetic
M designs, 103
Major depressive disorder diagnosis punitive disciplinary practices, 40
(MDD), 268 Parentoffspring adoption designs, 96
Mammalian target of rapamycin (mTor) Passive rGE, 113, 218
pathway, 161 Peripheral blood cells DNA, 186
Maternal deprivation, 187 Phenyl pyruvic acid, 10
Mental health Plasticity genes, 60, 7782
childhood adversity, 82 Population stratification, 134
neurons, 66 Posttraumatic stress disorder (PTSD),
plasticity genes, 78 187, 255
variability, 83 Precision Medicine Initiative, 251
Methionine, 79, 168 Prenatal anxiety, 187
Methylenetetrahydrofolate reductase Promoting Alternative Thinking Strategies
(MTHFR), 168, 170 (PATHS), 256
Mineralocorticoid receptor, 187 Psychiatric genetics, 49
Moderator, 243245 PTEN hamartoma syndrome, 161
Index 301
Q caregivers, 213
Qubec Suicide Brain Bank, 73 hypotheses, 213
intent-to-treat analysis, 213
intervention group, 213
R prevention trials, 213
Randomized control trials, 192
Rapamycin, 161
Reciprocal effect models, 124 T
Robert Guthrie a physician and Thalidomide, 163, 167
microbiologist, 10 Thomson-Reuters Web of Science, 37
Rubella virus, 163 Transcobalamin II (TCN2), 168
Transcription factor genes, 186
Transmethylation pathways, 168
S Traumatic brain injury (TBI), 38, 39
SAAFT randomized trial, 214, 215 Trichostatin A (TSA), 71, 72
Schizophrenia, 59, 131, 152, 170, 191, 289 Tuberous sclerosis (TSC), 161, 166, 167
gene-environment interactions, 170 Twin Research and Human Genetics, 273
Mauritius Child Health Project, 191
neurodevelopmental pathophysiology, 131
Serotonergic system, 184186, 193 U
Serotonin receptor 1D (HTR1D), 189 University of Minnesota, 12
Serotonin transporter polymorphisms (SERT), Upstream variable number of tandem repeat
149150 (uVNTR), 239
Simons Simplex Collection (SSC), 162, 163 US Academy of Science Panel on
Single nucleotide polymorphisms (SNPs), 23, Understanding Violent Behavior, 178
42, 43, 45, 75, 76, 82, 148, 244
Single nucleotide variants (SNVs), 162
SLC6A3, dopamine transporter, 189 V
SLC6A4, gene promoter, 186 Valproic acid (VPA), 163, 167
Society for Prevention Research, 2011, Variable nucleotide polymorphism (VNTR), 219
253, 256 Variably methylated regions (VMRs), 76
The Special Curriculum Opportunity Rating Vasopressin, 72, 187
Scale (SCORS), 166
Strong African American Families (SAAF)
program, 211 W
alcohol usage, 214 Wilson effect, 116, 120
analyses, 214 World Health Organisation (WHO), 178

(Advances in Development and Psychopathology_ Brain Research Foundation Symposium Series 2) Patrick H. Tolan, Bennett L. Leventhal (eds.)-Gene-Environment Transactions in Developmental Psychopathology.pdf

Hochgeladen von

Dokumentinformationen

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

(Advances in Development and Psychopathology_ Brain Research Foundation Symposium Series 2) Patrick H. Tolan, Bennett L. Leventhal (eds.)-Gene-Environment Transactions in Developmental Psychopathology.pdf

Hochgeladen von

Copyright:

Verfügbare Formate

Advances in Development and Psychopathology:

Brain Research Foundation Symposium Series

Series Editors: Patrick H.Tolan and Bennett L.Leventhal

More information about this series at http://www.springer.com/series/8544

Advances in Development and Psychopathology: Brain Research Foundation

Library of Congress Control Number: 2017932277

Springer International Publishing AG 2017

Printed on acid-free paper

This Springer imprint is published by Springer Nature

1 What Can andShould BetheRole ofIntervention Studies

Part I Gene-Environment Transaction Framework

Part II Application to Developmental Psychopathology

9 Challenge and Potential for Research on Gene-Environment

Part IIIIntervention Research: Implications for Gene

ChristopherR.Beam University of Southern California, Los Angeles, CA, USA

P.H. Tolan (*)

Springer International Publishing AG 2017 1

A second major focus of these discussions includes the notion of interventions

 ene-Environmental Transaction Framework

Radder, H. (2009). The philosophy of scientific experimentation: A review. Automated

