Page 1

This electronic thesis or dissertation has been

downloaded from the King’s Research Portal at

https://kclpure.kcl.ac.uk/portal/

Take down policy

If you believe that this document breaches copyright please contact librarypure@kcl.ac.uk providing

details, and we will remove access to the work immediately and investigate your claim.

END USER LICENCE AGREEMENT

Unless another licence is stated on the immediately following page this work is licensed

under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International

licence. https://creativecommons.org/licenses/by-nc-nd/4.0/

You are free to copy, distribute and transmit the work

Under the following conditions:

•

Attribution: You must attribute the work in the manner specified by the author (but not in any

way that suggests that they endorse you or your use of the work).

•

Non Commercial: You may not use this work for commercial purposes.

•

No Derivative Works - You may not alter, transform, or build upon this work.

Any of these conditions can be waived if you receive permission from the author. Your fair dealings and

other rights are in no way affected by the above.

The copyright of this thesis rests with the author and no quotation from it or information derived from it

may be published without proper acknowledgement.

Leveraging genome-wide data to understand the risk and treatment of common mental

disorders

ter Kuile, Abigail

Awarding institution:

King's College London

Download date: 11. Nov. 2023

Leveraging genome-wide data to

understand the risk and treatment

of common mental disorders

Abigail Roos ter Kuile

Social, Genetic and Developmental Psychiatry Centre

Institute of Psychiatry, Psychology and Neuroscience

King's College London

This thesis is submitted for the degree of

Doctor of Philosophy in Statistical Genetics

Supervised by Professor Thalia Eley and Professor Gerome

Breen

March 2023

2

Abstract

Common mental disorders, including anxiety and depression, have a detrimental impact on

global disease burden and quality of life. Both the risk and treatment of anxiety and depression

are associated with the complex interplay between genetic, psychological, and socio-

environmental factors. Integrating these factors has posed a challenge in research efforts

aimed at improving the prevention and treatment of these disorders. The recent growth of

genomic data and advancements in multi-trait methodology allows for the incorporation of

genetics with psycho-social factors at an unprecedented level. This thesis applies statistical

genetic approaches to further understand the genetic influences on the risk and resolution of

common mental disorders. The first empirical study (Chapter 2) presents the largest genome-

wide association study (GWAS) meta-analysis of lifetime fear-based anxiety disorders (N total

=188,812; N cases =30,861) and examines the shared and distinct genetic relationship with

generalised anxiety disorder (N total =172,248; N cases =54,928) and broad domains of other

complex traits. Chapter 3 builds on previous research finding a genetic component of reported

trauma, a major socio-environmental risk factor for anxiety and depression. By leveraging pre-

existing GWAS summary statistics, genetic correlations and genomic multiple regression

analyses are used to identify heritable psychological and behavioural traits that capture the

common genetic variant-based heritability of reported childhood maltreatment (N=185,414).

Chapter 4 represents the largest GWAS meta-analysis of outcomes following psychological

treatment for anxiety and depression (N total =15,131; N reported positive outcomes =11,408).

The utility of genetic factors is also assessed by incorporating polygenic scores of complex

traits into multivariable prediction models of psychological treatment outcomes alongside

known clinical and demographic predictors. The final chapter discusses the implications of the

findings from this thesis within the context of challenges emerging in anxiety and depression

genomics. With the ongoing expansion of GWAS data, consideration should be taken into how

phenotyping approaches influence downstream analyses, including the genetic structure

observed across psychopathologies and the psycho-social components involved in gene-

environment interplay.

3

Table of contents

Abstract ...................................................................................................................... 2

Table of contents ...................................................................................................... 3

List of tables and figures .......................................................................................... 5

Acknowledgments .................................................................................................... 7

Statement of authorship ........................................................................................... 8

Publications related to this thesis ......................................................................... 10

Chapter 1. General introduction ............................................................................ 12

Anxiety and depression comorbidity is extensive and associated with adverse outcomes

........................................................................................................................................... 12

Research into understanding drivers of comorbidity whilst accounting for clinical

heterogeneity ................................................................................................................. 14

Transdiagnostic and disorder-specific genetic influences ................................................. 16

Twin study findings ........................................................................................................ 16

From twin studies to genome-wide methods to understand biological aetiology ........... 19

Progress in depression and anxiety disorder genomics ................................................ 26

Leveraging genomic data to understand pervasive comorbidity and clinical heterogeneity

....................................................................................................................................... 28

Social environmental influences on anxiety and depression ............................................. 31

The heritability of environmental measures ................................................................... 32

Heritable psychological influences on anxiety and depression: from risk to treatment ..... 34

Heritable influences on subjective self-reports of adverse experiences ........................ 34

Heritable influences on positive experiences and application to psychological treatment

....................................................................................................................................... 35

Summary and aims ............................................................................................................ 38

References ........................................................................................................................ 39

Chapter 2. Exploring the genome-wide genetic overlap between anxiety and

fear disorders .......................................................................................................... 51

Abstract ............................................................................................................................. 52

Introduction ........................................................................................................................ 53

Methods ............................................................................................................................. 56

Results ............................................................................................................................... 62

Discussion ......................................................................................................................... 71

References ........................................................................................................................ 76

Chapter 3. Genetic decomposition of the heritable component of reported

childhood maltreatment ......................................................................................... 81

Abstract ............................................................................................................................. 82

Introduction ........................................................................................................................ 83

Methods and Materials ...................................................................................................... 86

Results ............................................................................................................................... 88

Discussion ......................................................................................................................... 93

4

References ........................................................................................................................ 99

Chapter 4. Genetic influences on self-reported outcomes following

psychological treatment for depression and anxiety disorders ....................... 103

Abstract ........................................................................................................................... 104

Introduction ...................................................................................................................... 105

Methods ........................................................................................................................... 109

Results ............................................................................................................................. 117

Discussion ....................................................................................................................... 123

References ...................................................................................................................... 129

Chapter 5. General discussion ............................................................................ 134

General overview of findings ........................................................................................... 134

Theme 1: Phenotyping depth and the utility of self-reported measures .......................... 137

Theme 2: the genetic structure of internalising disorders and implications for disorder

grouping ........................................................................................................................... 139

Theme 3: interpreting GWAS findings in the context of psychosocial influences ............ 143

Personality, behaviour and cognitive biases ............................................................... 144

Socioeconomic status .................................................................................................. 146

General Limitations .......................................................................................................... 147

Lack of replication and sources of bias ....................................................................... 147

Lack of ancestral diversity ........................................................................................... 149

Future directions .............................................................................................................. 150

Establishing causality, and disentangling individual from familial genomic effects ..... 150

Expanding and integrating genome-wide data ............................................................ 152

Conclusions ..................................................................................................................... 153

References ...................................................................................................................... 154

Appendix A. Supplementary materials for Chapter 2 ........................................ 160

Appendix B. Supplementary materials for Chapter 3 ........................................ 211

Appendix C. Supplementary materials for Chapter 4 ........................................ 241

5

List of tables and figures

Chapter 1

Figure 1: An overview of the three empirical chapters presented in this thesis and the

overlap among genomic, socio-environmental and psychological factors. ............................ 16

Box 1: Definitions and approaches to estimate trait heritability ............................................. 18

Box 2: Definitions of types and consequences of human genetic variation ........................... 20

Box 3: Structural properties of the human genome and related population genetic concepts

leveraged in genome-wide analyses ..................................................................................... 21

Figure 2: An illustration of the various stages of genome-wide analyses. ............................. 23

Box 4: Definitions of pleiotropy, bivariate and multi-trait genetic analyses ............................ 24

Figure 3: An illustration of the genomic continuum of the mood disorder spectrum from

mania genomics to internalising genomics. ........................................................................... 30

Figure 4: An illustration of the potential mechanisms underpinning the course and resolution

of common mental disorders, which are influenced by the interplay between genetic, socio-

environmental and psychological factors. .............................................................................. 31

Chapter 2

Table 1. Sample sizes in each study, GWAS dataset, and meta-analysis of fear-based

disorders and GAD. ............................................................................................................... 59

Figure 1: Genome-wide association study Manhattan plots for anxiety disorder phenotypes

meta-analysed across the GLAD+, QIMR and UKB datasets. .............................................. 63

Table 2: Independent genome-wide significant loci associated with anxiety disorder

phenotypes ............................................................................................................................ 64

Table 3: Gene-level associations with anxiety disorder phenotypes. .................................... 67

Table 4: Genetic correlations (rg ) between fear-based disorder and GAD phenotypes ....... 69

Figure 2: Genetic correlations between anxiety disorder phenotypes and external traits

estimated in LDSC regression. .............................................................................................. 70

Chapter 3

Figure 1. Top bivariate genetic correlations (rg) between reported childhood maltreatment

and various heritable traits. ................................................................................................... 89

Figure 2. Path diagrams representing results from genomic multiple regression analyses. . 91

6

Chapter 4

Table 1. Number of participants from the GLAD+ and AGDS datasets with data on each

measures of self-reported outcomes following therapy (Total N = 15,131) ......................... 112

Table 2. Descriptive table for GLAD+ and AGDS datasets, total N = 15,131 ...................... 117

Table 3. LDSC SNP-based heritability estimates (using summary-level data) in GLAD+ and

AGDS datasets, total N = 15,131 ........................................................................................ 118

Table 4. GCTA-GREML SNP-based heritability estimates (using individual-level data) in

GLAD+ and AGDS datasets ................................................................................................ 118

Figure 1: Q-Q plot (left) and Manhattan plot (right) of associations from a meta-analysis of

retrospectively self-reported treatment outcomes following psychological therapy (N =

15,131; 75% reported positive outcomes). .......................................................................... 119

Table 5. Univariable associations between predictors and retrospectively self-reported

treatment outcomes in the GLAD+ dataset (N = 4,439) ...................................................... 121

Table 6. Models predicting self-reported outcomes following psychological therapy for

depression/anxiety using genetic, sociodemographic and clinical predictors in the GLAD+

dataset (N = 4,439) .............................................................................................................. 122

Chapter 5

Figure 1: An illustration of the emergence of three key themes based on the findings of this

thesis and the way they relate to various stages of GWAS. ................................................ 137

Figure 2: An illustration of the genetic continuum across the mood disorder spectrum to the

internalising disorder spectrum, with the addition of fear and GAD. .................................... 141

7

Acknowledgments

Thank you to the study participants who shared their life experiences, and all the individuals

involved in designing and managing the studies and data, without whom the research

presented in this thesis would not have been possible. Additionally, I am thankful to the NIHR

Maudsley Biomedical Research Centre for providing me with the opportunity and funding to

pursue this PhD.

I am deeply grateful to my supervisors, Thalia and Gerome, for their unwavering support and

guidance. Your mentorship has provided me with numerous exciting opportunities to grow as

a researcher and has been instrumental in shaping my research and enabling me to achieve

my academic goals. To Thalia, thank you for your kindness, encouragement, and especially

your support during the challenging period when I was unwell with long covid. You gave me

invaluable insight to see the bigger picture of my research and communicate my findings more

broadly, which I will take with me throughout my career. To Gerome, thank you for helping me

find my passion and introducing me to the world of psychiatric genetics. You consistently had

faith in my ability to tackle challenges and encouraged me to persevere. I am thankful for the

opportunity to work with you both.

I feel fortunate to have had the opportunity to work at the SGDP Centre and meet so many

wonderful people. Despite the pandemic, the support and sense of community that I

experienced (even online) was truly special. I am grateful to my colleagues in the EDIT and

TNG research teams for being a part of my journey. Working alongside such a friendly,

sociable, and supportive group has been an enriching and rewarding experience.

A special thank you to Jess, Katie, Helena, Alicia, and Anna, whose friendship and

collaboration have been a highlight of my PhD journey. You all have been an inspiration to me

and have lifted me up during the most challenging times. Your support and encouragement

have helped me grow as a scientist and as a person, and I have learned so much from each

of you. Thank you for being there to celebrate every achievement and for encouraging me to

be as proud of myself as I am of each of you.

To my dear friends, and especially, my housemates at Greig and Sandlings, thank you for

taking care of me, reminding me to take breaks, and ensuring that I went outside to get some

fresh air during periods of intense focus. Your kindness and positivity were crucial in helping

me maintain my wellbeing.

I am also deeply grateful to my wonderful family - my Mum, Dad, and Naomi. Your unwavering

love, support, and inspiration have brought me to where I am today. We spent a considerable

portion of my PhD in lockdown together which I will always look back on fondly. Your care and

encouragement when I was unwell with long covid helped me to pace myself until I was back

to my normal self, enabling me to complete this hurdle.

Finally, to Oscar. I can’t thank you enough for everything. You have been my rock and

sounding board, providing me with love, encouragement, patience, and support every step of

the way. You have kept me grounded and reminded me that there is life outside of the PhD

when I needed it most. Thank you for being by my side throughout this journey.

8

Statement of authorship

All the work presented in this thesis is my own except where acknowledged in the text. Data

collection for all study samples was completed by the respective research teams. Genetic data

quality control (QC) was carried out by respective studies. In Chapters 2 and 4, I conducted

additional QC in subsamples of the GLAD+ datasets (Genetic Link to Anxiety and Depression

(GLAD) study and the COVID-19 Psychiatric and Neurological Genetics (COPING) study).

During the duration of this PhD, I was involved in setting up and maintaining three pipelines in

collaboration with colleagues to generate repositories of data used in this thesis and for future

studies. First, I contributed to the phenotypic data cleaning pipeline of the GLAD+ dataset

(used in Chapters 2 and 4). Second, I helped maintain and update a GWAS summary

statistics QC pipeline and repository of hundreds of complex traits (used in Chapters 2, 3 and

4). Third, using the GWAS summary statistics from this repository, I contributed to setting up

a polygenic score repository in the GLAD+ dataset (Chapter 4).

The studies presented in Chapters 2 and 3 were conceived and conducted by me as first

author (A.R.T.K), in collaboration with colleagues included in the author list presented at the

start of each chapter. The work presented in Chapter 4 was a joint first author study conducted

with Alicia J. Peel, reflecting equal contributions from myself and A.J.P. I contributed to the

work in Chapter 4 by leading on genome-wide association analyses in the GLAD+ dataset,

GWAS meta-analysis across datasets, post-GWAS analyses, genetic quality control

procedures and additionally the creation of polygenic risk scores for prediction modelling, with

input from A.J.P. I assisted A.J.P who led on preparation of phenotypic data, prediction

modelling, preparing the pre-registration and an original draft of the manuscript. Chapters 2

and 4 include GWAS meta-analyses that were conducted in collaboration with researchers

involved in the Australian Genetics of Depression study (AGDS) based at Queensland Institute

of Medical Research Berghofer (QIMR). Analyses in the AGDS dataset were conducted by

researchers at the QIMR. I led on coordinating and generating scripts to harmonise data

analysis with AGDS collaborators, including phenotypic definitions and genome-wide analysis

scripts. A summary of chapter-specific author contributions is shown below.

Abigail ter Kuile

9

Chapter 2

A.R.T.K, T.C.E and G.B. were responsible for the study conception and design. T.C.E, G.B,

G.K., M.R.D, M.H., N.R.W, S.E.M, E.M.B, N.G.M were responsible for the acquisition of the

data. A.R.T.K, C.H., A.J.P, H.L.D, J.M., Y.L., J.Z, A.P. were responsible for phenotypic data

preparation and cleaning. Preparation and quality control of genetic data was conducted by

A.R.T.K, S.H.L, J.R.I.C and B.A. B.M carried out data analysis in the AGDS QIMR dataset.

A.R.T.K conducted all other data analyses with support from G.M.V, S.H.L, H.L.D, J.M, M.S,

J.R.I.C and A.E.F. A.R.T.K was responsible for the drafting and revising of the manuscript,

under the close supervision of T.C.E. All co-authors reviewed the manuscript.

Chapter 3

A.R.T.K and G.B were responsible for the study conception and design, with input from all co-

authors. A.R.T.K, G.B, C.H, J.R.I.C, D.F.L, J.G and M.B.S were responsible for the acquisition

of the data. A.R.T.K conducted the analyses, with support from C.H. and C.R. A.R.T.K was

responsible for the drafting and revising of the manuscript, under the close supervision of G.B

and T.C.E. All co-authors contributed to the interpretation of the data, and reviewed and edited

the manuscript.

Chapter 4

A.R.T.K, A.J.P, C.R and T.C.E conceived and designed the study. T.C.E, G.B, G.K., M.R.D,

M.H., N.R.W, S.E.M, E.M.B, N.G.M were responsible for the acquisition of the data.

Phenotypic data in the GLAD+ dataset was processed and cleaned by A.J.P, A.R.T.K, C.H.,

H.L.D, J.M. Preparation and quality control of genetic data was conducted by A.R.T.K, S.H.L,

J.R.I.C and B.A. A.R.T.K led on genome-wide analyses, with support from J.M, M.S and

S.H.L. A.R.T.K generated polygenic scores in the GLAD+ dataset, with support from H.L.D

and Oliver Pain. A.J.P ran prediction modelling analysis, with support from A.R.T.K and Oliver

Pain. B.M and K.H carried out data analysis in the AGDS dataset. All authors contributed to

the interpretation of the results. A.J.P led on preparing an original draft of the manuscript

submitted in A.J.P’s PhD thesis, with input from A.R.T.K and supervisory input from T.C.E.

A.R.T.K was responsible for preparing and revising the current version of the manuscript

included in this thesis, with supervisory input from T.C.E and revised by G.B.

10

Publications related to this thesis

Publications resulting from the chapters in this thesis

Chapter 3 has been accepted for publication following peer-review in the journal Biological

Psychiatry Global Open Science.

ter Kuile, A. R., H�bel, C., Cheesman, R., Coleman, J. R. I., Peel, A. J., Levey, D. F., Stein,

M. B., Gelernter, J., Rayner, C., Eley, T. C., & Breen, G. (In press). Genetic

decomposition of the heritable component of reported childhood maltreatment.

Biological Psychiatry Global Open Science.

https://doi.org/10.1016/j.bpsgos.2023.03.003

Chapter 2 and 4 are currently in preparation for publication.

Chapter 2:

ter Kuile, A. R., Mitchell, B. L., Morneau-Vaillancourt, G., Lee, S. H., Davies, H. L., Mundy,

J., Peel, A. J., Skelton, M., H�bel, C., Davies, M. R., Coleman, J. R. I., F�rtjes, A. E.,

Ahmad, Z., Lin, Y., Adey, B. N., McGregor, T., Purves, K., Palmos, A., Zvrskovec, J.,

Hotopf, M., Kalsi, G., Jones, I. R., Smith, D. J., Veale, D., Walters, J. T. R., Armour,

C., Hirsch, C. R., McIntosh, A. M., Wray, N. R., Medland, S. E., Byrne, E. M., Martin,

N. G., Breen, G., & Eley, T. C. (In prep). Exploring the genome-wide genetic overlap

between anxiety and fear disorders.

Chapter 3:

Peel, A. J.*, ter Kuile, A. R.*, Rayner, C., Hopkins, K., Mitchell, B. L., Medland, S. E., Lee, S.

H., Adey, B. N., Armour, C., Buckman, J. E. J., Byrne, E. M., Coleman, J. R. I., Danese,

A., Davies, M. R., Davies, H. L., Hickie, I. B., Hirsch, C., Hotopf, M., H�bel, C., Jones,

I. R., Kalsi, G., Krebs, G., Martin, N. G., McIntosh, A. M., Mundy, J., Purves, K. L.,

Rogers, H., Skelton, M., Smith, D. J., Veale, D., Walters, J. T. R., Wray, N. R., Breen,

G., & Eley, T. C.(In prep). Genome-wide analysis and prediction of self-reported

outcomes following psychological treatment for depression and anxiety.

* Joint first authors

11

Additional publications

Co-authored publications in peer-reviewed journals arising from additional

work conducted during this PhD

Rayner, C., Coleman, J., Skelton, M., Armour, C., Bradley, J. R., Buckman, J. E. J., Davies,

M., Hirsch, C., Hotopf, M., H�bel, C., Jones, I. R., Kingston, N., Krebs, G., Lin, Y.,

Monssen, D., McIntosh, A., Mundy, J., Peel, A., Rimes, K., Rogers, H., Smith, D. J.,

ter Kuile, A. R, … Breen, G. & Eley, T.C., (2022). Patient characteristics associated

with retrospectively self-reported treatment outcomes following psychological therapy

for anxiety or depressive disorders - a cohort of GLAD Study participants. BMC

Psychiatry.

Young, K. S., Purves, K. L., H�bel, C., Davies, M. R., Thompson, K. N., Bristow, S., Krebs,

G., Danese, A., Hirsch, C., Parsons, C. E., Vassos, E., Adey, B. N., Bright, S.,

Hegemann, L., Lee, Y. T., Kalsi, G., Monssen, D., Mundy, J., Peel, A. J., Rayner, C.,

Rogers, H. C., ter Kuile, A. R., … , Eley, T. C. & Breen, G., (2022). Depression, anxiety

and PTSD symptoms before and during the COVID-19 pandemic in the UK.

Psychological Medicine.

Peel, A. J., Purves, K. L., Baldwin, J. R., Breen, G., Coleman, J. R. I., Pingault, J. B., Skelton,

M., ter Kuile, A. R., Danese, A. & Eley, T. C., (2022). Genetic and early environmental

predictors of adulthood self-reports of trauma. British Journal of Psychiatry.

Davies, M. R., Buckman, J. E. J., Adey, B. N., Armour, C., Bradley, J. R., Curzons, S. C. B.,

Davies, H. L., Davis, K. A. S., Goldsmith, K. A., Hirsch, C. R., Hotopf, M., H�bel, C.,

Jones, I. R., Kalsi, G., Krebs, G., Lin, Y., Marsh, I., McAtarsney-Kovacs, M., McIntosh,

A. M., Mundy, J., Monssen, D., Peel, A. J., Rogers, H. C., Skelton, M., Smith, D. J., ter

Kuile, A. R., … , Breen, G. & Eley, T. C., (2022). Comparison of symptom-based

versus self-reported diagnostic measures of anxiety and depression disorders in the

GLAD and COPING cohorts. Journal of Anxiety Disorders.

Peel, A. J., Armour, C., Buckman, J. E. J., Coleman, J. R. I., Curzons, S. C. B., Davies, M. R.,

H�bel, C., Jones, I., Kalsi, G., McAtarsney-Kovacs, M., McIntosh, A. M., Monssen, D.,

Mundy, J., Rayner, C., Rogers, H. C., Skelton, M., ter Kuile, A. R., Thompson, K. N.,

Breen, G., Danese, A., Eley, T. C., (2021). Comparison of depression and anxiety

symptom networks in reporters and non-reporters of lifetime trauma in two samples of

differing severity. Journal of Affective Disorders Reports.

Davies, H. L., H�bel, C., Herle, M., Kakar, S., Mundy, J., Peel, A. J., ter Kuile, A. R., … Breen,

G., (2022). Risk and protective factors for new-onset binge eating, low weight, and

self-harm symptoms in> 35,000 individuals in the UK during the COVID-19 pandemic.

The International Journal of Eating Disorders.

Davies, M. R., Glen, K., Mundy, J., ter Kuile, A. R., Adey, B. N., Armour, C., Assary, E.,

Coleman, J. R. I., Goldsmith, K. A., Hirsch, C. R., Hotopf, M., H�bel, C., Jones, I. R.,

Kalsi, G., Krebs, G., McIntosh, A. M., Morneau-Vaillancourt, G., Peel, A. J., Purves, K.

L., … Eley, T. C. (2023). Factors associated with anxiety disorder comorbidity. Journal

of Affective Disorders.

12

Chapter 1. General introduction

Anxiety and depression comorbidity is extensive and associated with

adverse outcomes

Common mental disorders account for a significant proportion of the global disease burden1.

They comprise multiple anxiety and depressive disorders and are a leading cause of disability,

with a global lifetime prevalence of approximately 16% and 11%, respectively2–4. Anxiety, fear

and sadness are typical emotions experienced in everyday life, but when experienced at

pathological levels can interfere with day-to-day functioning and become debilitating5. The

impairing features of anxiety and depression lead to a widespread disability affecting broad

aspects of quality of life, including problems maintaining relationships, poor educational

attainment, unemployment and financial difficulties5–8.

In addition to the detrimental impact on quality of life, there is a strong economic case for

improving the prevention and treatment of common mental disorders. In the UK, anxiety and

depression contribute 17.5% (�20.5 billion) and 22.5% (�26.6 billion), respectively, to a

minimum annual cost of �117.9 billion arising from mental health problems. This equates to

5% of the total gross domestic product. These costs include the effect of mental health

problems on disability, occupational productivity and health services. Moreover, current UK

mental health services are under extreme demand and struggle to meet the needs of all those

who seek a diagnosis or treatment9. When left untreated, anxiety disorders and depression

are often experienced for extensive periods, becoming either chronic or recurrent, contributing

to a large proportion of suicide1,10–12.

Anxiety disorders are characterised by excessive fear and anxiety, accompanied by impairing

features such as avoidance behaviours, physical anxiety reactions and panic attacks. In the

Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5), specific clusters

of symptoms differentiate one anxiety disorder from the other and include generalised anxiety

disorder (GAD), agoraphobia, social anxiety disorder, specific phobia and panic disorder13.

The breadth and focus of the threat that causes anxiety or fear differs across the disorders.

General anxiety is the anticipation and pathological preoccupation with a range of future

potential threats. Essential criteria for a GAD diagnosis include experiencing general anxiety

persistently for at least six months and difficulties controlling the worry. In contrast, fear is an

emotional and physiological reaction to immediate threats. The particular focus of perceived

threat distinguishes the phobias from each other and must consistently be feared or avoided

for a minimum period of six months for a diagnosis. The focus of fear is the broadest in

13

agoraphobia; the fear that escape might not be possible or help might not be available if

something were to go wrong in a range of situational exposures. In social anxiety disorder

(also known as social phobia), a narrower focus is placed on situations involving social

scrutiny. In specific phobias, the focus is limited to a certain situation or object that elicits fear.

A panic disorder diagnosis requires experiencing recurrent, unexpected panic attacks for at

least one month and consistent worry and avoidance of situations associated with having

panic attacks5.

Major depressive disorder (MDD) is the most common depressive disorder and has

considerable diagnostic heterogeneity. For a diagnosis of MDD, a low mood or the reduced

capacity to experience pleasure must be present and persist for at least two weeks, along with

the presence of other symptoms5. Clinical subtype specifiers can be symptom-based (with,

e.g. anxious-distress, mixed hypomanic, atypical, psychotic, melancholic features), temporal

or etiological-based (e.g. seasonal patterning, postpartum onset) and recurrence-based

(single episode or recurrent MDD)5,14.

Anxiety and depressive disorders are highly comorbid with one another, as well as other

psychiatric disorders and physical health problems. Estimates of comorbidity with another

mental health disorder are as high as 89% for anxiety disorders15 and approximately 75% for

major depressive disorder16,17. This includes comorbidity with a range of emotional and

behavioural psychopathologies, namely alcohol and substance use disorders, post-traumatic

stress disorder, obsessive-compulsive disorder, personality disorders and attention deficit

hyperactivity disorder5,15,18,19. However, the most common co-occurrence is among individual

anxiety disorders and depression. Approximately half of those who meet the diagnostic criteria

for one anxiety disorder have a history of another anxiety disorder10. Similarly, over half of

those with depression have at least one anxiety disorder and vice versa16,20,21.

In addition to anxiety and depression comorbidities being extensive, they are also associated

with various adverse clinical outcomes. A broad range of comorbid physical health problems

is common, including cardiovascular disease, diabetes, cancer, and irritable bowel

syndrome5,22–24. The presence of anxiety and depression is also linked with poor prognostic

outcomes in both chronic somatic diseases24 and, to some extent, treatment response in other

psychiatric disorders25. Compared to individuals with one psychiatric disorder, anxiety-

depression comorbidity is associated with increased symptom severity, impairment,

chronicity, the likelihood of suicide attempts, and is harder to treat20,21,26,27.

14

Research into understanding drivers of comorbidity whilst accounting for

clinical heterogeneity

Due to the magnitude and clinical severity of anxiety-depression comorbidity, examining

factors driving shared liability has been a focus in decades of research28. Comorbidity can be

modelled for clinical and research utility by examining covariance patterns within a structural

equation modelling (SEM) framework. Modelling patterns of comorbidity highlight the

distinctive and shared features between disorders and indicate overlapping or unique liability

factors29. Such research has implications for improving diagnostic systems and understanding

transdiagnostic or disorder-specific risk factors and treatment targets. Previous research

examining shared liability underpinning comorbidity patterns has challenged diagnostic

categories and continues to do so today28,30,31. These studies have aimed to address the

problem of pervasive comorbidity whilst taking into account clinical heterogeneity.

The extensiveness of anxiety-depression comorbidity is reflected in well-replicated disorder-

based and symptom-based factor analyses. These have revealed a broad shared liability to

both anxiety and depression along with other disorders characterised by emotional symptoms

(e.g. PTSD, OCD and eating disorders). Anxiety and depressive disorders are often

conceptualised as being core to this internalising disorder dimension, thought to be driven by

shared negative affect31. Disorders outside of this dimension are differentiated by other core

components of psychopathology. For example, disorders characterised primarily by

behavioural difficulties comprise an externalising disorder spectrum (e.g. substance use

disorders, borderline personality disorder). Disorders characterised by disturbances in thought

content with delusions, hallucinations or dissociative symptoms form the thought disorder

spectrum (i.e. schizophrenia spectrum disorders)32–34. The placement of some aspects of

disorders varies, for example, neurodevelopmental disorders (e.g. ASD, ADHD) and mania

(i.e. bipolar disorders)35,36. As discussed above, anxiety disorders and depression are also

comorbid with disorders in other spectra. This is captured by an overarching psychopathology

liability dimension, known as the ‘p-factor’, which accounts for shared features and high rates

of comorbidity observed across psychiatric disorders37.

Within these spectra lie subfactors that represent disorder-specific clinical heterogeneity.

These subfactors reflect differences in core features and comorbidity patterns among the

specific disorders. For example, GAD has higher rates of comorbidity with depression than

other anxiety disorders17. A well-replicated structure in the internalising spectrum is splitting

anxiety disorders and depression into two highly correlated subfactors: distress and fear.

Distress-based disorders are mainly characterised by broad, pervasive negative emotionality

15

(i.e. GAD is grouped with major depression). In contrast, fear-based disorders have more

context-specific features of distress, with fearful arousal and behavioural avoidance of a

narrower range of stimuli (i.e. agoraphobia, social anxiety disorder, specific phobia and panic

disorder)31,38. The splitting of internalising into distress and fear subfactors suggests an

overarching shared liability and some subfactor-specific risk factors that are not shared across

distress-fear dimensions.

The complex aetiology of common mental disorders poses a challenge to understanding what

drives comorbid psychopathologies and within-disorder heterogeneity. Both the risk and

treatment of anxiety and depression are associated with a broad range of genetic,

psychological and socio-environmental factors11,13. Such factors on their own are not enough

to cause the disorders or explain why some individuals respond better to certain treatments

than others. Furthermore, they do not act independently of one another, with a complex, multi-

layered interplay across different types of biological, psychological and social/environmental

factors. The interplay between these influences is dynamic, varying across the lifespan, from

person to person and across different psychiatric disorders39. Thus, resolving heterogeneity

and comorbidity problems in psychiatric research requires not only data on a broad range of

psychopathologies, but also a multi-level approach that integrates genetic, psychological, and

environmental factors.

In the next sections of this thesis introduction, progress in research of each facet of the bio-

psycho-social approach in the risk and treatment of anxiety and depression will be described.

By leveraging the complex interplay between genetic, environmental and psychological

influences, a focus will be placed on how genomic data can be used to understand each facet

and how this thesis aims to address key gaps in anxiety and depression genomics research,

depicted in Figure 1. A glossary of key statistical genetic terms and concepts relevant to this

thesis is provided in Boxes 1-4.

16

Figure 1: An overview of the three empirical chapters presented in this thesis and the

overlap among genomic, socio-environmental and psychological factors.

Through the application of genome-wide data, the overlap between genomic and

psychosocial factors can be harnessed to better understand risk factors and treatment

outcomes for common mental disorders.

Transdiagnostic and disorder-specific genetic influences

Twin study findings

Pivotal findings from family and twin studies formed the basis for understanding genetic

influences on anxiety and depression, which paved the way for future genomic studies.

Research dating to the early-mid 20th century showed that anxiety and depressive disorders

aggregate in families. A four to six-fold increased risk of anxiety disorders and a three-fold risk

of depression were observed in individuals with an affected first-degree relative compared to

unaffected controls40,41. Co-aggregation across individual anxiety disorders and depression

was also observed42.

Twin studies confirmed that the familial aggregation of anxiety and depressive disorders is

largely due to heritable factors and, to a lesser extent, environmental factors40–42. Definitions

Chapter 2: What

genomic factors

contribute to

subtype

heterogeneity?

Chapter 3: What

genomic factors

contribute to

environmental

vulnerability?

Chapter 4: What

genomic factors

contribute to

psychological

treatment

outcomes?

Common

mental

disorders

Biological

e.g. genotype

Social

e.g. adverse

environments

Psychological

e.g. poor

response to

psychological

treatment

17

and approaches to estimate heritability, including twin study estimates, are provided in Box 1.

Twin study heritability estimates of anxiety and depressive disorders are moderate, ranging

from 30-50%40–43. While substantial, these estimates are lower than rarer psychiatric disorders

such as schizophrenia and bipolar disorder (65-80%)43,44. Family and twin studies largely

support that familial risk and heritability estimates are similar across anxiety and MDD

subtypes; however, some find earlier-onset and recurrent MDD show greater familial risk and

heritability40,42,43,45. Twin studies have found that heritability estimates of anxiety and

depression measures are broadly similar across age groups43, although some report lower

heritability for childhood than adult depressive symptoms46. Stable genetic influences may

largely explain the moderate, long-term stability of anxiety and depressive symptoms across

the lifespan47,48. Furthermore, both stable and time-specific genetic factors are shared among

anxiety and depressive disorders49. These findings highlight how shared genetic influences

on anxiety and depression risk are important across the lifespan.

A major finding from twin studies is that high levels of shared genetic influences largely explain

the high comorbidity across anxiety and depressive disorders and related symptom

dimensions. Several twin studies found that GAD and MDD overlap almost entirely in genetic

variance50. A high genetic overlap across individual anxiety disorders is also consistently

greater than disorder-specific genetic influences42. Twin models of genetic covariation have

shown substantial shared genetic liability among anxiety and depression in childhood,

adolescence and adulthood50–53. A broad, internalising genetic factor is well-replicated across

twin datasets, supporting the overarching phenotypic structure of the internalising spectrum

discussed earlier51,53–55. The high genetic overlap between neuroticism, anxiety and

depression indicates genetic influences on shared negative affectivity drive internalising

psychopathology and anxiety-depression comorbidity56. Twin studies have further explored

the genetic relationship between anxiety and depression and disorders outside of the

internalising spectrum, showing pervasive genetic overlap with practically all forms of

psychopathology42,50. This overlap indicates cross-disorder genetics and the genetic factors

structure across disorders can be modelled as an overarching ‘p-factor’ liability that drives

comorbidity57,58. Studies also support partitioning psychiatric disorders into two distinct but

correlated lower-order internalising and externalising clusters59. As such, compared to their

overlap with other disorders, the genetic overlap is highest between anxiety and depression.

Notably, some multivariate twin modelling also supports the distress-fear subfactors structure

within the internalising dimension51,60,61. Multiple studies find that the extent of genetic overlap

between individual anxiety disorders and depression varies50. GAD and MDD show the

highest genetic overlap and are core to the internalising factor, indicating distress is a key

18

driver of general emotional problems59. Panic disorder and phobias have robust genetic

correlations with distress but also show some genetic distinction from GAD and MDD, forming

a genetic liability fear factor51,60,61. Together, twin study findings indicate that grouping

disorders is appropriate to identify higher-order genetic influences that broadly influence

internalising. However, genetic studies that group distress with fear disorders may also miss

important sources of heterogeneity and insight into mechanistic differences.

Box 1: Definitions and approaches to estimate trait heritability

Heritability: The proportion of total phenotypic variance observed within a population (VP)

explained by genetic factors, with estimates ranging from zero to one. Heritability estimates

of a trait can vary across time and populations due to differences in, for example,

environmental contexts, the type of phenotypic instrument and age at measurement62,63.

Broad-sense heritability: Defined as H2 = VG / VP with genetic values (VG) attributed to both

additive and non-additive genetic factors. Non-additive influences include interactions

between alleles located within the same genetic locus (dominance) or located on different

loci (epistasis)62–64.

Narrow-sense heritability: Defined as h2 = VA /VP with genetic values (VA) attributed only to

additive genetic factors; the sum of the average effect of multiple alleles contributing to a

trait62–64.

Twin study heritability: Twin studies are a commonly used family-based approach to

estimate trait heritability and environmental components. Phenotypic differences between

genetically identical monozygotic twins (MZ) and ~50% genetically similar dizygotic twins

(DZ) are compared. Traits that are more phenotypically similar in MZ than in DZ twins indicate

the role of genetic variation. Variance components modelled in twin designs include additive

genetics (A), non-additive genetics (D), the individual environment not shared among sibling

pairs (E), and the common environment shared among siblings growing up in the same

family, thus accounting for some within family resemblance (C). In the classical ACE twin

model, narrow-sense heritability is estimated based on the assumption that the non-additive

effects are zero. In the ADE model, the C component is replaced to estimate D, enabling

broad-sense heritability to be calculated from the combined variance of A and D64,65. Twin

studies have found stronger evidence for the role of additive genetic influences on complex

traits than non-additive43,66.

SNP-based heritability (h2

SNP): The proportion of phenotypic variance explained by a given

set of genome-wide common genetic variants67.

GREML h2

SNP: Genomic relatedness restricted maximum likelihood (GREML) is a commonly

used method to estimate narrow-sense h2

SNP using individual-level genomic data in unrelated

individuals68. GREML is implemented in the software GCTA69. Genome-wide common SNPs

are leveraged to construct a genomic relatedness matrix (GRM) containing the relatedness

coefficient between each pair of individuals across each SNP. The extent to which genomic

similarity can predict pair-wise phenotypic similarity for a given trait is used to deconstruct

trait variance into additive genetic variance and residual variance components (e.g. non-

additive genetics, environment, error)64,67.

19

Box 1: Definitions and approaches to estimate trait heritability

LDSC regression h2

SNP: Linkage disequilibrium score (LDSC) regression is a method to

estimate narrow-sense h2

SNP using GWAS summary statistics. LDSC regression assumes

under a polygenic model that common variants in higher LD regions are, on average, more

likely to be tagged by a causal variant in GWAS and therefore are more likely to be associated

with a given trait. An LD score quantifies the level of tagging of variants. A higher score

corresponds to a variant that tags more causal variants than those with a lower score and,

therefore, on average, has a higher association test statistic. LD scores are created using a

population reference panel representing the LD structure. For accurate estimates, the

population must be well-matched to the ancestry of the target GWAS population. An estimate

of h2

SNP is quantified as the regression coefficient from regressing the observed mean chi-

square association test statistic (Χ 2) against the LD score of each variant. The intercept can

be used to quantify bias from population stratification but can arise from an increase in sample

size and heritability of a trait. The attenuation ratio ([intercept -1]/ [Χ2 -1]) provides an estimate

of inflation not due to polygenicity, with a higher estimate indicating confounding67,70,71.

Liability scale h2

SNP: Observed scale h2

SNP estimates of categorical traits are converted to

the liability scale by accounting for the assumed population prevalence. The liability threshold

model presumes that genetic factors influence disorders under a normal distribution. An

individual meets case disorder status once reaching above a certain threshold of liability (on

the upper tail end of the normal distribution). The proportions of affected (cases) and

unaffected individuals (controls) in a GWAS sample often do not reflect the incidence of a

disorder observed in the population. The observed h2

SNP scale is often lower than the liability

h2

SNP scale, as information is lost by dichotomising the trait62,72.

From twin studies to genome-wide methods to understand biological aetiology

The heritable component and pervasive genetic overlap between anxiety and depressive

disorders showed the potential for employing genetic techniques to unravel biological

aetiology. Initial attempts to identify the genes associated with anxiety and depression

involved candidate gene and linkage analysis studies. As with all complex traits, these failed

to identify robust associations with anxiety and depression that survived replication42. As

knowledge of the human genome expanded, it was soon realised that psychiatric and other

complex traits are affected by many genetic variants, each of small effect size. In an attempt

to uncover this polygenicity, the era of genome-wide association studies (GWAS) began73.

A period of accelerated progress in genomics that led to the development of GWAS began

with the completion of the first human genome sequence, revealing millions of genetic

variants74,75. Box 2 provides a description of the various forms of human genetic variation.

Single nucleotide polymorphisms (SNP) are the most extensively analysed in GWAS, followed

by short insertion-deletion variants or ‘indels’. The majority of variants analysed in GWAS are

located within non-coding sequences of genes or in intergenic regions, which are potentially

involved in gene regulation76–78.

20

Box 2: Definitions of types and consequences of human genetic variation

Allele: Alternative versions of base pair changes of a variant79.

Polygenic: A phenotype influenced by many genetic variants79. Each genetic variant

contributes to phenotypic variation with a small effect size. The sum of each small effect,

along with environmental influences, contributes to the continuous distribution of complex

trait variation observed in a population80.

Polymorphism: Multiple definitions of genetic polymorphism exist in the literature81. Here,

polymorphism is defined as a population-specific term related to the occurrence of two or

more alleles at a position in a DNA sequence, with the rarer allele occurring in at least 1% of

a given population82–84.

Single nucleotide variant: Variation from a reference genome at a single base pair change.

The most abundant form of genetic variation in the human genome76,77.

Single nucleotide polymorphism (SNP): A single nucleotide variant with the minor allele

present in at least 1% of the population77,85.

Insertions and deletions (indels): A variant that corresponds to the insertion or deletion of

one or more base pairs not found in the reference genome. The second most abundant form

of genetic variation in the human genome76,77.

Copy number variation: An intermediate-large scale form of structural variation,

corresponding to additional copies (duplications) or deletions of segments of sequences >

1,000 base pairs85,86.

Tandem repeats: A segment of sequences with a number of base pair repeats. Variants

with repeat sequences of 2-6 bases are termed short tandem repeats, and those 7-10 bases

are termed variable number tandem repeats77.

Synonymous and non-synonymous variants: SNVs, and to a lesser extent SNPs, can

occur in protein-coding sequences within genes (exons), and cause a change in the amino

acid sequence, termed a non-synonymous variant. In contrast, variants within exons that do

not change the amino acid sequence are termed synonymous variants77.

Missense variant: A type of non-synonymous variant that alters one amino acid in the

protein, potentially influencing protein function77.

Nonsense variant: A type of non-synonymous variant that results in a premature STOP

codon and an incomplete and potentially non-functional protein product77.

Intergenic variant: A variant located between genes77.

Intronic variant: A variant that falls within an intron - the non-coding sequences of genes77.

Since the completion of the first human genome sequence, technological advancements have

provided cost-effective tools to rapidly conduct GWAS77. As the number of sequenced human

genomes increased, the depth and detail of publicly available reference genomes

improved87,88. Such advancements have led to an improved understanding of the structural

properties of the human genome and population genetic concepts that are leveraged for

genome-wide analyses (Box 3). The cost of generating genotyping data for GWAS through

microarrays has decreased substantially over the years. It is the most commonly used

approach to gather genotype data as it costs approximately ~95% less than next-generation

sequencing89. Reference genomes are then used for imputation to expand the number of

variants detected from microarrays, improving the overlap of variants across datasets using

21

different microarray technologies, thereby facilitating data harmonisation. The millions of

variants generated by these approaches are used in GWAS to conduct a hypothesis-free

study by testing across the entire genome if the frequency of alleles differs between cases

and controls (e.g. an allele is more common in individuals with anxiety disorder than in those

without), or along a trait dimension (e.g. an allele is associated with a higher GAD symptom

score)77. Variant associations may be causally or non-causally related to phenotype

expression, but are indirectly detected in GWAS due to being correlated with nearby causal

variants (in linkage disequilibrium [LD]). Haplotype blocks with regions of high LD are

leveraged to identify causal variants without the need for genotyping all causal SNPs90–92.

However, interpreting GWAS results is complicated by LD, giving rise to multiple associated

variants within a region, and subsequent analyses are required to determine causal

associations92–94.

Box 3: Structural properties of the human genome and related population genetic

concepts leveraged in genome-wide analyses

Reference panel: A database of densely genotyped haplotypes measured in a given set

of individuals from a specific population. Used to represent structural properties of the

genome in a given population, including allele frequencies and LD patterning. Examples of

commonly used reference panels include the Haplotype Reference Consortium, 1000

Genomes Project, and TOPMed95.

Allele frequencies: Allele frequency is the incidence of a certain allele in a population. In

a given population, minor allele frequency (MAF) is the frequency of the less common

allele, while major allele frequency is the frequency of the more common allele of a variant.

MAF is used to distinguish common from rare variants, with MAF > 1-5% considered

common79,90.