Springer International Publishing AG 2017 9

Defining Targets forBehavioral Intervention

Defining Groups ofIndividuals at Risk forDisease

Delineating Malleable Mechanisms ofGene Expression

Defining theTiming ofBehavioral Interventions

There is increasing interest in integrating both quantitative and molecular genetics

Genetic Influences onResponse toTreatment

A Few Ideas About theFuture

Future studies of genetic moderation of behavioral therapies need to be prospective and

Genetic inquiry has stimulated a strong interest in linking studies of behavioral

developments in gene expression assays currently attract monumental interest and

tex and T cells. The Journal of Neuroscience, 32(44), 1562615642. doi:10.1523/

What is GeneEnvironment Interaction?

M. McGue (*) B.E. Carey

Springer International Publishing AG 2017 35

Fig. 3.1 Alternative forms of genotypeenvironment interaction. In the diathesisstress model

genetic risk is minimized in environments in which the individuals capacity to

 hy is GenotypeEnvironment Interaction Considered

complex phenotype. Carriers of the 4 allele appear to be especially vulnerable to the

Methods forDetecting GenotypeEnvironment Interactions

GE occurs whenever the effect of the genotype is conditional on environmental

Candidate-Gene Environment Approach

The latent-variable gene environment interaction approach neither postulates nor

a target of considerable research on the genetics of aggression, on the X chromo

Issues withtheCandidate GE Approach

Before discussing concerns surrounding the power of cGE studies, it is informative

Table 3.1 Effect sizes reported in large-scale genome-wide association studies

Power in a cGE study is a complex function of the distribution and magnitude

Measurement error in any of the components of a cGE study (i.e., phenotype,

GE is Not Scale Invariant

A GE is the result of applying a particular statistical model to a set of data. As

Addressing theLimitations ofGE Research

Increasing Sample Size

been limited attempts to standardize the assessment of environmental risk (Hamilton

Alternative Approaches toAssessment

The sensitivity of tests for GE could be improved by increasing the accuracy of

preferentially sampling individuals at extreme levels of environmental exposure by,

Leveraging GWAS Findings

Prospects forDetecting GE inIntervention Research

Individualized or genomic medicine is premised on the existence of GE, specifi

The modern genomics era initially emphasized studies of genotypephenotype rela-

M.J. Meaney (*)

Springer International Publishing AG 2017 59

opposed to statistical inference based on estimates of heritability. GWAS is a start-

Part IIIIntervention Research: Implications for Gene

ene-Environmental Transaction Framework

Defining Targets forBehavioral Intervention

Defining Groups ofIndividuals at Risk forDisease

Delineating Malleable Mechanisms ofGene Expression

Defining theTiming ofBehavioral Interventions

Genetic Influences onResponse toTreatment

A Few Ideas About theFuture

What is GeneEnvironment Interaction?

hy is GenotypeEnvironment Interaction Considered

Methods forDetecting GenotypeEnvironment Interactions

Candidate-Gene Environment Approach

Issues withtheCandidate GE Approach

GE is Not Scale Invariant

Addressing theLimitations ofGE Research

Increasing Sample Size

Alternative Approaches toAssessment

Leveraging GWAS Findings

Prospects forDetecting GE inIntervention Research

Epigenetics andtheBiological Basis ofG E Interactions

Epigenetics andtheBiology ofG E Interactions

pigenetic Regulation ofGR Expression andFunction

Environmental Regulation oftheHuman Methylome

he Epigenome asaFunction ofGene Environment

Gene x Environment Effects: Plasticity Genes?

Implications forPrevention Programmes

Family-Based Behavioral Genetic Designs

Challenges toFamily-Based Behavioral Genetic Designs

he Continued Value ofFamily-Based Behavioral Genetic

Response toConceptual Criticisms

Response toTechnical Criticisms

The Value ofFamily-Based Behavioral Genetic Studies

he Implications ofFamily-Based Behavioral Genetic Studies

Conclusions andFuture Directions

A Brief but Successful Baseball Career

If Monkeys Could Play Baseball

Plomin, Loehlin andDeFries (1977)

Heritability Increases withAge

The Question ofHow

henotype toEnvironment Transmission andGene

Applications toChildhood Intervention

ethodological Issues inPrevious GxE Studies ofNDDs,

CT Strategies toExplore theRoles ofGxE intheEtiology