Linkage disequilibrium (LD): The non-random association between alleles located on

two distinct loci in a given population. Linkage equilibrium (LE) implies independence, in

which alleles at two loci are randomly associated by chance, with a theoretical

disequilibrium coefficient of 0. When the frequency of association between alleles at two

different loci deviates from expected when in LE, they are considered to be in LD. Levels

of LD are dependent on recombination events during meiosis that occur across

generations, natural selection, population bottlenecks, genetic drift, mutations, and

genomic inversion. LD can occur at a short-range, whereby LD is higher among physically

nearby genetic variants that are less likely to be separated by recombination events than

more distant variants. LD can occur at a long range, whereby the correlation between

genetic variants occurs at a larger distance, caused by various population processes91,96.

Haplotype block: Across the human genome are chromosomal regions with block-like

patterns of different LD levels. Low LD regions exist in recombination hot spots. In between

these hotspots are haplotype blocks, regions with high LD (from a few kb-100kb) and low

recombination, resulting in the inheritance of single-unit haplotype blocks across

generations. Haplotype blocks, therefore, reflect population history and geographical

subdivisions, with block sizes varying across ancestral populations91.

Genotype imputation: A statistical method that estimates genotypes of individuals at

unmeasured variants based on data from a more densely-genotyped population-specific

reference panel95,97.

22

Box 3: Structural properties of the human genome and related population genetic

concepts leveraged in genome-wide analyses

Genetic ancestry group: A label to denote groups of individuals who share more similar

genetic ancestors. Genetic ancestry is estimated in GWAS participants by comparing

genotypes with global reference panels. A continuum of genetic differences across

ancestry populations arises as those with more common ancestors have more similar

genomic structural patterns due to a number of influences. This includes ancestors mating

within nearby geographic locations and the sharing of more recombination events than

those of more distant ancestors. These differences confer variable frequencies of alleles

and patterning of linkage disequilibrium across ancestry populations. It is, therefore,

common practice to limit GWAS to more homogenous genetic ancestry groups. This

reduces false-positive associations arising from allele-frequency differences across

populations instead of reflecting phenotypic variation. It should be noted that ancestry

groups are an oversimplification of the complex dimension of human genetic variation and

demography95,98.

Population structure and stratification: Population structure is the genetic structure (i.e.

LD and allele frequencies) of an underlying population due to non-random mating and

restricted geographical movement. When a phenotype is correlated with a population

structure in a sample, this gives rise to population stratification that can confound genome-

wide analyses95.

Principal component analysis (PCA): A method used to estimate population structure

within a GWAS dataset. PCA reduces the dimensions of large-scale data (i.e. millions of

variants measured across thousands of individuals) into components that capture a large

proportion of variance, reflecting the underlying structure of a dataset. PCs are used to

estimate and assign participants to genetic ancestry groups by comparing the data with a

population reference panel of known ancestries. Additionally, including PCs as covariates

in genome-wide analyses is a commonly used approach to control for population

stratification95,99,100.

Genetic variants tested in GWAS are typically restricted to alleles more common in a given

population, with a minor allele frequency of at least 1-5%76,90. This approach has shown that

many common variants of small effect size spread across the genome contribute to the genetic

liability of complex traits and are, therefore, highly polygenic101. Consequently, large sample

sizes are required to detect robust associations of small effect sizes across the millions of

tested variants. To account for this multiple testing burden, a Bonferroni correction for one

million independent statistical tests (5 x 10-8) is applied to establish a genome-wide

significance threshold and reduce the chance of false positives. An approximately linear

relationship exists between the number of genome-wide significant associations and sample

size73,102.

Large, well-powered GWAS are pivotal to understanding the aetiology of complex traits.

Summary results can be made publicly available and used for various downstream analyses

(Figure 2), including elucidating the biological underpinnings of complex traits through

systems biology approaches. GWAS summary results can be integrated with various

functional genomic data types, including features from cells and tissues, such as gene

expression data. The small effect of common genetic variant associations can be aggregated

23

to the level of the gene and then combined into sets and translated to biological functions and

pathways enriched with a phenotype103.

Figure 2: An illustration of the various stages of genome-wide analyses.

Key population genetic concepts are applied to genome-wide association analyses (GWAS),

as described in Box 3. Summary results from GWAS can be used for several downstream

analyses. GWAS summary statistics are often made publicly available, enabling data on

various complex traits to be integrated into multiple-trait (e.g. multivariate or multivariable)

analyses. PC = principal component, LD = Linkage disequilibrium.

In addition to uncovering the genomic mechanisms associated with complex traits, GWAS

data can be used to estimate narrow-sense heritability captured by common genetic variants,

most often termed SNP-based heritability (see Box 1)67. Estimates of SNP-based heritability

can guide researchers on the sample size and power required to detect genome-wide

significant loci and the potential for future risk prediction based on GWAS data70. More

prevalent and highly polygenic psychiatric disorders with a lower SNP-based heritability, such

as anxiety and depression, require larger sample sizes than more heritable, rarer disorders,

such as schizophrenia, to achieve sufficient statistical power104. Estimates of SNP-based

heritability can be calculated directly from individual-level data used in GWAS, or through

summary-level GWAS results. As individual-level-based methods use full SNP data instead

of summary scores, they tend to yield higher estimates but are limited in computational speed,

and load with large sample sizes64,67. Using GWAS summary results is an alternative,

computationally efficient approach to estimating lower bounds of SNP-based heritability. This

has the additional benefit of not requiring access to individual-level genetic data for GWAS

datasets, often not possible in large-scale, collaborative meta-analyses. A widely used

summary-level method is linkage disequilibrium score (LDSC) regression (Box 1), which

Phenotyping

strategy

GWAS

SNP-based

heritability

genetic

correlations

polygenic

scoring

Multi-trait

analyses

Application of key concepts:

e.g. LD patterning, population structure

Population assignment

Genotype imputation

PC covariates

24

exploits the fact that some SNPs are more correlated with each other than others (i.e. are in

higher LD, as described in Box 3)67,70,71.

A genetic correlation between two traits can be estimated using extensions of the same

approaches used to estimate SNP-based heritability. To estimate genetic correlations across

a wide range of complex traits, the computationally light and flexible approach of using GWAS

summary statistics in LDSC regression is common. LDSC regression estimates global (Box

4) correlations by estimating genetic covariation across a pair of phenotypes (as captured by

common genetic variants) and scaling by genetic variances (i.e. from SNP-based

heritability)105,106. A significant global genetic correlation between two phenotypes arises when

the direction of the effect of common genetic variants is consistent across the genome105.

When genetic effects in both traits are, on average, in the same direction, this results in a

positive correlation of up to 1, or when in the opposite direction, a negative correlation of up

to -1107. Of note, genetic correlations cannot be used to determine causality but can be used

to generate novel hypotheses about the relationship between two traits to be then tested in

follow-up analyses.

Box 4: Definitions of pleiotropy, bivariate and multi-trait genetic analyses

Pleiotropy: When a genetic variant (or locus, gene) is associated with more than one

trait106.

Horizontal pleiotropy: When a genetic variant influences two traits, either directly or

indirectly, through an additional intermediate phenotype. Indicative of shared biological

processes between two traits (also termed biological pleiotropy)106.

Vertical pleiotropy: When a genetic variant influences one trait, which in turn is causally

associated with a second trait (i.e. one trait mediates the effect of the genetic variant on

another trait). Reverse causation of a causal cascade between two phenotypes is also

possible. The presence of vertical pleiotropy can be useful for intervention purposes106.

Environmentally mediated pleiotropy: A type of vertical pleiotropy when a genetic variant

influences a phenotype that shapes the environment, which in turn influences a second

phenotype108.

Spurious pleiotropy: When a genetic variant is falsely associated with two traits. This can

arise due to multiple different sources of biases. For example, design artefacts such as

phenotype misclassification or ascertainment bias. High LD can also lead to spurious

pleiotropy by a variant tagging multiple causal variants located in different genes with

different functions, which are not causally related to both phenotypes106,109. Variants in

regions of extreme LD (e.g. the major histocompatibility complex) are sometimes removed

for genetic correlation analyses110.

Global genetic correlations: A bivariate global genetic correlation between two traits is

the average effect of pleiotropy across the whole genome106.

Local genetic correlations: A bivariate local genetic correlation between two traits

analyses region-specific pleiotropy (e.g. at a given locus). Different regions may show

genetic correlations in opposite directions, contributing to a non-significant global genetic

correlation111.

25

Multi-trait analyses: Pervasive pleiotropy observed across multiple complex traits can be

leveraged through multi-trait analyses112–114. For example, multivariate genetic analyses (of

more than one outcome phenotype) can be used to extend bivariate genetic correlations

into multivariate Genomic Structural Equation Modelling115. Multivariable genetic analyses

(of one outcome phenotype and multiple predictors) can be used to model multiple

polygenic scores to predict a trait outcome jointly113,116.

A significant global genetic correlation can reflect multiple forms of underlying mechanisms

(Box 4). They may capture variants with a causal effect on multiple disorders, indicating

shared biological pathways70,106. Genetic correlations between psychiatric disorders and

health-related lifestyle factors may reflect causal relationships70. Genetic correlations

observed between personality and psychiatric disorders may be environmentally mediated108.

As such, measuring genetic correlations from GWAS across broad domains of phenotypes is

useful for understanding not only shared biological aetiology but also other sources of genetic

signals107.

To further strengthen hypotheses on shared and distinct mechanisms, a genetic overlap

across phenotypes can be harnessed for several downstream multi-trait analyses (Box 4).

Genetic correlations can be used to assess the genetic structure of traits related to psychiatric

disorders through multivariate modelling of genetic covariances. One example is the software

Genomic Structural Equation Modelling (Genomic SEM), which extends multivariate twin

modelling to genomic data using GWAS summary statistics115. This method constructs

estimates of SNP-based heritabilities and genetic correlations into matrices, often calculated

in LDSC regression. The genetic covariance matrix contains SNP-based heritabilities of each

trait and genetic correlations scaled relative to the heritabilities. A second sampling covariance

matrix contains the precision of these estimates (i.e. the sampling variance of each estimate

in the genetic covariance matrix), and the association between sampling errors to account for

GWAS sample overlap. A number of different structural equation models can then be fit to

these matrices. For example, this method can be used to evaluate hypotheses on why a group

of phenotypes are correlated or the extent to which other traits explain the genetic variance of

a trait in a model, or are unique to that phenotype115.

An alternative approach that leverages a genetic overlap between traits to build multi-trait

models is through polygenic scoring113. Polygenic scores are an aggregate of the number of

phenotype-associated alleles an individual carries. For example, for common mental

disorders, both individuals affected and unaffected by a disorder will carry a range of alleles

across the genome associated with an increased risk (coded as a 0, 1 or 2). To calculate a

polygenic score, alleles are summed and weighted by the effect size estimate of their

association with a given phenotype, derived from GWAS summary statistics in an independent

26

sample117. A polygenic score for one trait can be used to explain the phenotypic variance of

another trait. This is particularly useful for examining the genetic overlap between two traits

where one trait does not have a well-powered GWAS available for genetic correlation analysis.

Even for polygenic scores derived from well-powered GWAS, the amount of phenotypic

variance explained is relatively small73,102. For example, one of the most powerful psychiatric

polygenic scores is for schizophrenia, which explains 8.5% of the variance in schizophrenia

liability118. As such, polygenic scores are not yet sufficiently powered for prediction at the

individual level. Prediction at the group level can be improved by combining multiple well-

powered polygenic scores and non-genetic predictors into multivariable models. As GWAS

sample sizes grow, the hope is that the power of polygenic scores will improve to the extent

that multivariable predictive models can be implemented for clinical application119.

Progress in depression and anxiety disorder genomics

Since the start of the GWAS era approximately 16 years ago, significant progress has been

made in psychiatric genomics research120. Collaborative efforts and large-scale data collection

have enabled sample sizes to reach the level of power needed to start identifying the

abundance of common genetic variants associated with psychiatric disorders. This has been

facilitated by the Psychiatric Genomics Consortium (PGC), where large-scale meta-analyses

of multiple GWAS have transformed psychiatric genomics research. Large national cohorts

such as the UK Biobank and the Million’s Veteran Programme (MVP) have also been crucial

in these developments by contributing data to the largest GWAS of depression and anxiety to

date45,121,122.

Considerable progress has been made in depression genomics and led to a broad array of

findings on the neurobiology of depression. The largest MDD GWAS identified 178

independent loci in a sample of approximately 370,000 cases123. The translation of depression

GWAS results have highlighted the role of several brain regions, nervous system

development, synaptic processes, and immune-related function123–126. Such findings have

progressed our understanding of the aetiology of depression by corroborating previous

hypotheses and generating new ones.

Anxiety disorder genomics, and thus an understanding of genomic-associated systems

biology, is somewhat behind compared to depression. A lack of datasets with measures on

the multiple types of anxiety disorders has hindered progress. This may partly be due to the

complexity of phenotyping anxiety, a highly comorbid trait with somewhat vague boundaries

between ‘normal’ to pathological states127,128. The largest GWAS of anxiety identified five

genome-wide significant loci associated with GAD symptoms in 175,000 individuals129. Two

27

additional loci were associated with a case-control phenotype of anxiety and panic disorder in

34,000 cases129. A follow-up study applied a novel phenotyping approach to increase

statistical power by imputing GAD symptoms, identifying a further seven independent loci.

Integrating these GWAS results with functional genomic data found associations with the

dorsolateral prefrontal cortex and GABAergic neurons130. Although these studies represent

significant progress, larger datasets are still needed to further our understanding of the

genomic mechanisms underpinning anxiety disorders. Efforts are underway to publish the first

large-scale PGC anxiety disorder GWAS, representing a milestone in anxiety disorder

genomics127.

The SNP-based heritability estimates for anxiety and depression are broadly similar but are

also notably smaller than twin study estimates. Individual-level methods report ~26% SNP-

based heritability for any anxiety disorder131, and ~20% for MDD132,133. The ‘missing’ heritability

between SNP-based heritability and twin heritability estimates of complex traits are thought to

be explained by the role of genetic factors contributing to heritability not measured in GWAS

data, such as rare variants, non-additive genetics or gene-environment interactions73. As

noted earlier, summary-level SNP-based heritability estimates are even smaller than those

calculated from individual-level SNP data and twin data. The LDSC SNP-based heritability

estimate for MDD is 10.5%70,124 and 9% for broad depression125. Similarly, the largest GWAS

of anxiety reported an LDSC SNP-based heritability of 8.8%134, which is in line with other

anxiety disorder GWAS 135 and neuroticism GWAS136. However, meta-analysis SNP-based

heritability estimates of psychiatric disorders may also be biased downwards due to incorrect

conversions to the liability scale, not taking into account cohort-specific ascertainment (e.g.

an increase from 10.5% to 11.5% was observed for MDD)137. Furthermore, smaller, single-

study GWAS of anxiety disorders reported LDSC SNP-based heritability estimates as high as

28%138. This highlights how SNP-based heritability can vary across populations and

phenotypic measures73. Such heterogeneity across samples may, in part, limit the ability to

detect SNP-based heritability and contribute to missing heritability102.

As GWAS of anxiety and depression increase, a growing issue in the field is the balance

between increasing sample size and consistent and detailed phenotyping121,139,140. Many

large-scale genetic datasets use brief phenotypic measures, often limited to a few self-report

items to assess a broad psychiatric phenotype, as this approach is time and cost-effective.

Brief phenotyping has been key for making progress in loci discovery for both anxiety and

depression and will likely be vital for future risk prediction using GWAS data. The predictive

power of depression polygenic scores derived from GWAS is more dependent on increased

sample size than the detail of phenotyping. This indicates that sample size should be

28

prioritised over phenotyping depth for risk estimation141. However, when GWAS sample sizes

are equal, polygenic scores for more detailed measures have captured more specific genetic

influences on MDD141,142. Loci identified by GWAS of briefly phenotyped broad depression are

also less specific to MDD140,142. A tradeoff thus exists between maximising sample size and

the depth of phenotyping.

While research on the depth of GWAS phenotyping for anxiety disorders is lacking, lessons

learnt from phenotyping strategies in depression genomics may apply to anxiety genomics.

Larger samples ascertained from brief phenotyping may increase loci discovery, but some

associations may not be disorder-specific. An improvement in power may lead to disorder-

specific biology or non-specific treatment targets of broader phenotypes such as neuroticism.

Genetic risk prediction may improve, but identify a broader group of individuals than the

specific disorder under analysis140. Therefore, both brief and detailed phenotyping are of

research value. Genetic datasets with both brief and detailed phenotyping are needed to make

significant progress in anxiety disorder genomics and ultimately translate GWAS findings for

clinical utility139.

Leveraging genomic data to understand pervasive comorbidity and clinical

heterogeneity

As discussed earlier, with evidence from twin studies, the pervasive comorbidity across

anxiety disorders and depression is likely partly driven by genetic overlap. Findings from

GWAS data further support this notion. GWAS consistently report substantial polygenic

overlap between anxiety and depression and other psychopathologies123,124,131,134. The largest

GWASs of anxiety and MDD show a genetic correlation of 0.72123,134. Estimates from other

GWAS are similar, with 0.78 reported between a clinically ascertained MDD phenotype and

lifetime anxiety disorder and up to 0.88 with GAD symptoms124,131. Neuroticism also shows

high genetic correlations with anxiety and depression (0.69-0.73)123,125,131,134, providing further

support for genetic influences on shared negative affectivity driving comorbidity. Significant

global genetic correlations have also been reported between anxiety and depression and a

range of other psychiatric disorders, health, and behavioural traits123,125,131,134. This broad array

of significant genetic correlations calculated from GWAS data can be leveraged for various

analyses to better understand shared and distinct aetiology106.

Unlike twin studies, molecular genetic assessment of disorder subfactors, such as the

distress-fear subdomain, is still in its infancy. A major limiting factor is the availability of

detailed phenotyping to measure subtypes for GWAS sufficiently. Anxiety and depression

29

disorder subtype GWASs are needed to discern subtype-specific from transdiagnostic

common genetic influences. Failure to consider clinical heterogeneity may result in

overlooking specific biological pathways that can provide important insights into the genomic

influences on psychopathology. Furthermore, variable clinical heterogeneity across studies,

for example, inconsistent groupings of subtypes, may lead to inconsistent findings140.

More progress has been made in examining the genomic differences between the subtypes

of depression than for anxiety disorders. Subtyping of depression has found some evidence

that genetic heterogeneity reflects clinical heterogeneity143–145, although to a lesser extent in

some studies133,146,147. Differences in SNP-based heritability estimates have been reported.

More severe subtypes, including those treated with ECT, atypical, recurrent, and depression

with anxiety comorbidity, have shown higher heritability estimates than less clinically severe

forms of depression143,145. Genetic correlations across a range of subtypes from one study

suggested that 30-70% of genetic influences are shared143. Testing genetic differences

between subtypes of differing severity with a range of other complex traits has revealed

significant differences in genetic overlap with other psychiatric disorders, BMI, personality and

cognitive-related traits143,145,148. The more severe subtypes of depression are genetically less

similar to neuroticism145 and more similar to bipolar disorder and schizophrenia132. Such

analyses shed light on the genetic spectrum from manic to distress clusters (Figure 3),

showing both shared and unique genetic influences underlying depression and bipolar

disorder subtypes148. GWAS of depression subtypes have been largely underpowered for loci-

discovery and thus comparison149, but recent GWAS reveal evidence for some subtype-

specific loci143. Inconsistencies in subtyping have slowed progress in depression-subtype

genomics140.

30

Figure 3: An illustration of the genomic continuum of the mood disorder spectrum

from mania genomics to internalising genomics.

Adapted from Coleman et al.148.

Examining anxiety disorder subtypes at the genome-wide level has been especially limited in

psychiatric genomics. This is even though diagnostic subtypes of anxiety disorders are more

well-established than depression subtypes5. Anxiety genomics has lagged behind due to the

lack of genetic datasets with detailed phenotyping on anxiety. Datasets with detailed

measures on all five anxiety disorders are even more scarce. Thus, a more prominent focus

has been placed on ‘any’ anxiety disorder groupings to maximise power122,131,134,135. It remains

unclear what genomic influences are shared or specific to individual anxiety disorders without

assessing comorbid conditions.

GWAS of individual anxiety disorders has been largely underpowered for loci-discovery and

genetic correlation analyses to assess disorder-specific and transdiagnostic genetic liability.

Recent progress has been made on GWAS of current GAD symptoms130,134, but well-powered

GWAS of lifetime GAD has been limited150. GWAS of individual fear-based disorders is further

lacking151. One genome-wide significant locus has been reported in a GWAS of panic

disorder152, and one in a GWAS of agoraphobia symptoms153. Therefore, a comprehensive

assessment of the distress-fear model at the molecular genetic level is lacking. Preliminary

GWAS findings in the UK Biobank indicate some genetic specificity among the fear-based

disorders that is not shared with distress disorders150. Chapter 2 expands upon this study by

conducting GWAS meta-analyses of fear-based disorders and GAD with newly established

Bipolar

disorder

type I

Major

depressive

disorder

Bipolar

disorder

type II

Mania

Internalising

31

datasets that have detailed phenotyping on all five anxiety disorders. An improvement in

GWAS power enables a more comprehensive assessment of the common genetic variant

architecture of GAD and fear-based disorders and their genetic overlap with depression and

broad domains of other complex traits.

Social environmental influences on anxiety and depression

Research on the genetic basis of anxiety and depression is important for understanding

disorder aetiology, particularly for advancing our understanding of biology. However,

incorporating environmental information into genetic analyses is necessary to understand the

nature of genetic signals and the interplay between biological and social factors140. The

following sections of this chapter will first discuss how exposure to social environmental factors

is not independent of genetic influences. Psychological factors are discussed second, which

are the lens through which we experience environmental exposures and act as a bridge

between genetic and environmental influences on psychopathology. Figure 4 illustrates the

mechanisms involved in the interplay between genomics and different types of experiences in

the risk and treatment of common mental disorders.

Figure 4: An illustration of the potential mechanisms underpinning the course and

resolution of common mental disorders, which are influenced by the interplay

between genetic, socio-environmental and psychological factors.

Brown arrows represent genetic influences on biological pathways associated with the risk of

developing subtypes of common mental disorders, measured through genome-wide

association studies (GWAS). Experiences of negative and positive exposures and their

influence on the risk and treatment of psychopathology do not act independently of genetics.

Genomics can therefore be leveraged to understand these experiences. Blue arrows

represent the interplay between heritable characteristics associated with shaping adverse

environmental exposures and the risk of psychopathology, known as gene-environment

correlation. Green arrows represent the interaction between genetic liability to a disorder and

a psychological intervention (depicted here as a positive environmental exposure), leading to

variation in experiences and response, such as with psychological treatment.

32

Approximately half of the phenotypic variability of anxiety and depressive disorders is not

explained by genetics. Twin studies (see Box 1) show that the remaining variance is primarily

attributed to the individual environment (sometimes called “non-shared”). The common

environment explains some variance during childhood but declines with age and is no longer

present during adulthood47,154. Individual-specific environmental liability is partially shared

across anxiety disorders and depression, though some disorder-specific influences are also

observed51. As such, unique individual-specific environmental factors are thought to explain

why anxiety disorders and depression present as distinct conditions155. Anxiety and

depression are associated with both positive and negative social environmental contexts.

Positive environments can mitigate the effects of adverse environmental risk factors. For

example, social support is a particularly important transdiagnostic buffer for the effects of

severe adverse environments, notably traumatic experiences156.

Psychologically traumatic experiences, which can be defined as those perceived by the

individual as devastating and overwhelming, are major risk factors for internalising disorders

and psychopathology more broadly5. Traumas experienced during childhood are the most

robust transdiagnostic environmental risk factors31,157. Childhood trauma is reported to

increase psychopathology risk by approximately two-fold158,159. The mechanisms underlying

the causal effects of trauma on psychopathology are an ongoing area of research. A recent

meta-analysis found that after controlling for pre-existing risk factors such as environmental

adversities and genetics, an attenuated but significant causal effect of childhood trauma on

broad psychopathology remained160. Transdiagnostic theories on the long-term biological

effects of early-life trauma include altered brain structure and function, such as disruption to

the development of the hypothalamic-pituitary-adrenal axis. This may lead to elevated

glucocorticoid signalling in response to stress and abnormal brain development with long-

lasting effects on emotion and behaviour161,162. Difficulties with emotions include dysregulation

and increased emotional reactivity, which, together with elevated informational processing of

social cues as threatening, can influence maladaptive behavioural responses156. Given its

transdiagnostic effects, understanding the risk and consequences of trauma is an important

area of research. Furthermore, targeting trauma exposure in prevention and intervention

strategies may be more effective in reducing psychopathology risk than disorder-specific

environmental risk factors157.

The heritability of environmental measures

A complexity of disentangling the causal effects of environmental exposures on

psychopathology is that the environment does not operate independently of genetics. Twin

studies show that almost all measures of the environment are themselves genetically

33

influenced163. This includes the reporting of positive, negative, and traumatic social exposures,

which have a heritability ranging from 24-62%163–167. Significant SNP-based heritability

estimates have also been found for reported traumatic and stressful exposures (6-

30%)132,168,169. Twin studies find that life events dependent on one’s behaviour are more

heritable than independent events163,170,171, likely due to heritable behavioural characteristics

influencing the environments we are exposed to. These environments can be shaped

passively by our parents during childhood, or we actively select them in later life and by how

others engage with us. Correlations between genetic factors and environmental exposures

can arise through these mechanisms, termed gene-environment correlation172,173. As such,

the association between environmental measures and psychopathology may partly be

explained by shared genetic influences on both the likelihood of environmental exposure and

the development of psychopathology.

Identifying the genetically influenced traits that contribute to the heritability of environmental

measures would elucidate key traits involved in gene-environment correlations. Such findings

help narrow down characteristics to target in follow-up analyses assessing causal pathways

and ultimately inform intervention or prevention strategies. Multivariate twin modelling has

been used to deconstruct the heritability of an environmental measure by jointly modelling

genetically correlated personality and behavioural traits174,175. For example, one study found

that genetic propensity for non-cognitive traits is as important as genetic influences on

cognitive ability in explaining the heritability of educational attainment. This study highlighted

plausible alternative candidates for non-cognitive interventions to improve engagement with

the schooling environment174. As noted earlier, recent advancements in genomics research

enable the systematic assessment and multi-trait modelling of genetic correlations between a

broad range of traits, including environmental exposures and psychiatric and behavioural

traits115. For example, genomic multiple regression models in Genomic SEM have been used

to estimate genetic correlations between psychiatric disorders and smoking behaviours,

independent of other heritable behavioural traits176. This method could also be used to

deconstruct the SNP-based heritability of adverse environments and identify uniquely

genetically associated psychiatric and behavioural traits. Thus, such approaches provide the

opportunity to better understand key environmental risk factors for psychopathology at the

genomic level. Chapter 3 explores the genomic influences of multiple traits on vulnerability to

childhood trauma and aims to identify the traits that may be involved in gene-environment

correlation mechanisms through genomic multiple regression modelling.

34

Heritable psychological influences on anxiety and depression: from risk to

treatment

As outlined earlier, twin studies show that the individual environment explains more

phenotypic variation in anxiety and depression than the common environment. This is partly

due to the important role of individual differences in psychological influences that shape how

people experience their environment, which is unique to each person177,178. Genetics also

plays a role in how sensitive we are and the context in which we perceive exposures, which

in turn influences our response179. Genetics can therefore be leveraged to understand

characteristics associated with vulnerability to experiencing adverse exposures as traumatic

and a greater risk of subsequent psychopathology180.

Heritable influences on subjective self-reports of adverse experiences

Various genetically influenced psychological and cognitive factors impact sensitivity to

adverse environments. The stress-diathesis model describes sensitivity to negative (i.e.

stressful) environmental experiences and explains why not all those exposed to adverse

environments develop a common mental health problem. Genetic pathogenic effects depend

on environmental exposure and vice versa13,179,181,182. It is plausible that psychological factors

that influence sensitivity to adverse environments also influence reporting of events as

stressful or traumatic. As such, using self-reports to capture environmental exposures is often

conceptualised as a limitation to studies due to potential biases in reporting183. Meta-analyses

show that subjective and objective measures of adversity identify largely different groups of

individuals184. Thus, if research aims only to understand the influences on objective exposure,

then subjective reporting may capture influences on reporting biases as opposed to

mechanisms associated with exposure. Such research is important for prevention strategies.

However, understanding the genetic components contributing to subjective reporting is also

valuable. The subjective experience of trauma, as captured by retrospective self-reports, is

key for developing psychopathology, more so than trauma exposure. Some individuals

identified as exposed through objective measures, such as court documentation of childhood

maltreatment, do not subjectively report trauma in later life. These individuals are less likely

to develop post-traumatic psychopathology than those who retrospectively self-report

trauma184. This may reflect a group of individuals with low sensitivity to adverse environments,

which acts as a protective factor and resilience to adversity. Thus, understanding the

psychological and cognitive factors associated with retrospective and subjective reporting may

be informative for psychopathology intervention.

35

Psychological factors, including personality traits and cognitive biases, may influence how

individuals experience and engage with adverse environments and later risk of

psychopathology. Neuroticism is associated with cognitive biases towards negative stimuli,

elevated stress levels, and environmental sensitivity185,186. Twin studies suggest that genetics

primarily explain the correlation between neuroticism and adverse environmental sensitivity186,

highlighting the role of genetics in the subjective appraisal of adversity. Individuals with higher

levels of neuroticism tend to self-report more adversity in the absence of prospective records.

This is in contrast to agreeableness, which is associated with lower rates of retrospective

reports than indicated by the level of prospective accounts187. Higher neuroticism is also linked

with reporting traumatic and stressful events as more central to one's identity, a risk factor for

later psychopathology188,189. A potential mechanism explaining the link between neuroticism

and posttraumatic psychopathology is the increased emotional availability of trauma memories

through more frequent retrieval and rehearsal and, thus, maintenance of the memory189. A

genetic overlap has been observed between neuroticism, depressive symptoms, and

perceived stress190. Thus, genetic influences on neuroticism may influence the reporting of

events as traumatic166. Indeed, genetic correlations between GWASs of neuroticism and

retrospective self-reports of trauma have been observed132. Chapter 3 explores the extent to

which genomic influences on psychological traits such as neuroticism can explain the

heritability of retrospectively self-reporting adverse environments, aiming to identify key traits

involved in the experience of adverse environments.

Heritable influences on positive experiences and application to psychological

treatment

Genetic influences on psychological factors impact not only vulnerability to respond negatively

following trauma (as in the diathesis-stress model) but also sensitivity to positive

environments. The differential susceptibility model extends environmental sensitivity to

genetic influences on psychological factors affecting the experience and response to both

negative and positive environments. Under this framework, we can leverage genetic

information to understand who is more likely to benefit from support environments, providing

a more positive outlook on intervention research of psychopathology. Indeed, a core focus on

anxiety and depression genomics is understanding risk factors. However, genomics can also

be harnessed to improve treatment selection. For example, individuals with high

environmental sensitivity may translate to a better response to protective and supportive

exposures, such as psychological intervention179,191,192.

There is substantial individual variability observed in treatment response for anxiety and

depression. Given the overlap in risk factors, it is not surprising that treatments for anxiety and

36

depression are highly similar, with comparable treatment response rates across disorders31.

Medication and psychological therapy are both first-line treatments for anxiety and depression.

However, not all individuals respond to them equally, with approximately only 50% responding

to current treatments193,194. Predicting who will respond best to which treatment is poor and is

therefore often prescribed in a trial and error fashion13,194. Research understanding the factors

that influence treatment response is needed to aid in identifying suitable predictors to guide

treatment selection.

Genetic influences on antidepressant response show the potential for using genetic

information in guiding treatment. Psychiatric pharmacogenomics, which aims to harness

genetic variants associated with variability in response to medication to guide treatment

selection, has grown over recent years with some progress in antidepressant response

research. Although findings have been mixed, a meta-analysis of randomised controlled trials

of pharmacogenomic testing for antidepressant response showed a 71% increased likelihood

of depressive symptom remission in individuals receiving pharmaco-guided antidepressant

prescribing compared to standard practice prescribing195. Based on individual-level estimates,

the SNP-based heritability of antidepressant response is approximately 42%196 and 13% for

depressive symptom remission following antidepressant treatment197. These findings

represent considerable progress in the field of antidepressant response genetics. However,

antidepressants are not always beneficial to patients, and an integrated approach with

psychological therapy and medication outperforms pharmacotherapy alone. Thus, a combined

understanding of factors associated with response to both forms of treatment would be of even

greater clinical utility in guiding treatment selection.

Identifying predictors of improvement in anxiety and depressive symptoms following

psychological therapy is an ongoing area of research. Cognitive behavioural therapy (CBT) is

the gold standard psychological treatment for anxiety and depression198, including disorder-

specific and transdiagnostic approaches199, which are equally as effective200,201. Grouping

internalising disorders may prove useful in improving the prediction of psychological treatment

outcomes31. The ultimate goal is to incorporate a range of variables into predictive models to

aid in treatment decision-making202. Adding genetic factors to these models requires

advancements in examining the genetics of psychological treatment outcomes.

Considering there are genetic influences on a range of behavioural, psychological and

cognitive traits, which in turn influence response to positive environments163, there is likely a

genetic component to psychological treatment. No twin studies have found a significant

heritability of outcomes following psychological therapy, also known as ‘therapygenetics’.

However, evidence from examining related phenotypes suggests that a heritable component

37

exists. Twin studies report a significant heritability of fear extinction of 15-36%, a key

mechanism involved in exposure-based CBT for anxiety disorders203,204. Half of the genetic

influences on fear acquisition also overlap with fear extinction. This suggests that some

genetic influences on anxiety-related traits could be used to understand responses to

psychological therapy203. For example, polygenic scores from existing well-powered GWAS of

anxiety or depression may be useful in predicting psychological therapy outcomes alongside

other predictors. As such, incorporating pre-existing polygenic scores from well-powered

GWAS of related complex traits may be a promising avenue for therapygenetics research205.

Therapygenetics can be understood through the framework of gene-environment interaction,

in which psychological intervention is the environmental exposure and genetics influence

sensitivity to the exposure206. Preliminary findings using polygenic scores derived from a small

GWAS of environmental sensitivity in twins indicates that a higher genetic propensity for

environmental sensitivity is associated with a better response to more intensive CBT than less

intensive207. As was the case for many polygenic traits in candidate gene studies, early

attempts to identify specific genetic variants with robust gene-environment interaction effects

on psychological treatment outcomes were unsuccessful206,208. Response to psychological

therapy is unlikely to be influenced by a few genes, but rather highly polygenic, as is the case

for all heritable psychological traits.

A challenge for genome-wide therapygenetics is phenotyping at the scale required for GWAS.

GWAS of prognostic outcomes following CBT for anxiety and depression have been limited in

power due to the difficulty in ascertaining such phenotypes. The largest published GWAS

meta-analysis thus far was small, with 2,724 participants, and did not detect a significant SNP-

based heritability or genome-wide significant loci209. Alternative phenotyping strategies are

required to achieve sufficient power to detect the thousands of loci underpinning complex

traits. As with anxiety and depressive disorder genomics research, an improvement in power

has largely been achieved through brief phenotyping. The utility of brief phenotyping in GWAS

of psychological therapy outcomes has yet to be explored. Furthermore, subjective self-

reports of benefitting from psychological therapy may be useful in capturing genetic influences

that impact the perception of experiences, as with self-reports of environmental adversity.

Chapter 4 uses brief self-reports of psychological treatment outcomes to increase the sample

size for GWAS and prediction models that incorporate polygenic scores of related traits.

38

Summary and aims

The overarching aim of this thesis was to further the understanding of the risk and treatment

of common mental disorders by integrating bio-psycho-social components and leveraging

large genomic datasets (Figure 1). Chapters 2 and 3 focus on understanding genetic

influences on risk factors, while Chapter 4 explores genetic influences on treatment outcomes

(see Figure 4). Chapter 2 focuses more on the biological and genomic mechanisms

associated with clinical heterogeneity, whilst Chapters 3 and 4 harness genomic data to

understand psychological influences on environmental experiences. To improve statistical

power for genomic analyses of phenotypes that currently lack GWAS with sufficient sample

sizes, Chapters 2 and 4 leverage novel genotyped datasets with detailed measures of anxiety

and depression outcomes. All three empirical chapters harness pre-existing, well-powered

GWAS summary statistics across bio-psycho-social domains of complex traits related to

common mental disorders. This includes psychopathologies and physical health problems

commonly co-occurring with anxiety and depression, reported socio-environmental

experiences, and cognitive-related and psychological traits. Examining a broad range of

complex traits enables the assessment of shared and distinct genetic influences on common

mental disorders and the interplay with psychosocial influences at a large scale.

39

References

1. Organization, W. H. & Others. Depression and other common mental disorders:

global health estimates.

https://apps.who.int/iris/bitstream/handle/10665/254610/WHO-MSD-MER-2017.2-

eng.pdf (2017).

2. Kessler, R. C. et al. The global burden of mental disorders: an update from the

WHO World Mental Health (WMH) surveys. Epidemiol. Psichiatr. Soc. 18, 23–33

(2009).

3. Lim, G. Y. et al. Prevalence of Depression in the Community from 30 Countries

between 1994 and 2014. Sci. Rep. 8, 2861 (2018).

4. Somers, J. M., Goldner, E. M., Waraich, P. & Hsu, L. Prevalence and incidence

studies of anxiety disorders: a systematic review of the literature. Can. J. Psychiatry

51, 100–113 (2006).

5. American Psychiatric Association. Diagnostic and Statistical Manual of Mental

Disorders (DSM-5�). (American Psychiatric Pub, 2013).

6. Aquino, J. P. D., De Aquino, J. P., Londono, A. & Carvalho, A. F. An Update on the

Epidemiology of Major Depressive Disorder Across Cultures. Understanding

Depression 309–315 Preprint at https://doi.org/10.1007/978-981-10-6580-4_25

(2018).

7. Breslau, J., Lane, M., Sampson, N. & Kessler, R. C. Mental disorders and

subsequent educational attainment in a US national sample. J. Psychiatr. Res. 42,

708–716 (2008).

8. Fergusson, D. M. & Woodward, L. J. Mental health, educational, and social role

outcomes of adolescents with depression. Arch. Gen. Psychiatry 59, 225–231

(2002).

9. McDaid, D., Park, A. L., Davidson, G., John, A. & Knifton, L. The economic case for

investing in the prevention of mental health conditions in the UK. (2022).

10. Craske, M. G. & Stein, M. B. Anxiety. Focus 15, 199–210 (2017).

11. Malhi, G. S. & John Mann, J. Depression. The Lancet vol. 392 2299–2312 Preprint

at https://doi.org/10.1016/s0140-6736(18)31948-2 (2018).

12. Too, L. S. et al. The association between mental disorders and suicide: A systematic

review and meta-analysis of record linkage studies. J. Affect. Disord. 259, 302–313

(2019).

13. Craske, M. G. et al. Anxiety disorders. Nat Rev Dis Primers 3, 17024 (2017).

14. Harald, B. & Gordon, P. Meta-review of depressive subtyping models. J. Affect.

Disord. 139, 126–140 (2012).

15. Toft, T. et al. Mental disorders in primary care: prevalence and co-morbidity among

disorders. results from the functional illness in primary care (FIP) study. Psychol.

Med. 35, 1175–1184 (2005).

16. Kessler, R. C., Merikangas, K. R. & Wang, P. S. Prevalence, comorbidity, and

service utilization for mood disorders in the United States at the beginning of the

twenty-first century. Annu. Rev. Clin. Psychol. 3, 137–158 (2007).

17. Kessler, R. C. et al. Comorbidity of DSM–III–R Major Depressive Disorder in the

General Population: Results from the US National Comorbidity Survey. Br. J.

Psychiatry 168, 17–30 (1996).

40

18. Kessler, R. C. et al. The epidemiology of major depressive disorder: results from the

National Comorbidity Survey Replication (NCS-R). JAMA 289, 3095–3105 (2003).

19. Kessler, R. C. et al. Lifetime prevalence and age-of-onset distributions of DSM-IV

disorders in the National Comorbidity Survey Replication. Arch. Gen. Psychiatry 62,

593–602 (2005).

20. Hirschfeld, R. M. A. The Comorbidity of Major Depression and Anxiety Disorders:

Recognition and Management in Primary Care. Prim. Care Companion J. Clin.

Psychiatry 3, 244–254 (2001).

21. Lamers, F. et al. Comorbidity patterns of anxiety and depressive disorders in a large

cohort study: the Netherlands Study of Depression and Anxiety (NESDA). J. Clin.

Psychiatry 72, 341–348 (2011).

22. Steffen, A., N�bel, J., Jacobi, F., B�tzing, J. & Holstiege, J. Mental and somatic

comorbidity of depression: a comprehensive cross-sectional analysis of 202

diagnosis groups using German nationwide ambulatory claims data. BMC

Psychiatry 20, 142 (2020).

23. Penninx, B. W., Pine, D. S., Holmes, E. A. & Reif, A. Anxiety disorders. Lancet 397,

914–927 (2021).

24. Prince, M. et al. No health without mental health. Lancet 370, 859–877 (2007).

25. Kraus, C., Kadriu, B., Lanzenberger, R., Zarate, C. A., Jr & Kasper, S. Prognosis

and Improved Outcomes in Major Depression: A Review. Focus 18, 220–235

(2020).

26. Noyes, R., Jr. Comorbidity in generalized anxiety disorder. Psychiatr. Clin. North

Am. 24, 41–55 (2001).

27. Kalin, N. H. The Critical Relationship Between Anxiety and Depression. Am. J.

Psychiatry 177, 365–367 (2020).

28. Clark, L. A. & Watson, D. Distress and fear disorders: an alternative empirically

based taxonomy of the ‘mood’ and ‘anxiety’ disorders. Br. J. Psychiatry 189, 481–

483 (2006).

29. Eysenck, M. W. & Fajkowska, M. Anxiety and depression: toward overlapping and

distinctive features. Cogn. Emot. 32, 1391–1400 (2018).

30. Crocq, M.-A. The history of generalized anxiety disorder as a diagnostic category.

Dialogues Clin. Neurosci. 19, 107–116 (2017).

31. Watson, D. et al. Validity and utility of Hierarchical Taxonomy of Psychopathology

(HiTOP): III. Emotional dysfunction superspectrum. World Psychiatry 21, 26–54

(2022).

32. Conway, C. C., Forbes, M. K., South, S. C. & HiTOP Consortium. A Hierarchical

Taxonomy of Psychopathology (HiTOP) Primer for Mental Health Researchers. Clin.

Psychol. Sci. 10, 236–258 (2022).

33. Krueger, R. F. et al. Validity and utility of Hierarchical Taxonomy of Psychopathology

(HiTOP): II. Externalizing superspectrum. World Psychiatry vol. 20 171–193 Preprint

at https://doi.org/10.1002/wps.20844 (2021).

34. Kotov, R. et al. Validity and utility of Hierarchical Taxonomy of Psychopathology

(HiTOP): I. Psychosis superspectrum. World Psychiatry 19, 151–172 (2020).

35. Stanton, K. et al. An Initial Investigation of the Joint Classification of Hypomania-

and Neurodevelopmental Disorder-Relevant Dimensions Within the Hierarchical

Taxonomy of Psychopathology. Assessment 30, 414–432 (2023).

36. Stanton, K., DeLucia, E. A., Brown, M. F. D. & McDonnell, C. G. Advancing

understanding of the classification of broad autism phenotype and attention-

41

deficit/hyperactivity disorder symptom dimensions within the Hierarchical Taxonomy

of Psychopathology. Personal. Ment. Health 15, 113–123 (2021).

37. Caspi, A. et al. The p Factor: One General Psychopathology Factor in the Structure

of Psychiatric Disorders? Clin. Psychol. Sci. 2, 119–137 (2014).

38. Watson, D. Rethinking the mood and anxiety disorders: a quantitative hierarchical

model for DSM-V. J. Abnorm. Psychol. 114, 522–536 (2005).

39. Kendler, K. S. The structure of psychiatric science. Am. J. Psychiatry 171, 931–938

(2014).

40. Hettema, J. M., Neale, M. C. & Kendler, K. S. A review and meta-analysis of the

genetic epidemiology of anxiety disorders. Am. J. Psychiatry 158, 1568–1578

(2001).

41. Sullivan, P. F., Neale, M. C. & Kendler, K. S. Genetic epidemiology of major

depression: review and meta-analysis. Am. J. Psychiatry 157, 1552–1562 (2000).

42. Smoller, J. W. The Genetics of Stress-Related Disorders: PTSD, Depression, and

Anxiety Disorders. Neuropsychopharmacology 41, 297–319 (2016).

43. Polderman, T. J. C. et al. Meta-analysis of the heritability of human traits based on

fifty years of twin studies. Nat. Genet. 47, 702–709 (2015).

44. Craddock, N., O’Donovan, M. C. & Owen, M. J. The genetics of schizophrenia and

bipolar disorder: dissecting psychosis. J. Med. Genet. 42, 193–204 (2005).

45. Kendall, K. M. et al. The genetic basis of major depression. Psychol. Med. 51,

2217–2230 (2021).

46. Rice, F. Genetics of childhood and adolescent depression: insights into etiological

heterogeneity and challenges for future genomic research. Genome Med. 2, 68

(2010).

47. Nivard, M. G. et al. Stability in symptoms of anxiety and depression as a function of

genotype and environment: a longitudinal twin study from ages 3 to 63 years.

Psychol. Med. 45, 1039–1049 (2015).

48. Hannigan, L. J., Walaker, N., Waszczuk, M. A., McAdams, T. A. & Eley, T. C.

Aetiological influences on stability and change in emotional and behavioural

problems across development: a systematic review. Psychopathol Rev 4, 52–108

(2017).

49. Waszczuk, M. A., Zavos, H. M. S., Gregory, A. M. & Eley, T. C. The stability and

change of etiological influences on depression, anxiety symptoms and their co-

occurrence across adolescence and young adulthood. Psychol. Med. 46, 161–175

(2016).

50. Cerd�, M., Sagdeo, A., Johnson, J. & Galea, S. Genetic and environmental

influences on psychiatric comorbidity: a systematic review. J. Affect. Disord. 126,

14–38 (2010).

51. Waszczuk, M. A., Zavos, H. M. S., Gregory, A. M. & Eley, T. C. The phenotypic and

genetic structure of depression and anxiety disorder symptoms in childhood,

adolescence, and young adulthood. JAMA Psychiatry 71, 905–916 (2014).

52. Middeldorp, C. M., Cath, D. C., Van Dyck, R. & Boomsma, D. I. The co-morbidity of

anxiety and depression in the perspective of genetic epidemiology. A review of twin

and family studies. Psychol. Med. 35, 611–624 (2005).

53. Mosing, M. A. et al. Genetic and environmental influences on the co-morbidity

between depression, panic disorder, agoraphobia, and social phobia: a twin study.

Depress. Anxiety 26, 1004–1011 (2009).

54. Hettema, J. M. et al. Genetic and environmental risk structure of internalizing

42

psychopathology in youth. Depress. Anxiety 37, 540–548 (2020).

55. Kendler, K. S. et al. The structure of genetic and environmental risk factors for

syndromal and subsyndromal common DSM-IV axis I and all axis II disorders. Am.

J. Psychiatry 168, 29–39 (2011).

56. Hettema, J. M., Neale, M. C., Myers, J. M., Prescott, C. A. & Kendler, K. S. A

population-based twin study of the relationship between neuroticism and

internalizing disorders. Am. J. Psychiatry 163, 857–864 (2006).

57. Tackett, J. L. et al. Common genetic influences on negative emotionality and a

general psychopathology factor in childhood and adolescence. J. Abnorm. Psychol.

122, 1142–1153 (2013).

58. Pettersson, E., Larsson, H. & Lichtenstein, P. Common psychiatric disorders share

the same genetic origin: a multivariate sibling study of the Swedish population. Mol.

Psychiatry 21, 717–721 (2016).

59. Kendler, K. S. & Myers, J. The boundaries of the internalizing and externalizing

genetic spectra in men and women. Psychol. Med. 44, 647–655 (2014).

60. Hettema, J. M., Prescott, C. A., Myers, J. M., Neale, M. C. & Kendler, K. S. The

structure of genetic and environmental risk factors for anxiety disorders in men and

women. Arch. Gen. Psychiatry 62, 182–189 (2005).

61. Kendler, K. S., Prescott, C. A., Myers, J. & Neale, M. C. The structure of genetic and

environmental risk factors for common psychiatric and substance use disorders in

men and women. Arch. Gen. Psychiatry 60, 929–937 (2003).

62. Visscher, P. M., Hill, W. G. & Wray, N. R. Heritability in the genomics era--concepts

and misconceptions. Nat. Rev. Genet. 9, 255–266 (2008).

63. Wray, N. & Visscher, P. Estimating trait heritability. Nature education.

64. Barry, C.-J. S. et al. How to estimate heritability: a guide for genetic epidemiologists.

Int. J. Epidemiol. (2022) doi:10.1093/ije/dyac224.

65. Mayhew, A. J. & Meyre, D. Assessing the Heritability of Complex Traits in Humans:

Methodological Challenges and Opportunities. Curr. Genomics 18, 332–340 (2017).

66. Hill, W. G., Goddard, M. E. & Visscher, P. M. Data and theory point to mainly

additive genetic variance for complex traits. PLoS Genet. 4, e1000008 (2008).

67. Yang, J., Zeng, J., Goddard, M. E., Wray, N. R. & Visscher, P. M. Concepts,

estimation and interpretation of SNP-based heritability. Nat. Genet. 49, 1304–1310

(2017).

68. Yang, J. et al. Common SNPs explain a large proportion of the heritability for human

height. Nat. Genet. 42, 565–569 (2010).

69. Yang, J., Lee, S. H., Goddard, M. E. & Visscher, P. M. GCTA: a tool for genome-

wide complex trait analysis. Am. J. Hum. Genet. 88, 76–82 (2011).

70. Baselmans, B. M. L., Yengo, L., van Rheenen, W. & Wray, N. R. Risk in Relatives,

Heritability, SNP-Based Heritability, and Genetic Correlations in Psychiatric

Disorders: A Review. Biol. Psychiatry 89, 11–19 (2021).

71. Bulik-Sullivan, B. K. et al. LD Score regression distinguishes confounding from

polygenicity in genome-wide association studies. Nature Genetics vol. 47 291–295

Preprint at https://doi.org/10.1038/ng.3211 (2015).

72. Lee, S. H., Wray, N. R., Goddard, M. E. & Visscher, P. M. Estimating missing

heritability for disease from genome-wide association studies. Am. J. Hum. Genet.

88, 294–305 (2011).

73. Andreassen, O. A., Hindley, G. F. L., Frei, O. & Smeland, O. B. New insights from

the last decade of research in psychiatric genetics: discoveries, challenges and

43

clinical implications. World Psychiatry 22, 4–24 (2023).

74. Roses, A. D. The genome era begins. Nat. Genet. 33, 217–217 (2003).

75. Ikegawa, S. A short history of the genome-wide association study: where we were

and where we are going. Genomics Inform. 10, 220–225 (2012).

76. Hofker, M. H., Fu, J. & Wijmenga, C. The genome revolution and its role in

understanding complex diseases. Biochim. Biophys. Acta 1842, 1889–1895 (2014).

77. Michaelson, J. J. Genetic Approaches to Understanding Psychiatric Disease.

Neurotherapeutics 14, 564–581 (2017).

78. Edwards, S. L., Beesley, J., French, J. D. & Dunning, A. M. Beyond GWASs:

illuminating the dark road from association to function. Am. J. Hum. Genet. 93, 779–

797 (2013).

79. Knopik, V. S., Neiderhiser, J. M., DeFries, J. C. & Plomin, R. Behavioral Genetics.

(Macmillan Learning, 2016).

80. Visscher, P. M., Yengo, L., Cox, N. J. & Wray, N. R. Discovery and implications of

polygenicity of common diseases. Science 373, 1468–1473 (2021).

81. Karki, R., Pandya, D., Elston, R. C. & Ferlini, C. Defining ‘mutation’ and

‘polymorphism’ in the era of personal genomics. BMC Med. Genomics 8, 37 (2015).

82. Brookes, A. J. The essence of SNPs. Gene 234, 177–186 (1999).

83. Cavalli-Sforza, L. L. & Bodmer, W. F. The Genetics of Human Populations. (Courier

Corporation, 1999).

84. Arias, T. D., Jorge, L. F. & Barrantes, R. Uses and misuses of definitions of genetic

polymorphism. A perspective from population pharmacogenetics. Br. J. Clin.

Pharmacol. 31, 117–119 (1991).

85. Day, I. N. M. dbSNP in the detail and copy number complexities. Hum. Mutat. 31, 2–

4 (2010).

86. Eichler, E. E. Copy number variation and human disease. Nat Educ.

87. 1000 Genomes Project Consortium et al. An integrated map of genetic variation

from 1,092 human genomes. Nature 491, 56–65 (2012).

88. Consortium, †the International Hapmap & †The International HapMap Consortium.

The International HapMap Project. Nature vol. 426 789–796 Preprint at

https://doi.org/10.1038/nature02168 (2003).

89. Zhao, S., Jiang, L., Yu, H. & Guo, Y. GTQC: Automated Genotyping Array Quality

Control and Report. J Genomics 10, 39–44 (2022).

90. Visscher, P. M. et al. 10 Years of GWAS Discovery: Biology, Function, and

Translation. Am. J. Hum. Genet. 101, 5–22 (2017).

91. Slatkin, M. Linkage disequilibrium--understanding the evolutionary past and

mapping the medical future. Nat. Rev. Genet. 9, 477–485 (2008).

92. Schaid, D. J., Chen, W. & Larson, N. B. From genome-wide associations to

candidate causal variants by statistical fine-mapping. Nat. Rev. Genet. 19, 491–504

(2018).

93. Cano-Gamez, E. & Trynka, G. From GWAS to Function: Using Functional Genomics

to Identify the Mechanisms Underlying Complex Diseases. Frontiers in Genetics vol.

11 Preprint at https://doi.org/10.3389/fgene.2020.00424 (2020).

94. Watanabe, K., Taskesen, E., van Bochoven, A. & Posthuma, D. Functional mapping

and annotation of genetic associations with FUMA. Nat. Commun. 8, 1826 (2017).

95. Peterson, R. E. et al. Genome-wide Association Studies in Ancestrally Diverse

Populations: Opportunities, Methods, Pitfalls, and Recommendations. Cell 179,

589–603 (2019).

44

96. Reich, D. E. et al. Linkage disequilibrium in the human genome. Nature 411, 199–

204 (2001).

97. Uffelmann, E. et al. Genome-wide association studies. Nature Reviews Methods

Primers 1, 1–21 (2021).

98. National Academies of Sciences, Engineering & Medicine. Using Population

Descriptors in Genetics and Genomics Research: A New Framework for an Evolving

Field. (2023).

99. Price, A. L. et al. Principal components analysis corrects for stratification in genome-

wide association studies. Nat. Genet. 38, 904–909 (2006).

100. Novembre, J. & Stephens, M. Interpreting principal component analyses of spatial

population genetic variation. Nat. Genet. 40, 646–649 (2008).

101. Visscher, P. M., Goddard, M. E., Derks, E. M. & Wray, N. R. Evidence-based

psychiatric genetics, AKA the false dichotomy between common and rare variant

hypotheses. Mol. Psychiatry 17, 474–485 (2012).

102. Fabbri, C. Genetics in psychiatry: Methods, clinical applications and future

perspectives. PCN Rep. 1, (2022).

103. Uffelmann, E. & Posthuma, D. Emerging Methods and Resources for Biological

Interrogation of Neuropsychiatric Polygenic Signal. Biol. Psychiatry 89, 41–53

(2021).

104. Wray, N. R. et al. Genome-wide association study of major depressive disorder: new

results, meta-analysis, and lessons learned. Mol. Psychiatry 17, 36–48 (2012).

105. Bulik-Sullivan, B. et al. An atlas of genetic correlations across human diseases and

traits. Nat. Genet. 47, 1236–1241 (2015).

106. van Rheenen, W., Peyrot, W. J., Schork, A. J., Lee, S. H. & Wray, N. R. Genetic

correlations of polygenic disease traits: from theory to practice. Nat. Rev. Genet.

(2019) doi:10.1038/s41576-019-0137-z.

107. Abdellaoui, A. & Verweij, K. J. H. Dissecting polygenic signals from genome-wide

association studies on human behaviour. Nat Hum Behav 5, 686–694 (2021).

108. Avinun, R. The E Is in the G: Gene-Environment-Trait Correlations and Findings

From Genome-Wide Association Studies. Perspect. Psychol. Sci. 15, 81–89 (2020).

109. Solovieff, N., Cotsapas, C., Lee, P. H., Purcell, S. M. & Smoller, J. W. Pleiotropy in

complex traits: challenges and strategies. Nat. Rev. Genet. 14, 483–495 (2013).

110. Zheng, J. et al. LD Hub: a centralized database and web interface to perform LD

score regression that maximizes the potential of summary level GWAS data for SNP

heritability and genetic correlation analysis. Bioinformatics vol. 33 272–279 Preprint

at https://doi.org/10.1093/bioinformatics/btw613 (2017).

111. Ronald, A. & Pain, O. A Revolution Is Brewing in How We Understand the Shared

Genetic Causes of Psychiatric Disorders. The American journal of psychiatry vol.

179 791–793 (2022).

112. Visscher, P. M. & Yang, J. A plethora of pleiotropy across complex traits. Nature

genetics vol. 48 707–708 (2016).

113. Krapohl, E. et al. Multi-polygenic score approach to trait prediction. Mol. Psychiatry

23, 1368–1374 (2018).

114. Grotzinger, A. D. Shared genetic architecture across psychiatric disorders. Psychol.

Med. 51, 2210–2216 (2021).

115. Grotzinger, A. D. et al. Genomic structural equation modelling provides insights into

the multivariate genetic architecture of complex traits. Nat Hum Behav 3, 513–525

(2019).

45

116. Allegrini, A. G. et al. Multivariable G-E interplay in the prediction of educational

achievement. PLoS Genet. 16, e1009153 (2020).

117. Choi, S. W., Mak, T. S.-H. & O’Reilly, P. F. Tutorial: a guide to performing polygenic

risk score analyses. Nat. Protoc. 15, 2759–2772 (2020).

118. Trubetskoy, V. et al. Mapping genomic loci implicates genes and synaptic biology in

schizophrenia. Nature 604, 502–508 (2022).

119. Lewis, C. M. & Vassos, E. Polygenic risk scores: from research tools to clinical

instruments. Genome Med. 12, 44 (2020).

120. Psychiatric GWAS Consortium Coordinating Committee et al. Genomewide

association studies: history, rationale, and prospects for psychiatric disorders. Am.

J. Psychiatry 166, 540–556 (2009).

121. Sullivan, P. F. & Kendler, K. S. The state of the science in psychiatric genomics.

Psychol. Med. 51, 2145–2147 (2021).

122. Ask, H. et al. Genetic contributions to anxiety disorders: where we are and where we

are heading. Psychol. Med. 51, 2231–2246 (2021).

123. Levey, D. F. et al. Bi-ancestral depression GWAS in the Million Veteran Program

and meta-analysis in >1.2 million individuals highlight new therapeutic directions.

Nat. Neurosci. 24, 954–963 (2021).

124. Wray, N. R. et al. Genome-wide association analyses identify 44 risk variants and

refine the genetic architecture of major depression. Nat. Genet. 50, 668–681 (2018).

125. Howard, D. M. et al. Genome-wide meta-analysis of depression identifies 102

independent variants and highlights the importance of the prefrontal brain regions.

Nat. Neurosci. 22, 343–352 (2019).

126. Tubbs, J. D., Ding, J., Baum, L. & Sham, P. C. Immune dysregulation in depression:

Evidence from genome-wide association. Brain Behav Immun Health 7, 100108

(2020).

127. Smoller, J. W. Anxiety Genetics Goes Genomic. The American journal of psychiatry

vol. 177 190–194 (2020).

128. Smoller, J. W. Who’s afraid of anxiety genetics? Biological psychiatry vol. 69 506–

507 (2011).

129. Levey, D. F., Gelernter, J. & Polimanti, R. Reproducible genetic risk loci for anxiety:

results from∼ 200,000 participants in the Million Veteran Program. American Journal

of (2020).

130. Wendt, F. R. et al. Using phenotype risk scores to enhance gene discovery for

generalized anxiety disorder and posttraumatic stress disorder. Mol. Psychiatry

(2022) doi:10.1038/s41380-022-01469-y.

131. Purves, K. L. et al. A major role for common genetic variation in anxiety disorders.

Mol. Psychiatry 25, 3292–3303 (2020).

132. Coleman, J. R. I. et al. Genome-wide gene-environment analyses of major

depressive disorder and reported lifetime traumatic experiences in UK Biobank. Mol.

Psychiatry (2020) doi:10.1038/s41380-019-0546-6.

133. Hall, L. S. et al. Genome-wide meta-analyses of stratified depression in Generation

Scotland and UK Biobank. Transl. Psychiatry 8, 9 (2018).

134. Levey, D. F. et al. Reproducible Genetic Risk Loci for Anxiety: Results From

∼200,000 Participants in the Million Veteran Program. Am. J. Psychiatry 177, 223–

232 (2020).

135. Otowa, T. et al. Meta-analysis of genome-wide association studies of anxiety

disorders. Mol. Psychiatry 21, 1391–1399 (2016).

46

136. Nagel, M. et al. Meta-analysis of genome-wide association studies for neuroticism in

449,484 individuals identifies novel genetic loci and pathways. Nat. Genet. 50, 920–

927 (2018).

137. Grotzinger, A. D., Fuente, J. de la, Priv�, F., Nivard, M. G. & Tucker-Drob, E. M.

Pervasive Downward Bias in Estimates of Liability-Scale Heritability in Genome-

wide Association Study Meta-analysis: A Simple Solution. Biol. Psychiatry 93, 29–36

(2023).

138. Meier, S. M. et al. Genetic Variants Associated With Anxiety and Stress-Related

Disorders: A Genome-Wide Association Study and Mouse-Model Study. JAMA

Psychiatry (2019) doi:10.1001/jamapsychiatry.2019.1119.

139. Derks, E. M., Thorp, J. G. & Gerring, Z. F. Ten challenges for clinical translation in

psychiatric genetics. Nat. Genet. 54, 1457–1465 (2022).

140. Flint, J. The genetic basis of major depressive disorder. Mol. Psychiatry (2023)

doi:10.1038/s41380-023-01957-9.

141. Mitchell, B. L. et al. Polygenic Risk Scores Derived From Varying Definitions of

Depression and Risk of Depression. JAMA Psychiatry 78, 1152–1160 (2021).

142. Cai, N. et al. Minimal phenotyping yields genome-wide association signals of low

specificity for major depression. Nat. Genet. 52, 437–447 (2020).

143. Nguyen, T.-D. et al. Genetic heterogeneity and subtypes of major depression. Mol.

Psychiatry 27, 1667–1675 (2022).

144. Mitchell, B. L. et al. The Australian Genetics of Depression Study: New Risk Loci

and Dissecting Heterogeneity Between Subtypes. Biol. Psychiatry (2021)

doi:10.1016/j.biopsych.2021.10.021.

145. Clements, C. C. et al. Genome-wide association study of patients with a severe

major depressive episode treated with electroconvulsive therapy. Mol. Psychiatry

26, 2429–2439 (2021).

146. Jermy, B. S., Glanville, K. P., Coleman, J. R. I., Lewis, C. M. & Vassos, E. Exploring

the genetic heterogeneity in major depression across diagnostic criteria. Mol.

Psychiatry 26, 7337–7345 (2021).

147. Howard, D. M. et al. Genetic stratification of depression in UK Biobank. Transl.

Psychiatry 10, 163 (2020).

148. Coleman, J. R. I. et al. The Genetics of the Mood Disorder Spectrum: Genome-wide

Association Analyses of More Than 185,000 Cases and 439,000 Controls. Biol.

Psychiatry (2019) doi:10.1016/j.biopsych.2019.10.015.

149. Cai, N., Choi, K. W. & Fried, E. I. Reviewing the genetics of heterogeneity in

depression: operationalizations, manifestations and etiologies. Hum. Mol. Genet. 29,

R10–R18 (2020).

150. Morneau-Vaillancourt, G. et al. The genetic and environmental hierarchical structure

of anxiety and depression in the UK Biobank. Depress. Anxiety 37, 512–520 (2020).

151. Forstner, A. J. et al. Genome-wide association study of panic disorder reveals

genetic overlap with neuroticism and depression. Mol. Psychiatry 26, 4179–4190

(2019).

152. Erhardt, A. et al. TMEM132D, a new candidate for anxiety phenotypes: evidence

from human and mouse studies. Mol. Psychiatry 16, 647–663 (2011).

153. Deckert, J. et al. GLRB allelic variation associated with agoraphobic cognitions,

increased startle response and fear network activation: a potential neurogenetic

pathway to panic disorder. Mol. Psychiatry 22, 1431–1439 (2017).

154. Cheesman, R., Rayner, C. & Eley, T. The Genetic Basis of Child and Adolescent

47

Anxiety. Pediatric Anxiety Disorders 17–46 Preprint at https://doi.org/10.1016/b978-

0-12-813004-9.00003-7 (2019).

155. van Dongen, J., Slagboom, P. E., Draisma, H. H. M., Martin, N. G. & Boomsma, D. I.

The continuing value of twin studies in the omics era. Nat. Rev. Genet. 13, 640–653

(2012).

156. McLaughlin, K. A., Colich, N. L., Rodman, A. M. & Weissman, D. G. Mechanisms

linking childhood trauma exposure and psychopathology: a transdiagnostic model of

risk and resilience. BMC Med. 18, 96 (2020).

157. Waszczuk, M. A., Kotov, R., Ruggero, C., Gamez, W. & Watson, D. Hierarchical

structure of emotional disorders: From individual symptoms to the spectrum. J.

Abnorm. Psychol. 126, 613–634 (2017).

158. Lewis, S. J. et al. The epidemiology of trauma and post-traumatic stress disorder in

a representative cohort of young people in England and Wales. Lancet Psychiatry 6,

247–256 (2019).

159. McLaughlin, K. A. et al. Childhood adversities and first onset of psychiatric disorders

in a national sample of US adolescents. Arch. Gen. Psychiatry 69, 1151–1160

(2012).

160. Baldwin, J. R. et al. Childhood Maltreatment and Mental Health Problems: A

Systematic Review and Meta-Analysis of Quasi-Experimental Studies. Am. J.

Psychiatry appiajp20220174 (2023).

161. Koss, K. J. & Gunnar, M. R. Annual Research Review: Early adversity, the

hypothalamic–pituitary–adrenocortical axis, and child psychopathology. Journal of

Child Psychology and Psychiatry vol. 59 327–346 Preprint at

https://doi.org/10.1111/jcpp.12784 (2018).

162. Danese, A. & Baldwin, J. R. Hidden Wounds? Inflammatory Links Between

Childhood Trauma and Psychopathology. Annu. Rev. Psychol. 68, 517–544 (2017).

163. Kendler, K. S. & Baker, J. H. Genetic influences on measures of the environment: a

systematic review. Psychol. Med. 37, 615–626 (2007).

164. Wootton, R. E., Davis, O. S. P., Mottershaw, A. L., Wang, R. A. H. & Haworth, C. M.

A. Genetic and environmental correlations between subjective wellbeing and

experience of life events in adolescence. Eur. Child Adolesc. Psychiatry 26, 1119–

1127 (2017).

165. Peel, A. J. et al. A multivariate genetic analysis of anxiety sensitivity, environmental

sensitivity and reported life events in adolescents. J. Child Psychol. Psychiatry 64,

289–298 (2023).

166. Stein, M. B., Jang, K. L., Taylor, S., Vernon, P. A. & Livesley, W. J. Genetic and

environmental influences on trauma exposure and posttraumatic stress disorder

symptoms: a twin study. Am. J. Psychiatry 159, 1675–1681 (2002).

167. Pezzoli, P., Antfolk, J., Hatoum, A. S. & Santtila, P. Genetic Vulnerability to

Experiencing Child Maltreatment. Front. Genet. 10, 852 (2019).

168. Power, R. A. et al. Estimating the heritability of reporting stressful life events

captured by common genetic variants. Psychol. Med. 43, 1965–1971 (2013).

169. Dalvie, S. et al. Genomic influences on self-reported childhood maltreatment. Transl.

Psychiatry 10, 38 (2020).

170. Saudino, K. J., Pedersen, N. L., Lichtenstein, P., McClearn, G. E. & Plomin, R. Can

personality explain genetic influences on life events? J. Pers. Soc. Psychol. 72,

196–206 (1997).

171. Kendler, K. S., Karkowski, L. M. & Prescott, C. A. The assessment of dependence in

48

the study of stressful life events: validation using a twin design. Psychol. Med. 29,

1455–1460 (1999).

172. Knafo, A. & Jaffee, S. R. Gene–environment correlation in developmental

psychopathology. Dev. Psychopathol. 25, 1–6 (2013).

173. Jaffee, S. R. & Price, T. S. Gene–environment correlations: a review of the evidence

and implications for prevention of mental illness. Mol. Psychiatry 12, 432–442

(2007).

174. Krapohl, E. et al. The high heritability of educational achievement reflects many

genetically influenced traits, not just intelligence. Proc. Natl. Acad. Sci. U. S. A. 111,

15273–15278 (2014).

175. McAdams, T. A., Gregory, A. M. & Eley, T. C. Genes of experience: explaining the

heritability of putative environmental variables through their association with

behavioural and emotional traits. Behav. Genet. 43, 314–328 (2013).

176. Barkhuizen, W., Dudbridge, F. & Ronald, A. Genetic overlap and causal

associations between smoking behaviours and mental health. Sci. Rep. 11, 14871

(2021).

177. Werme, J., van der Sluis, S., Posthuma, D. & de Leeuw, C. A. Genome-wide gene-

environment interactions in neuroticism: an exploratory study across 25

environments. Transl. Psychiatry 11, 180 (2021).

178. Turkheimer, E. Three Laws of Behavior Genetics and What They Mean. Curr. Dir.

Psychol. Sci. 9, 160–164 (2000).

179. Assary, E., Vincent, J. P., Keers, R. & Pluess, M. Gene-environment interaction and

psychiatric disorders: Review and future directions. Semin. Cell Dev. Biol. 77, 133–

143 (2018).

180. Peel, A. J. et al. Genetic and early environmental predictors of adulthood self-

reports of trauma. Br. J. Psychiatry 1–8 (2022).

181. Kendler, K. S. et al. Stressful life events, genetic liability, and onset of an episode of

major depression in women. Am. J. Psychiatry 152, 833–842 (1995).

182. Dick, D. M. Gene-environment interaction in psychological traits and disorders.

Annu. Rev. Clin. Psychol. 7, 383–409 (2011).

183. Pingault, J.-B. et al. Research Review: How to interpret associations between

polygenic scores, environmental risks, and phenotypes. J. Child Psychol. Psychiatry

63, 1125–1139 (2022).

184. Danese, A. & Widom, C. S. Objective and subjective experiences of child

maltreatment and their relationships with psychopathology. Nat Hum Behav 4, 811–

818 (2020).

185. Uliaszek, A. A. et al. The role of neuroticism and extraversion in the stress-anxiety

and stress-depression relationships. Anxiety Stress Coping 23, 363–381 (2010).

186. Assary, E., Zavos, H. M. S., Krapohl, E., Keers, R. & Pluess, M. Genetic architecture

of Environmental Sensitivity reflects multiple heritable components: a twin study with

adolescents. Mol. Psychiatry 26, 4896–4904 (2021).

187. Reuben, A. et al. Lest we forget: comparing retrospective and prospective

assessments of adverse childhood experiences in the prediction of adult health. J.

Child Psychol. Psychiatry 57, 1103–1112 (2016).

188. Gehrt, T. B., Berntsen, D., Hoyle, R. H. & Rubin, D. C. Psychological and clinical

correlates of the Centrality of Event Scale: A systematic review. Clin. Psychol. Rev.

65, 57–80 (2018).

189. Ogle, C. M., Siegler, I. C., Beckham, J. C. & Rubin, D. C. Neuroticism Increases

49

PTSD Symptom Severity by Amplifying the Emotionality, Rehearsal, and Centrality

of Trauma Memories. J. Pers. 85, 702–715 (2017).

190. Rietschel, L. et al. Perceived stress has genetic influences distinct from neuroticism

and depression. Behav. Genet. 44, 639–645 (2014).

191. Belsky, J. & Pluess, M. Beyond diathesis stress: differential susceptibility to

environmental influences. Psychol. Bull. 135, 885–908 (2009).

192. Jolicoeur-Martineau, A. et al. Distinguishing differential susceptibility, diathesis-

stress, and vantage sensitivity: Beyond the single gene and environment model.

Dev. Psychopathol. 32, 73–83 (2020).

193. Cuijpers, P., Cristea, I. A., Karyotaki, E., Reijnders, M. & Huibers, M. J. H. How

effective are cognitive behavior therapies for major depression and anxiety

disorders? A meta-analytic update of the evidence. World Psychiatry vol. 15 245–

258 Preprint at https://doi.org/10.1002/wps.20346 (2016).

194. Rush, A. J. et al. Acute and longer-term outcomes in depressed outpatients

requiring one or several treatment steps: a STAR*D report. Am. J. Psychiatry 163,

1905–1917 (2006).

195. Bousman, C. A., Arandjelovic, K., Mancuso, S. G., Eyre, H. A. & Dunlop, B. W.

Pharmacogenetic tests and depressive symptom remission: a meta-analysis of

randomized controlled trials. Pharmacogenomics 20, 37–47 (2019).

196. Tansey, K. E. et al. Contribution of common genetic variants to antidepressant

response. Biol. Psychiatry 73, 679–682 (2013).

197. Pain, O. et al. Identifying the Common Genetic Basis of Antidepressant Response.

Biol Psychiatry Glob Open Sci 2, 115–126 (2022).

198. David, D., Cristea, I. & Hofmann, S. G. Why Cognitive Behavioral Therapy Is the

Current Gold Standard of Psychotherapy. Front. Psychiatry 9, 4 (2018).

199. Powers, M. B., de Kleine, R. A. & Smits, J. A. J. Core Mechanisms of Cognitive

Behavioral Therapy for Anxiety and Depression: A Review. Psychiatr. Clin. North

Am. 40, 611–623 (2017).

200. Newby, J. M., McKinnon, A., Kuyken, W., Gilbody, S. & Dalgleish, T. Systematic

review and meta-analysis of transdiagnostic psychological treatments for anxiety

and depressive disorders in adulthood. Clin. Psychol. Rev. 40, 91–110 (2015).

201. Garc�a-Escalera, J., Chorot, P., Valiente, R. M., Reales, J. M. & Sand�n, B. Efficacy

of transdiagnostic cognitive-behavioral therapy for anxiety and depression in adults,

children and adolescents: A meta-analysis. Revista de Psicopatolog�a y Psicolog�a

Cl�nica 21, 147–175 (2016).

202. Tanguay-Sela, M. et al. A systematic meta-review of patient-level predictors of

psychological therapy outcome in major depressive disorder. J. Affect. Disord. 317,

307–318 (2022).

203. Purves, K. L. et al. Evidence for distinct genetic and environmental influences on

fear acquisition and extinction. Psychol. Med. 1–9 (2021).

204. Hettema, J. M., Annas, P., Neale, M. C., Kendler, K. S. & Fredrikson, M. A twin

study of the genetics of fear conditioning. Arch. Gen. Psychiatry 60, 702–708

(2003).

205. Zanardi, R., Carminati, M., Attanasio, F., Fabbri, C. & Serretti, A. Genetics of

nonpharmacological treatments of depression. Psychiatr. Genet. 33, 1–7 (2023).

206. Lester, K. J. & Eley, T. C. Therapygenetics: Using genetic markers to predict

response to psychological treatment for mood and anxiety disorders. Biol. Mood

Anxiety Disord. 3, 4 (2013).

50

207. Keers, R. et al. A Genome-Wide Test of the Differential Susceptibility Hypothesis

Reveals a Genetic Predictor of Differential Response to Psychological Treatments

for Child Anxiety Disorders. Psychother. Psychosom. 85, 146–158 (2016).

208. Lester, K. J. et al. Non-replication of the association between 5HTTLPR and

response to psychological therapy for child anxiety disorders. Br. J. Psychiatry 208,

182–188 (2016).

209. Rayner, C. et al. A genome-wide association meta-analysis of prognostic outcomes

following cognitive behavioural therapy in individuals with anxiety and depressive

disorders. Transl. Psychiatry 9, 150 (2019).

51

Chapter 2. Exploring the genome-wide genetic

overlap between anxiety and fear disorders

This chapter is a manuscript that is in preparation for peer-review. Supplementary materials

for this chapter, as detailed in the text, are included in Appendix A.

Authors: Abigail R. ter Kuile1,2, Brittany L. Mitchell3, Genevi�ve Morneau-Vaillancourt1, Sang

Hyuck Lee1, Helena L. Davies1, Jessica Mundy1,2, Alicia J. Peel1, Megan Skelton1,2,

Christopher H�bel1,2,9, Molly R. Davies1,2, Jonathan R. I. Coleman1,2, Anna E. F�rtjes1, Zain

Ahmad1, Yuhao Lin1, Brett N. Adey1,2, Thomas McGregor1, Kirstin Purves1,2, Alish Palmos1,

Johan Zvrskovec1,2, Matthew Hotopf1,2,8, Gursharan Kalsi1,2, Ian R. Jones10, Daniel J. Smith11,

David Veale1,2,8, James T. R. Walters10, Ch�rie Armour5, Colette R. Hirsch1,2,8, Andrew M.

McIntosh11, Naomi R. Wray12,13, Sarah E. Medland3, Enda M. Byrne6, Nicholas G. Martin3,

Gerome Breen1,2, Thalia C. Eley1,2

1. Institute of Psychiatry, Psychology and Neuroscience, King's College London,

Denmark Hill, Camberwell, London, UK

2. National Institute for Health and Care Research (NIHR) Maudsley Biomedical

Research Centre, South London and Maudsley NHS Foundation Trust, London, UK

3. QIMR Berghofer Medical Research Institute, Brisbane, Australia

4. School of Biomedical Sciences, Faculty of Medicine, The University of Queensland,

Brisbane, Australia.

5. Research Centre for Stress, Trauma, and Related Conditions (STARC), School of

Psychology, Queen’s University Belfast (QUB), Belfast, Northern Ireland, UK

6. Child Health Research Centre, The University of Queensland, Brisbane, Australia

7. Brain and Mind Centre, The University of Sydney, Sydney, New South Wales,

Australia

8. South London and Maudsley NHS Foundation Trust, Bethlem Royal Hospital, Monks

Orchard Road, Beckenham, Kent, UK

9. National Centre for Register-based Research, Aarhus Business and Social Sciences,

Aarhus University, Aarhus, Denmark

10. MRC Centre for Neuropsychiatric Genetics and Genomics, National Centre for

Mental Health, Cardiff University, Cardiff,UK

11. Division of Psychiatry, Centre for Clinical Brain Sciences, University of Edinburgh,

Edinburgh, UK

12. Institute for Molecular Bioscience, The University of Queensland, Brisbane, Australia

13. Queensland Brain Institute, The University of Queensland, Brisbane, Australia

Abstract

Background: Twin studies have consistently shown a high genetic overlap amongst anxiety

disorders and depression that contributes to the internalising spectrum. Research has also

identified modest genetic specificity to fear-based anxiety disorders, not shared with general

anxiety and depression (often grouped as distress disorders). Due to the lack of datasets with

detailed phenotyping of all five anxiety disorders, genome-wide analysis of anxiety has

typically been limited to “any anxiety diagnosis”. Further genome-wide evidence is needed to

establish if fear-based disorders are genetically distinct from distress disorders.

Methods: We undertook a genome-wide association study (GWAS) meta-analysis of fear-

based disorders (panic, social anxiety disorder, specific phobia, and agoraphobia) and

generalised anxiety disorder (GAD). Cases and controls were defined using a combination of

brief single-item and detailed symptom-based diagnoses from three large datasets. We

explored two approaches to define controls. First, we screened for any anxiety disorder and

depression. Second, we screened specifically for fear or GAD. We estimated genetic

correlations between fear and GAD and compared their correlations with broad domains of

other complex traits.

Results: Our GWAS meta-analyses identified a total of three independent loci associated with

fear (up to 30,861 Ncases; 157,951 Ncontrols) and four with GAD (up to 54,928 Ncases; 117,320

Ncontrols). The genetic correlation between fear and GAD was not significantly different from

one, except for when excluding a depression-enriched dataset and screening controls

specifically (rg = 0.85; P = 4.57 � 10-3). Most complex traits did not have a significantly different

genetic correlation with fear versus GAD, including depression. The exceptions to this

included general cognitive ability, educational attainment, and coronary artery disease with

stronger negative genetic correlations with fear than GAD. Bipolar disorder type I, anorexia

nervosa, and neuroticism had stronger positive genetic correlations with GAD than fear.

Conclusions: Our findings partially support a distress-fear genetic distinction. However, we

found stronger evidence for an overarching genetic liability to internalising psychopathology

that drives comorbidity across anxiety disorders and depression.

53

Introduction

Anxiety disorders, including panic disorder, agoraphobia, specific phobia, social anxiety

disorder, and generalised anxiety disorder (GAD), are among the most common mental health

conditions, with an international lifetime prevalence of 16%1,2. The core symptoms of anxiety

disorders are persistent and excessive fear and worry, accompanied by impairing features

such as avoidance behaviours, physiological arousal, and panic attacks3. Individual disorder

diagnoses across the anxiety spectrum are based upon their distinct clinical features, although

high rates of comorbidity exist between them4,5. GAD is characterised by excessive and

chronic worry about a broad range of future potential threats. Panic disorder is an extreme

version of a flight or fight response with repeated experiences of a sudden unexpected feeling

of fear in the absence of a threat (i.e. panic attack). Recurrent, unexpected panic attacks are

necessary for a diagnosis of panic disorder. Panic attacks, however, are also transdiagnostic

and can occur in certain situations, as in phobic disorders. The phobias, including

agoraphobia, social anxiety disorder and specific phobia, are characterised by excessive fear

and avoidance of particular signals of perceived threat4,6. As they are all characterised by fear,

panic disorder and the phobias are often grouped together as ‘fear-based’ disorders.

Shared risk factors for anxiety disorders include being female, exposure to stressful

experiences, and a family history of anxiety and depression3. Familial aggregation of these

disorders reflects a partial genetic basis, with twin study estimates of heritability ranging from

20-60%3,7,8. Twin studies consistently show that most genetic influences on anxiety disorders

overlap, contributing to a broad anxiety-related genetic factor, which has considerable overlap

with genetic risk for depression9–12. However, evidence suggests this higher-order internalising

genetic liability can be differentiated into lower-order genetic components. Fear-based

disorders are reported to be partly genetically distinct from those characterised by distress

(GAD and depression), forming two distinct but correlated genetic liability fear-distress

subfactors9,13. Overall, twin studies show that anxiety disorders mostly share common genetic

factors but also have some genetic specificity.

Although twin studies have been crucial in revealing the overall genetic structure of the anxiety

disorder spectrum, they do not identify the specific genetic variants that contribute to the

heritability and high genetic overlap between the disorders. Genome-wide association studies

(GWAS) have begun to identify common genetic variants underpinning anxiety disorders.

Fewer GWAS have been performed in anxiety disorders than in other psychiatric disorders14.

However, recent developments have enabled several large-scale GWAS of broadly defined

anxiety and GAD symptoms15–18. Such studies have identified significant genetic correlations

with all other psychiatric disorders, including depression, schizophrenia, bipolar disorder, and

54

attention deficit hyperactivity disorder (ADHD)15–17. Consistent with evidence from twin

studies9, preliminary genome-wide findings indicate that a lifetime diagnosis of any fear-based

disorder (as a grouping) and depression are less genetically correlated with one another than

they each are with lifetime GAD19. However, the common genetic variants associated with

fear-based disorders remain unclear, which limits our understanding of their genetic overlap

at the molecular level.

Previous GWAS of fear-based disorders have been underpowered for loci discovery and to

assess genetic correlations with a broad range of traits19–21. Such analyses would provide

more molecular genetic evidence and fine-grained detail to the current hierarchical structures

of the psychiatric spectrum, which could have implications for improving diagnostic systems22.

A formal comparison of fear and GAD genetic correlations with other psychiatric, behavioural,

cognitive, or health traits could suggest transdiagnostic or subtype-specific mechanisms.

Further well-powered genome-wide evidence of the genetic distinction between distress and

fear is needed in the broader context of their relationships with other complex traits.

The way in which anxiety disorders are measured is an important consideration when

identifying genetic similarities and differences. Dealing with comorbidity is a particular

challenge in identifying anxiety-specific genetic influences, as anxiety disorders are among

the most highly comorbid disorders across psychopathology23. Screening controls for co-

occurring traits in GWAS could inflate genetic correlations24. A comparison of different control

screening approaches is needed. Furthermore, genetic influences captured by using brief

phenotyping measures may be less specific to the focal trait than those captured via detailed

phenotyping, reflecting more general psychopathology25,26. The previous GWAS reporting

distress-fear genetic correlations was limited to brief, single-item reports of receipt of a

diagnosis by a health professional for each fear disorder19. A recent study found that rates of

fear-based disorder diagnoses were lower when using brief self-report diagnoses than

detailed symptom-based measures, and observed the opposite pattern for GAD27. Since

identifying genetic variants requires large sample sizes, a trade-off exists between maximising

sample size and the level of detail in phenotyping. One way to achieve larger sample sizes

while retaining phenotypic specificity is by combining detailed symptom-based diagnoses with

brief single-item self-report diagnoses28.

Here, we aimed to assess the shared and non-shared genetic influences on fear-based

disorders and GAD at the level of common genetic variants. We increased sample size and

thus power for loci-discovery and genetic correlation analyses by combining detailed

symptom-based diagnoses with brief single-item diagnoses. We examined differences in

genetic correlations between fear-based disorders and GAD with a range of other complex

55

traits, and also assessed broad, and more specific approaches to screening controls. Our

study represents progress in the anxiety disorder genomics field and adds further detail on the

distress-fear distinction at the genome-wide level.

56

Methods

Phenotypes

Anxiety disorder phenotypes

Analyses were part of a pre-registered plan on the Open Science Framework (accessible at

https://tinyurl.com/OSFfearGAD). We conducted GWAS meta-analyses of two lifetime anxiety

disorder phenotypes: i) fear-based disorders (referred to as ‘fear’ hereafter; panic,

agoraphobia, specific phobia and social anxiety disorder; N cases = 30,861) and ii) GAD (N

cases = 54,928). We used the term “panic” instead of “panic disorder” as brief diagnostic

measures of self-reported panic attacks were used in addition to brief and detailed diagnostic

measures of panic disorder. This increased power and data harmonisation, as the UK Biobank

Mental Health Questionaire only had brief measures on panic attacks29 (Supplementary

Tables 1 & 2). Although panic attacks are not specific to panic disorder, the brief diagnostic

measure of panic attacks has shown reasonable agreement with detailed diagnostic measures

of panic disorder27. Five samples were included in the GWAS meta-analyses. The five

samples resulted in three case and control datasets, as detailed below (see Table 1 for a

summary of sample sizes). Individuals were screened for an anxiety disorder using a

combination of detailed symptom-based diagnoses and brief, self-report diagnoses. Fear

cases were defined as participants who endorsed at least one fear disorder.

Detailed diagnostic measures

Detailed symptom-based measures of anxiety disorders were derived using previously

described algorithms27. Cases for each phenotype were defined as those who met DSM-5

criteria for a lifetime symptom-based disorder diagnosis. Each anxiety disorder was assessed

using an online, self-report questionnaire version of the Composite International Diagnostic

Interview short-form (CIDI-SF)30,31. The availability of questionnaires in each study is

summarised in Supplementary Table 1.

Brief diagnostic measures

Individuals meeting case status on brief measures endorsed a single-item self-report question

for the corresponding disorder, e.g. “Have you ever been diagnosed with one or more of the

following mental health problems by a professional, even if you do not have it currently?”. The

exact phrasing of the question and responses in each study, as well as mapping to disorders,

are provided in Supplementary Table 2.

57

Control groups

We explored two different ways of defining control groups. First, controls were screened for

any anxiety disorder or depression (fearanx-dep and GADanx-dep). Due to the high genetic

correlation between these disorders19, this was the most powerful approach for loci discovery.

Detailed and brief measures were used to screen controls for any anxiety disorder

(Supplementary Tables 1 & 2). Controls were screened for depression using brief, self-report

diagnoses in all datasets. Second, to better distinguish genetic differences between GAD and

fear, we defined specifically screened controls whereby participants were only screened for

the specific disorder being analysed (fearspecific and GADspecific). This is because both fear and

GAD have high genetic overlap with one another and with depression, thus screening controls

for any anxiety disorder and depression could further increase the genetic correlation between

them. Fearspecific GWAS controls were only screened for any fear disorder, and GADspecific

GWAS controls were only screened for GAD where possible. Screening controls specifically

for only GAD or only fear-based disorders was not possible in the QIMR dataset as we were

limited to using a broad, single-item self-report diagnosis of “anxiety”. Controls were not

screened for the presence of other psychiatric disorders as this can bias genetic correlation

estimates24.

Samples

The GLAD Study and COPING Study (GLAD+ dataset)

Data from two studies within the National Institute for Health and Care Research (NIHR)

BioResource were included. Cases were primarily ascertained from the Genetic Links to

Anxiety and Depression (GLAD) study32. The GLAD study is an online research platform

launched in September 2018 to recruit participants with lifetime anxiety and/or depressive

disorder. Recruitment is ongoing and is open to individuals residing in the UK and aged at

least 16 years. Initial recruitment involved a widespread media campaign, followed by ongoing

social media advertising and participating NHS trusts and GP practices. Controls for analyses

with GLAD cases were ascertained from a longitudinal survey run by the GLAD team,

collecting data within GLAD and other NIHR BioResource cohorts called the COVID-19

Psychiatric and Neurological Genetics (COPING) study33. The phenotyping in COPING for

anxiety disorders was identical to GLAD. A proportion of COPING study participants met the

criteria for an anxiety disorder and were defined as cases (Table 1). Cases and controls in the

GLAD and COPING studies were defined using both brief and detailed measures of all five

anxiety disorders. We refer here to the GLAD and COPING combined study samples as the

GLAD+ dataset.

58

Queensland Institute of Medical Research (QIMR) Berghofer dataset

A second unrelated set of case-control analyses included data from two Australian QIMR

Berghofer studies. Cases were obtained from the Australian Genetics of Depression Study

(AGDS), launched in September 2016, to collect genetic and phenotypic data on individuals

with a lifetime experience of depression31. Recruitment is ongoing and has been conducted

via pharmaceutical prescription history provided by the Australian government and through

media campaigns. Participation is open to individuals aged at least 18 years across Australia.

AGDS cases were defined using detailed and brief self-report measures of all five anxiety

disorders. Controls for analyses with AGDS cases were obtained from the QIMR Berghofer

QSkin Sun & Health Study34, defined using a brief self-report measure of any anxiety disorder.

The UK Biobank dataset

The UK Biobank is a population-based dataset of over 500,000 participants aged between 40

and 69 recruited from 2006 to 201035. In 2017, a subset of participants (N = 157,366)

completed the online follow-up Mental Health Questionnaire (MHQ) assessing mental health

and well-being through self-report measures29. Participants were defined as cases or controls

using brief measures of fear disorders and brief and detailed measures of GAD.

59

Table 1. Sample sizes in each study, GWAS dataset, and meta-analysis of fear-based disorders and GAD.

Control criteria

N dataset total


(Ncases + Ncontrols)

Anxiety disorder

GWAS

Datasets

Study

Sample

Ncases

Anx-dep controls

Specific controls

Anx-

dep

Specific

Ncontrols

Neffective

Ncontrols

Neffective

Fear

GLAD+

GLAD

12,243

-

13,300

-

16,961

17,848

19,626

COPING

1,186

4,419

6197

UKB

10,789

83,236

38,204

137,209

40,010

94,025

147,998

QIMR

AGDS

6,643

-

17,457

-

18,241

19,365

21,188

Qskin

-

12,722

14,545

Total meta-analysis N

(GLAD+, UKB & QIMR)

30,861

100,377

68,960

157,951

75,212

131,238

188,812

Total meta-analysis N

(GLAD+ & UKB)

24,218

87,655

51,504

143,406

56,971

111,873

167,624

GAD

GLAD+

GLAD

14,339

-

13,857

-

19,606

20,452

23,093

COPING

1,694

4,419

7060

UKB

26,067

83,236

79,402

95,715

81,950

109,303

121,782

QIMR

AGDS

12,828

-

25,550

-

27,265

25,550

27,373

Qskin

-

12,722

14,545

Total meta-analysis N

GLAD+, UKB & QIMR)

54,928

100,377

118,808

117,320

128,822

155,305

172,248

Total meta-analysis N

(GLAD+ & UKB)

42,100

87,655

93,259

102,775

101,556

129,755

144,875

Numbers shown in bold are total sample size used in GWAS meta-analyses. Effective sample size was calculated as: Neffective = 4/[(1/ncase) +

(1/ncontrol)], which converts the power of a study sample size to one with a balanced case-control ratio (i.e. equivalent to a study with a sample

prevalence of 50%, with Ncases = Neffective and Ncontrols = Neffective).

60

Genome-wide association analyses

Genotyping and quality control was performed separately in each study (Supplementary

Materials). Common genetic variants in the GLAD+ and QIMR datasets were imputed to the

Top Med imputation panel (Version R2 on GRC38) and the UK Biobank to the Haplotype

Reference Consortium and the UK10K Consortium reference panels. GWAS analyses were

restricted to common genetic variants (MAF > 1%) and high confidence-imputed variants

(INFO score > 0.3 for TOPMed imputed variants; > 0.4 for UK Biobank).

We ran four anxiety disorder GWASs; fearspecific, GADspecific, fearanx-dep and GADanx-dep. A

separate GWAS was conducted for each of the three datasets (GLAD+, UK Biobank, QIMR).

Sample sizes for each GWAS are shown in Table 1. The software REGENIE v3.1.3 was used

for analyses in the GLAD+ and the UK Biobank datasets, whilst SAIGE v0.44 was used in the

QIMR dataset36,37. The first ten principal components and genotyping batch were included as

covariates. Additionally, the assessment centre was included as a covariate for GWAS in the

UK Biobank. GWAS were limited to participants of European-associated genetic ancestry

clusters, defined using principal component analysis (PCA).

Meta-analyses and annotation

We used the software METAL to perform inverse-variance weighted meta-analyses of each

anxiety disorder phenotype38. In our primary analyses, we meta-analysed GWAS results from

the GLAD+, QIMR Berghofer, and UK Biobank datasets (termed full meta-analysis

throughout). As all cases from the AGDS also had comorbid depression, we conducted a

second set of meta-analyses excluding the QIMR datasets (referred to as GLAD+ and UKB

meta-analysis). A depression-enriched dataset might increase depression/distress genetics in

anxiety disorder GWAS, elevating the genetic correlation between fear, GAD, and depression.

Meta-analysis sample sizes are shown in Table 1. We restricted common genetic variants to

those overlapping across datasets in each meta-analysis. Associated variants were mapped

and annotated using FUMA v1.4.0 (with default parameters applied) and the UKB release2b

10K European reference panel39. MAGMA v1.08 was used with default parameters to identify

gene-level and biological pathway associations. Gene-level association analysis aggregates

the combined effect of associated common genetic variants at the level of the gene. Results

from gene-level analyses were then applied to gene-set analysis to conduct biological pathway

analyses40.

61

SNP-based heritability

We estimated the heritability captured by common genetic variants (h2

SNP) of fear and GAD.

Population prevalences for liability scale estimates were based on the assumption of accurate

sampling from the COPING study, in which both detailed and brief measures were available

to define cases and controls for all five anxiety disorders (Supplementary Table 3). GWAS

meta-analysis summary results were used in LDSC regression to estimate h2

SNP

41.

Genetic correlations

Using LDSC regression42, we calculated genetic correlations (rg) between i) fear and GAD and

ii) fear and GAD with 345 external complex traits respectively. Traits were considered

sufficiently powered for genetic correlation analysis if they had a GWAS mean Χ2 > 1.02 and

a heritability Z score > 4, calculated in LDSC regression. We tested if the genetic correlation

between fear and GAD was significantly different from 0 (using default parameters in bivariate

LDSC regression) and significantly different from 1 (calculated in R using the chi-squared

distribution function and [(|rg|−1)/se]2 ). To correct for multiple testing, we applied a Bonferroni

corrected P-value threshold (𝛼[0.05] / the number of fear-GAD rg tested [4] = P  ≤ 0.0125).

Fear and GAD GWAS with a genetic correlationsignificantly different from 1 were then tested

for genetic correlations with external traits. We selected external traits in a hypothesis-free

manner and applied a Bonferroni-corrected significance threshold (0.05 / number of external

traits tested [345] = P ≤ 1.45 � 10-4). We tested for differences between fear versus GAD in

terms of their respective genetic correlations with external traits using the block-jackknife

method implemented in LDSC regression42,43. Fear and GAD were considered to have a

significantly different genetic correlation with an external trait if the P-value derived from the

block-jackknife Z statistic result exceeded the Bonferroni-corrected threshold

(P  ≤ 1.45 � 10-4).

62

Results

Genome-wide association meta-analyses

Manhattan plots for the most well-powered GWAS are shown in Figure 1 (fearanx-dep and

GADanx-dep in the full meta-analysis). Manhattan plots for all other GWAS with genome-wide

significant loci are shown in Supplementary Figures 2, 4, 14 and 16. We found little evidence

of confounding in all GWAS (LDSC intercept = 0.96-1.04). LDSC intercepts >1 indicate some

inflation, which is expected as the GWAS sample size increases44. The majority of inflation in

our meta-analyses was due to polygenicity (90-97%), as indicated by LDSC attenuation ratio

calculations. The genomic inflation factor, LDSC intercept, attenuation ratio, and Q-Q plots for

each GWAS are shown in Supplementary Figures 1, 3, 12, 13 and 15. Regional plots for

independent genome-wide significant loci are also in Supplementary Figures 5-11.

Fear-based disorders

Across the four GWASs, we found a total of three independent genome-wide significant loci

associated with fear. In each fear GWAS, a different locus was identified, except for fearspecific

in the GLAD+ and UKB meta-analysis, where no loci reached P < 5 � 10−8. The locus with the

most significant lead variant was rs10047892 on chromosome 14 in the fearanx-dep GWAS in

the full meta-analysis (P = 2.99 � 10−8; Figure 1, upper panel). The fearspecific GWAS in the

full meta-analysis showed a locus on chromosome 5 as genome-wide significant (lead variant

rs3996354; P = 4.71 � 10−8). A locus on chromosome 1 with lead variant rs11576254 was

found in the fearanx-dep

GWAS in the GLAD+ and UKB meta-analysis

(P = 4.70 � 10−8; Table 2).

Generalised anxiety disorder

In all, we identified four independent genome-wide significant loci associated with GAD

(P < 5 � 10−8), although not all four loci reached genome-wide significance in each GAD

GWAS. The most significantly associated locus was on chromosome 9 (lead variant

rs10120318; P = 3.1 � 10−13; Figure 1, lower panel) and was genome-wide significant in all

four GAD GWASs (Table 2). The locus on chromosome 6 was the second most significant

(lead variant rs3858; P = 1.1 � 10−9; Figure 1, lower panel) and reached genome-wide

significance in all GWAS meta-analyses of GAD, except for GADspecific in the UKB and GLAD+

meta-analysis. The locus on chromosome 2 exceeded genome-wide significance in the

GADspecific GWAS in the full meta-analysis (lead variant rs11688767; P = 4.20 � 10−8), and a

locus on chromosome 5 in the GADspecific GWAS in the UKB and GLAD+ meta-analysis (lead

variant rs3095951; P = 4.33 � 10−8; Table 2).

63

Figure 1: Genome-wide association study Manhattan plots for anxiety disorder

phenotypes meta-analysed across the GLAD+, QIMR and UKB datasets.

Fear-based disorder GWAS (upper panel) and GAD GWAS (lower panel) results from

phenotypes with controls screened for any anxiety disorder and depression. Dashed line

red; common genetic variant genome-wide significance threshold (P < 5 � 10−8), black;

suggestive significance threshold (P < 1 � 10−5).

64

Table 2: Independent genome-wide significant loci associated with anxiety disorder phenotypes

Anxiety

disorder Datasets

Control

screening

criteria

Locus

no. Region CHR Lead SNP Base pair A1:A2 Func; gene (kb)

P

OR

N cand.

SNPs

Ind. Sig.

SNPs

Previous report in

GWAS catalog

Fear

QIMR,

GLAD+ &

UKB

Anx-dep

1 14q24.3 14 rs10047892 75111346 T:C

intergenic;

AREL1 (8793)

2.99E-08 1.07 26 rs10047892

EA, worry, systolic blood

pressure

Specific

2 5q31.3

5 rs3996354 143264699 A:G

intergenic; CTB-

57H20.1 (56361)

4.71E-08 0.93 20 rs3996354

Novel

GLAD+ &

UKB

Anx-dep

3 1p34.1

1 rs11576254 44835833 T:C

intergenic; ERI3

(14900)

4.70E-08 1.09 64 rs11576254

None at genome-wide

significance. At suggestive

threshold: metabolite

levels, reaction time,

anxiety

Specific

-

-

-

-

-

-

-

-

-

-

GAD

QIMR,

GLAD+ &

UKB

Anx-dep

4

9p23

9 rs10120318 11645069 A:T

intergenic; RP11-

23D5.1 (368754)

3.07E-13 0.93 301

rs10120318;

rs11515172

Neuroticism, depression,

wellbeing, any anxiety

disorder, GAD symptoms,

worry, BMI

5 6p22.1

6

rs385816 29480224 A:G

intergenic;

XXbac-

BPG13B8.10

(1890)

1.08E-09 0.92 184 rs385816

SCZ, HDL, BIP, cognitive

performance, depression,

cardiometabolic &

hematological traits, lung

cancer, worry, smoking,

neuroticism

Specific

4

9p23

9 rs10960024 11616820 C:G

intergenic; RP11-

23D5.1 (340505)

7.17E-13 0.93 301

rs10960024;

rs11515172

As locus 4 above

5 6p22.1

6

rs385816 29480224 A:G

intergenic;

XXbac-

BPG13B8.10

(1890)

7.32E-09 0.93 184 rs385816

As locus 5 above

6 2p16.1

2 rs11688767 57988194 A:T

ncRNA intronic;

CTD-2026C7.1

(0)

4.20E-08 0.95 19 rs11688767

Cross-disorder, sleep,

SCZ, cognitive ability,

depression, neuroticism,

epilepsy, neuroticism

65

Table 2: Independent genome-wide significant loci associated with anxiety disorder phenotypes

Anxiety

disorder Datasets

Control

screening

criteria

Locus

no. Region CHR Lead SNP Base pair A1:A2 Func; gene (kb)

P

OR

N cand.

SNPs

Ind. Sig.

SNPs

Previous report in

GWAS catalog

GAD

GLAD+ &

UKB

Anx-dep

4

9p23

9 rs17189482 11513617 T:G

intergenic; RP11-

23D5.1 (237302)

1.24E-11 1.09 295

rs17189482;

rs11515172

As locus 4 above

5 6p22.1

6 rs3117427 29274136 T:C

upstream;

OR14J1 (266)

8.70E-09 1.09 219 rs3117427

As locus 5 above

7

5q34

5 rs3095951 164616460 A:G

intergenic; CTC-

340A15.2

(17810)

4.33E-08 1.06 142 rs3095951

Any anxiety disorder,

depression, neuroticism,

wellbeing, BMI

Specific

4

9p23

9 rs10960024 11616820 C:G

intergenic; RP11-

23D5.1 (340505)

1.63E-11 0.92 250 rs10960024

As locus 4 above

Locus no. = annotated locus number; CHR = chromosome; A1 = effect allele; A2 = noneffect allele; Func gene (kb) = functional consequence of the SNP on the nearest gene

with distance in kilobase; OR = odds ratio; N cand. SNPs = Number of genome-wide candidate SNPs that are in LD (r2 = 0.6) with Ind. Sig. SNPs; Ind. Sig. SNPs = independent

significant SNPs in the locus. Previous report = GWASs of other phenotypes that reported a genome-wide association with one or more of the candidate SNPs identified in this

study (full results shown in Supplementary Tables 4 & 5for fear and GAD loci, respectively). EA = educational attainment; BMI = body mass index; SCZ = schizophrenia, HDL

= high-density lipoprotein cholesterol, BIP = bipolar disorder.

66

Gene-level and gene-set association analyses

Fear-based disorders

Three genes were significantly associated with fear-based disorders after Bonferroni

correction for the number of genes tested (see Table 3 for P-value thresholds). The gene

SNX29 on chromosome 16 was significantly associated with fear in all GWASs. The gene

YWHAG on chromosome 7 was only significantly associated with both fear GWAS in the full

meta-analysis, whilst the gene ERI3 on chromosome 1 was only significantly associated with

both fear GWAS in the GLAD+ and UKB meta-analysis (Table 3). The gene ERI3 includes

the SNP-level genome-wide significant locus 1p34.1. No gene sets were significantly

associated with fear (Bonferroni correction threshold for 10678 gene sets = P ≤ 4.68 � 10−6).

Generalised anxiety disorder

A total of three genes were significantly associated with GAD. All three genes were identified

in the GADanx-dep GWAS in the full meta-analysis. The most significantly associated gene was

TMEM106B on chromosome 7. This gene was also found in GADspecific GWAS in the full meta-

analysis and the GADanx-dep GWAS in the GLAD+ and UKB meta-analysis. The gene TRIM31

on chromosome 6 was also associated with GADspecific GWAS in the full meta-analysis. This

gene region includes common genetic variants on locus 6p22.1 identified in SNP-level

association analyses. The gene SORCS3 on chromosome 10 was only associated with

GADanx-dep GWAS in the full meta-analysis (Table 3). No gene sets were significantly

associated with GAD (P ≤ 4.68 � 10−6).

67

Table 3: Gene-level associations with anxiety disorder phenotypes.

Anxiety

disorder Datasets

Control

screening

criteria

Gene

symbol CHR

N

SNPS Z

P

P Bonf.

Threshold

(0.05/

N genes) Previous report in GWAS catalog

Gene description

Fear

QIMR,

GLAD+ &

UKB

Anx-dep

SNX29

16 1560 5.09 1.77E-07

2.702E-6

(18502

genes)

EA, insomnia, smoking, IQ

Sorting nexin 29

YWHAG

7

46 4.67 1.49E-06

Blood protein levels. Suggestive

significance: SCZ, multiple sclerosis

Tyrosine 3-

monooxygenase/tryptophan 5-

monooxygenase activation protein

gamma

Specific

SNX29

16 1560 4.97 3.31E-07 2.702E-

6
(18502)

As above

As above

YWHAG

7

46 4.64 1.75E-06

As above

As above

GLAD+ &

UKB

Anx-dep

SNX29

16 1561 5.17 1.19E-07

2.698E-6

(18531)

As above

As above

ERI3

1 159 5.05 2.23E-07

EA, alcohol consumption

ERI1 exoribonuclease family member

3

Specific

SNX29

16 1562 5.07 1.96E-07 2.698E-6

(18531)

As above

As above

ERI3

1 159 4.62 1.94E-06

As above

As above

GAD

QIMR,

GLAD+ &

UKB

Anx-dep

TMEM106B 7 125 4.98 3.22E-07

2.702E-6

(18503)

Depression, EA, wellbeing, insomnia,

Alzheimer’s, neuroticism, GAD, PTSD,

HDL

Transmembrane protein 106B

SORCS3 10 1422 4.92 4.33E-07

EA, depression, blood pressure,

externalising behaviours, CAD,

insomnia, neuroticism, smoking, ADHD

Sortilin related VPS10 domain

containing receptor 3

TRIM31

6

45 4.81 7.74E-07

Cholesterol, depression, insomnia, IQ

Tripartite motif containing 31

Specific

TMEM106B 7 125 4.71 1.27E-06 2.702E-6

(18503)

As above

As above

TRIM31

6

45 4.58 2.31E-06

As above

GLAD+ &

UKB

Anx-dep TMEM106B 7 125 4.58 2.27E-06

2.698E-6

(18529)

As above

As above

Specific

-

-

-

-

-

2.698E-6

(18530)

-

-

Gene-level association analyses conducted in MAGMA. CHR = chromosome; Previous report = GWASs of other phenotypes that report genome-wide significant

loci that map onto the gene-level associations identified in this study (Supplementary Tables 6 & 7); EA = educational attainment; IQ = general cognitive ability;

SCZ = schizophrenia; GAD = generalised anxiety disorder; PTSD = post-traumatic stress disorder; HDL = high-density lipoprotein cholesterol; CAD = coronary

artery disease; ADHD = Attention deficit hyperactivity disorder.

68

SNP-based heritability

We report results here for the most powerful GWAS in the full meta-analyses, as measured

by GWAS mean Χ2, and present results for other GWAS in the supplementary. We identified

modest liability-scale LDSC h2

SNP for Fearanx-dep of 9.6% (SE = 0.006, mean Χ2 = 1.17) and

10% for GADanx-dep (SE = 0.006, mean Χ2 = 1.25), assuming a population prevalence of 8.8%

and 12.9%, respectively. Population prevalence was based on COPING study sample

prevalence (Supplementary Table 3). We note that this is not a population-based study, and

a proportion of participants were recruited through an inflammatory bowel disease study45.

However, COPING study anxiety disorder prevalence estimates were within the range of

epidemiological studies, which report broad estimates for panic disorder (1-5%), agoraphobia

(1-3%), social anxiety disorder (3-13%), specific phobia (8-14%), and GAD (3-13%)46,47. To

our knowledge, no previous epidemiological studies have estimated the population prevalence

of fear-based disorders as a group. Supplementary Table 8 shows h2

SNP liability-scale

estimates calculated from prevalences 10% higher and lower than COPING study sample

prevalences. Supplementary Table 8 also reports h2

SNP liability-scale estimates for

phenotypes with specifically screened controls, and in the meta-analysis excluding QIMR.

Genetic correlation between fear-based disorders and GAD

The genetic correlation between fear and GAD was high (rg= 0.85-0.96) and significantly

different from 0 when analysing both GWAS phenotypes (fearspecific–GADspecific and fearanx-dep–

GADanx-dep) in both GWAS meta-analyses datasets (full meta-analysis and GLAD+ and UKB

meta-analysis excluding depression-enriched QIMR dataset). The genetic correlation between

fear and GAD was significantly less than 1 when controls were screened specifically and when

the QIMR dataset was excluded from the meta-analysis. All other fear–GAD genetic

correlations were not significantly different from 1 (Table 4; Bonferroni-corrected significance

threshold P ≤ 0.0125).

69

Table 4: Genetic correlations (rg ) between fear-based disorder and GAD phenotypes

Datasets Control criteria

rg

se

Z

P (0)

P (1)

GLAD+ &

UKB

Specific

0.85

0.05

15.63

4.90E-55

4.57E-03

Any anx-dep

0.92

0.03

27.52

9.65E-167

0.0215

QIMR,

GLAD+ &

UKB

Specific

0.95

0.03

28.77

4.77E-182

0.1282

Any anx-dep

0.97

0.02

47.03

0.00E+00

0.1808

Genetic correlations estimated using LDSC regression. P (0) - P-value for test of genetic

correlation different from 0. P (1) - P-value for test of genetic correlation different from 1.

Genetic correlations with external phenotypes

We calculated genetic correlations between fear and GAD with 345 external phenotypes. Only

fearspecific and GADspecific in the GWAS meta-analysis excluding the depression-enriched QIMR

dataset were tested for genetic correlations with external phenotypes (as the fear–GAD

genetic correlation was significantly different from 1). There was no statistically significant

difference in genetic correlations with most external traits across fearspecific and GADspecific

(Bonferroni-corrected significance threshold P ≤ 1.45 � 10-4; Figure 2; left panel). Exceptions

were observed for educational attainment, general cognitive ability, which showed significantly

stronger negative genetic correlations with fearspecific than with GADspecific. Coronary artery

disease also showed a significantly higher positive genetic correlation with fearspecific than with

GADspecific. Conversely, a diagnosis of any bipolar disorder (type I and type II), bipolar disorder

type I, anorexia nervosa, and neuroticism showed significantly stronger positive genetic

correlationswith GADspecific than with fearspecific (Figure 2; right panel). Full genetic correlation

results with external phenotypes are shown in Supplementary Table 9.

70

Figure 2: Genetic correlations between anxiety disorder phenotypes and external traits estimated in LDSC regression.

Fear-based disorder GWAS and GAD GWAS results from phenotypes with specifically screened controls, meta-analysed across the UKB and GLAD+ datasets.

Genetic correlations with 345 phenotypes were tested (Bonferroni-corrected significance threshold P ≤ 1.45 � 10-4 ). Only traits with genetic correlation P-values

significantly different from 0 with at least one of the anxiety disorder phenotypes are shown (after Bonferonni correction for multiple testing). Full results are shown in

Supplementary Table 9. External phenotypes were tested for significantly different genetic correlations with fear-based disorders versus GAD using a block

jackknife (Bonferroni correction for significance [P ≤ 1.45 � 10-4]). Left panel: phenotypes with no statistically significant different genetic correlations between fear-

based disorders and GAD, though in the lower two segments, there was a significant association with one trait and not the other. Right panel: phenotypes with

statistically significant different genetic correlations between fear-based disorders and GAD. Bars represent standard errors. ADHD = attention deficit hyperactivity

disorder; ASD = autism spectrum disorder; BIP = bipolar disorder; GAD = generalised anxiety disorder; MDD = major depressive disorder; OCD = obsessive-

compulsive disorder; PGC = Psychiatric Genomics Consortium; PTSD = post-traumatic stress disorder; SCZ = schizophrenia; UKB = UK Biobank.

.8 1.0

#

#

#

#

#

#

#

#

#

GAD & fear rg significantly different from each other

Significantly stronger rg with fear

Significantly stronger rg with GAD

−0.6−0.4−0.20.0 0.2 0.4 0.6 0.8 1.0

Educational Attainment

College/university degree attainment

General cognitive ability

Coronary Artery Disease

Anorexia nervosa

Bipolar disorder type I

Bipolar disorder type I & II

Cross disorders (ADHD, ASD, BIP, SCZ)

Neuroticism UKB

Genetic correlation rg

#

#

Fear−based disorders: significant

GAD: significant

Fear−based disorders: non−significant

GAD: non−significant

#

#

#

#

#

#

#

#

#

#

#

#

#

#

#

#

#

#

#

#

#

#

#

#

#

#

#

#

#

GAD & fear rg not significantly different from each other

Both significantly different from 0

Only fear signif.

Only GAD signif.

−0.6−0.4−0.2 0.0 0.2 0.4 0.6 0.8 1.0

Ever smoker

Automobile speeding

ADHD

Schizophrenia

Insomnia

Childhood maltreatment

Self−rated health

Bipolar disorder type II

Subjective well−being

Self−reported tiredness

PTSD symptoms

GAD symptoms

Neuroticism, no psychiatric illness

MDD (PGC2, no 23andme, no UKB)

Depressive symptoms

Cross 12 psychiatric disorders

MDD (PGC2 + 23andme + UKB)

Lifetime anxiety

Body fat percentage

Verbal−numerical reasoning

Father's age at death

Mother's age at death

Age first birth

Undersleeper

Household income

Problematic consequences of drinking

Autism spectrum disorder

OCD

Alcohol dependence

Genetic correlation rg

College

Cross disor

71

Discussion

This is the first study to report genome-wide significant loci (three in total) associated with any

lifetime fear-based disorder diagnosis, hereafter referred to as fear. In addition, we identified

four genome-wide significant loci for lifetime GAD diagnosis, previously identified in other

GWAS of anxiety15,48 and distress-related traits49–51. The increase in power compared to

previous GWAS of fear19 was achieved by combining detailed and brief diagnostic measures.

In agreement with twin studies, we found that fear and GAD were highly genetically correlated

with each other and with depression9,11,12. When we excluded the depression-enriched dataset

and only screened controls for the specific anxiety disorder being analysed, genetic

correlations between these disorders were reduced, suggesting some genetic specificity

between them. Differences in common genetic variant-based genetic correlations between

fear and GAD with broad domains of other complex traits were explored. Compared with GAD,

fear showed significantly stronger negative genetic correlations with general cognitive ability

and educational attainment and stronger positive genetic correlations with coronary artery

disease. Conversely, neuroticism, bipolar disorder type I and anorexia nervosa were among

the traits with significantly higher positive genetic correlations with GAD than fear.

We found independent loci and gene-level associations that support previous GWAS findings

of other phenotypes and some novel associations. We identified a novel variant associated

with fear that has not been previously reported in GWASs of complex traits. The nearest gene

to the intergenic variant is the uncharacterised, non-coding gene CTB-57H20.1. Other loci and

gene-level associations with fear that we identified have been previously implicated in GWAS

of educational attainment52, general cognitive ability53, insomnia54 and broad depression49,

including sorting nexin 29. The sorting nexin family of proteins is thought to play a role in

neuronal function, synaptic plasticity, learning, and memory55. One locus associated with fear

was previously reported in a worry-neuroticism GWAS in the UK Biobank50. No fear loci or

gene-level associations overlapped with our GAD results. As GWASs of specific anxiety

disorders improve in power, we expect this overlap to increase.

Some of our loci and gene-level associations with GAD were reported in previous UK Biobank

GWASs of lifetime anxiety and GAD symptoms15,48, which is expected given the sample

overlap. Loci and gene-level associations with GAD were consistently replicated in previous

GWASs of other distress-related phenotypes, with all four loci previously identified in

depression and neuroticism49–51. This includes the SORCS3 gene, which is highly expressed

in the CA1 hippocampus region and has been implicated in aversive memory extinction in

mice56. SORCS3 was also associated with anxiety-like behaviour in mice, mediated by

changes in gut microbiome measures suggesting the role of the gut-brain axis57. Our finding

72

warrants further animal modelling and functional genomic interrogation of the role of SORCS3

in distress liability and translational potential.

The high genetic correlations we observed between fear, GAD, and depression are consistent

with previous twin studies9,11,12. Our results are compatible with our previous preliminary

GWAS in the UK Biobank19 but add further detail to the genetic structure of fear and GAD by

assessing a broad range of other complex traits. We verified that screening controls

specifically for the disorder grouping being analysed improved our ability to distinguish genetic

differences between highly genetically correlated disorders (e.g. controls in the fear GWAS

were only screened for any fear-based disorder). We found evidence of some genetic

differences between fear and GAD once we screened controls specifically and after excluding

the depression-ascertained QIMR dataset. This highlights the necessity for careful

consideration of how to screen controls when attempting to identify disorder-specific genetics.

Genetic correlations with depression were high and not significantly different between fear

and GAD, supporting a common genetic liability factor shared among these disorders9,11,12,58.

A common liability factor is often conceptualised as being driven by shared negative affectivity,

which is also well captured by measures of the personality trait neuroticism22,59. We found

neuroticism had a significantly stronger genetic correlation with GAD than fear. This is in line

with twin studies that reported genetic influences on GAD (and depression) were more core

to a higher-order dimension of genetic liability to negative affectivity than fear60. General

negative affectivity also influences fear-based disorders but is less central to them; instead,

they are characterised by more specific elements of acute fearful and physiological

hyperarousal61. Our findings partially support a distress-fear genetic distinction. However, we

found stronger evidence for an overarching genetic liability to internalising that drives

comorbidity across anxiety disorders and depression.

Compared with GAD, fear had stronger negative genetic correlations with educational

attainment and general cognitive ability and a higher positive genetic correlation with coronary

artery disease (heart disease). These correlations were non-significant with GAD. The specific

genetic relationship observed between fear and cognitive-related traits may reflect shared

cognitive mechanisms, such as those involved in learning, that are less central to GAD.

Associative learning is a key mechanism that distinguishes fear-based disorders from GAD6.

Our finding that heart disease is uniquely genetically correlated with fear may be explained by

genetics playing a bigger part in driving comorbidity between heart disease and fear than with

GAD. Core symptoms of fear-based disorders, including experiencing phobic fears and

somatic arousal, are linked to a higher risk of heart disease and cardiac fatality 62–65. Our

findings align with a phenotypic study that revealed fear-based disorders were stronger

73

predictors of heart disease development than distress disorders66. The earlier age of onset of

chronically experienced fear-based disorders compared to distress may give rise to longer-

term exposure to mechanisms associated with heart disease risk1,66,67. Overlapping genetics

with heart disease may underpin shared physiological mechanisms in inflammatory pathways

or autonomic dysfunction68.

GAD showed significantly stronger positive genetic correlations than fear with a diagnosis of

any bipolar disorder (type I or II). Bipolar type II disorder involves episodes of hypomania and

is more similar to depression, whereas bipolar type I disorder is characterised by more severe

manic episodes4. When we analysed bipolar disorder I and II separately, we found GAD and

fear had no significantly different genetic correlation with bipolar II, whereas bipolar I was

significantly more positively genetically correlated with GAD. This indicates that bipolar II

drives down the genetic correlations with fear when analysed together with bipolar I. This

aligns with previous studies that found bipolar I had a lower genetic correlation with depression

than bipolar II and was more genetically similar to rarer psychiatric illnesses, including

schizophrenia69,70. Genetic correlation analyses of the mood disorder spectrum reflect a

genetic continuum from manic to depressive clusters. Bipolar I sits on one end of the spectrum,

and depressive disorders on the other, with bipolar type II connecting the two disorders71. Our

novel finding adds further fine-grained detail to this genetic structure by highlighting

heterogeneity at the internalising pole of the genetic continuum. On the opposite end of the

spectrum to bipolar type I lies fear-based disorders, with distress disorders (GAD and

depression) sitting between bipolar and fear-based disorders.

We found that fear had a significantly weaker genetic correlation with anorexia nervosa (~0.10)

than GAD (~0.30). This is somewhat unexpected given that fear-based disorders have similar

rates of comorbidity with anorexia nervosa compared to GAD, thought to be driven by an

overlap in genetic susceptibility and fear-conditioning mechanisms72–74. Few studies have

assessed genetic correlations specifically between anorexia nervosa and the individual

anxiety disorder subtypes, with analyses restricted to GAD75. However, the genetic overlap

between other eating disorder phenotypes and fear-based disorders has been reported76,77.

One twin study found that a broad eating disorder phenotype had moderate genetic

correlations with panic disorder (0.62) and GAD (0.51) and a slightly lower genetic correlation

with specific phobias (0.40), agoraphobia (0.38), and social anxiety disorder (0.33)77. Thus,

further research is required to establish whether genetic correlations with anorexia nervosa

differ between individual fear-based disorders. There may also be shared loci that affect traits

in different directions, making us unable to detect a global genetic correlation. Studies

conducting local genetic correlation analyses may identify specific shared regions of the

74

genome supporting shared mechanisms that link fear-based disorders with anorexia

nervosa78,79.

Our study is the first to identify fear-based disorder-associated genetic loci. However, findings

should be considered in light of limitations that impacted our ability to detect fear-specific

genetic influences. First, we were underpowered to assess the fear-based disorders

individually. Some twin studies suggest the proportion of disorder-specific genetic influences

on individual fear-based disorders varies11,12, with specific phobia being less genetically similar

to GAD than panic disorder and agoraphobia11. Second, results may have differed if we had

sufficient sample sizes to only use panic disorder diagnosis instead of combining it with panic

attack measures. Panic attacks are transdiagnostic and can be included as a specifier for a

range of DSM-5 psychiatric disorders4. Using self-reports of panic attacks may have elevated

the genetic correlations between fear and other psychiatric disorders. However,

epidemiological estimates of lifetime panic disorder comorbidity with other mental disorders

are also high (80%)80. Third, specifically screening controls for a certain anxiety disorder was

not possible in the QIMR data as we were limited to using a single-item brief measure of any

anxiety disorder. Fourth, although a strength of our study was incorporating detailed measures

of anxiety disorders, we could not compare genetic differences between brief versus detailed

diagnostic measures. Studies suggest that more detailed measures increase genetic

specificity25. Combining detailed with brief measures may have limited genetic specificity and

elevated the genetic correlation between fear, GAD, and related disorders such as depression.

Finally, we could not account for the high comorbidity observed across anxiety disorders and

depression. We previously found that cases rarely met criteria for only one anxiety disorder

and no comorbid MDD (5%) in the GLAD+ dataset81. Thus, ascertaining data for well-powered

GWAS of a single anxiety disorder without comorbidities will be challenging and are less

representative of the broader lived experience of these disorders. We did not include heritable

comorbid traits as GWAS covariates as this can bias results82. Future studies with well-

powered GWAS of anxiety subtypes could discern subtype-specific genetics through

multivariate genomic structural equation modelling83. As individual-level methods advance84,85,

this will enable testing genetic covariances across comorbid disorders and comparing

measures in smaller samples than is required for GWAS summary-level modelling.

In summary, our results add further evidence that fear-based disorders and GAD share a high

proportion of genetic liability, as well as with depression, while also revealing some genetic

differences between them. We identified differences in the genetic relationship between fear

and GAD, including those with neuroticism, bipolar type I, anorexia nervosa, general cognitive

ability, and coronary artery disease. Our findings add more fine-grained detail to the proposed

75

hierarchical structure of internalising disorders and partially support fear-based disorders and

GAD as separate subfactors of distress and fear. Quantitatively-based dimensional modelling

of shared and distinct distress and fear symptoms on the genomic level would further complete

this hierarchical structure22. The growth of datasets with detailed phenotyping of all anxiety

disorders, such as those used in this study, will be key for further identifying subtype-specific

and transdiagnostic genetic factors of the full anxiety disorder spectrum.

76

References

1. Kessler, R. C. et al. Lifetime prevalence and age-of-onset distributions of DSM-IV

disorders in the National Comorbidity Survey Replication. Arch. Gen. Psychiatry 62,

593–602 (2005).

2. Somers, J. M., Goldner, E. M., Waraich, P. & Hsu, L. Prevalence and incidence

studies of anxiety disorders: a systematic review of the literature. Can. J. Psychiatry

51, 100–113 (2006).

3. Craske, M. G. et al. Anxiety disorders. Nat Rev Dis Primers 3, 17024 (2017).

4. American Psychiatric Association. Diagnostic and Statistical Manual of Mental

Disorders (DSM-5�). (American Psychiatric Pub, 2013).

5. Kessler, R. C. The global burden of anxiety and mood disorders: putting the

European Study of the Epidemiology of Mental Disorders (ESEMeD) findings into

perspective. J. Clin. Psychiatry 68 Suppl 2, 10–19 (2007).

6. Craske, M. G. & Waters, A. M. Panic Disorder, Phobias, and Generalized Anxiety

Disorder. Annual Review of Clinical Psychology vol. 1 197–225 Preprint at

https://doi.org/10.1146/annurev.clinpsy.1.102803.143857 (2005).

7. Polderman, T. J. C. et al. Meta-analysis of the heritability of human traits based on

fifty years of twin studies. Nat. Genet. 47, 702–709 (2015).

8. Ask, H. et al. Genetic contributions to anxiety disorders: where we are and where we

are heading. Psychol. Med. 51, 2231–2246 (2021).

9. Waszczuk, M. A., Zavos, H. M. S., Gregory, A. M. & Eley, T. C. The phenotypic and

genetic structure of depression and anxiety disorder symptoms in childhood,

adolescence, and young adulthood. JAMA Psychiatry 71, 905–916 (2014).

10. Roberson-Nay, R., Eaves, L. J., Hettema, J. M., Kendler, K. S. & Silberg, J. L.

Childhood separation anxiety disorder and adult onset panic attacks share a common

genetic diathesis. Depress. Anxiety 29, 320–327 (2012).

11. Hettema, J. M., Prescott, C. A., Myers, J. M., Neale, M. C. & Kendler, K. S. The

structure of genetic and environmental risk factors for anxiety disorders in men and

women. Arch. Gen. Psychiatry 62, 182–189 (2005).

12. Tambs, K. et al. Structure of genetic and environmental risk factors for dimensional

representations of DSM–IV anxiety disorders. Br. J. Psychiatry 195, 301–307 (2009).

13. Kendler, K. S., Prescott, C. A., Myers, J. & Neale, M. C. The structure of genetic and

environmental risk factors for common psychiatric and substance use disorders in

men and women. Arch. Gen. Psychiatry 60, 929–937 (2003).

14. Smoller, J. W. Anxiety Genetics Goes Genomic. The American journal of psychiatry

vol. 177 190–194 (2020).

15. Purves, K. L. et al. A major role for common genetic variation in anxiety disorders.

Mol. Psychiatry 25, 3292–3303 (2020).

16. Levey, D. F. et al. Reproducible Genetic Risk Loci for Anxiety: Results From

∼200,000 Participants in the Million Veteran Program. Am. J. Psychiatry 177, 223–

232 (2020).

17. Meier, S. M. et al. Genetic Variants Associated With Anxiety and Stress-Related

Disorders: A Genome-Wide Association Study and Mouse-Model Study. JAMA

Psychiatry (2019) doi:10.1001/jamapsychiatry.2019.1119.

18. Wendt, F. R. et al. Using phenotype risk scores to enhance gene discovery for

generalized anxiety disorder and posttraumatic stress disorder. Mol. Psychiatry

(2022) doi:10.1038/s41380-022-01469-y.

19. Morneau-Vaillancourt, G. et al. The genetic and environmental hierarchical structure

of anxiety and depression in the UK Biobank. Depress. Anxiety 37, 512–520 (2020).

20. Deckert, J. et al. GLRB allelic variation associated with agoraphobic cognitions,

increased startle response and fear network activation: a potential neurogenetic

pathway to panic disorder. Mol. Psychiatry 22, 1431–1439 (2017).

77

21. Erhardt, A. et al. TMEM132D, a new candidate for anxiety phenotypes: evidence

from human and mouse studies. Mol. Psychiatry 16, 647–663 (2011).

22. Watson, D. et al. Validity and utility of Hierarchical Taxonomy of Psychopathology

(HiTOP): III. Emotional dysfunction superspectrum. World Psychiatry 21, 26–54

(2022).

23. Toft, T. et al. Mental disorders in primary care: prevalence and co-morbidity among

disorders. results from the functional illness in primary care (FIP) study. Psychol.

Med. 35, 1175–1184 (2005).

24. Kendler, K. S., Chatzinakos, C. & Bacanu, S.-A. The impact on estimations of genetic

correlations by the use of super-normal, unscreened, and family-history screened

controls in genome wide case-control studies. Genet. Epidemiol. 44, 283–289 (2020).

25. Cai, N., Choi, K. W. & Fried, E. I. Reviewing the genetics of heterogeneity in

depression: operationalizations, manifestations and etiologies. Hum. Mol. Genet. 29,

R10–R18 (2020).

26. Flint, J. The genetic basis of major depressive disorder. Mol. Psychiatry (2023)

doi:10.1038/s41380-023-01957-9.

27. Davies, M. R. et al. Comparison of symptom-based versus self-reported diagnostic

measures of anxiety and depression disorders in the GLAD and COPING cohorts. J.

Anxiety Disord. 85, 102491 (2022).

28. Mitchell, B. L. et al. The Australian Genetics of Depression Study: New Risk Loci and

Dissecting Heterogeneity Between Subtypes. Biol. Psychiatry (2021)

doi:10.1016/j.biopsych.2021.10.021.

29. Davis, K. A. S. et al. Mental health in UK Biobank - development, implementation and

results from an online questionnaire completed by 157 366 participants: a reanalysis.

BJPsych Open 6, e18 (2020).

30. Gigantesco, A. & Morosini, P. Development, reliability and factor analysis of a self-

administered questionnaire which originates from the World Health Organization’s

Composite International Diagnostic Interview – Short Form (CIDI-SF) for assessing

mental disorders. Clinical Practice and Epidemiology in Mental Health vol. 4 8

Preprint at https://doi.org/10.1186/1745-0179-4-8 (2008).

31. Byrne, E. M. et al. Cohort profile: the Australian genetics of depression study. BMJ

Open 10, e032580 (2020).

32. Davies, M. R. et al. The Genetic Links to Anxiety and Depression (GLAD) Study:

Online recruitment into the largest recontactable study of depression and anxiety.

Behav. Res. Ther. 123, 103503 (2019).

33. Young, K. S. et al. Depression, anxiety and PTSD symptoms before and during the

COVID-19 pandemic in the UK. Psychol. Med. 1–14 (2022).

34. Olsen, C. M. et al. Cohort profile: the QSkin Sun and Health Study. Int. J. Epidemiol.

41, 929–929i (2012).

35. Sudlow, C. et al. UK biobank: an open access resource for identifying the causes of a

wide range of complex diseases of middle and old age. PLoS Med. 12, e1001779

(2015).

36. Mbatchou, J. et al. Computationally efficient whole-genome regression for

quantitative and binary traits. Nat. Genet. 53, 1097–1103 (2021).

37. Zhou, W. et al. Efficiently controlling for case-control imbalance and sample

relatedness in large-scale genetic association studies. Nat. Genet. 50, 1335–1341

(2018).

38. Willer, C. J., Li, Y. & Abecasis, G. R. METAL: fast and efficient meta-analysis of

genomewide association scans. Bioinformatics 26, 2190–2191 (2010).

39. Watanabe, K., Taskesen, E., van Bochoven, A. & Posthuma, D. Functional mapping

and annotation of genetic associations with FUMA. Nat. Commun. 8, 1826 (2017).

40. de Leeuw, C. A., Mooij, J. M., Heskes, T. & Posthuma, D. MAGMA: generalized

gene-set analysis of GWAS data. PLoS Comput. Biol. 11, e1004219 (2015).

78

41. Bulik-Sullivan, B. K. et al. LD Score regression distinguishes confounding from

polygenicity in genome-wide association studies. Nature Genetics vol. 47 291–295

Preprint at https://doi.org/10.1038/ng.3211 (2015).

42. Bulik-Sullivan, B. et al. An atlas of genetic correlations across human diseases and

traits. Nat. Genet. 47, 1236–1241 (2015).

43. H�bel, C. et al. Genomics of body fat percentage may contribute to sex bias in

anorexia nervosa. Am. J. Med. Genet. B Neuropsychiatr. Genet. 180, 428–438

(2019).

44. Loh, P.-R., Kichaev, G., Gazal, S., Schoech, A. P. & Price, A. L. Mixed-model

association for biobank-scale datasets. Nat. Genet. 50, 906–908 (2018).

45. Davies, H. L. et al. Risk and protective factors for new-onset binge eating, low

weight, and self-harm symptoms in> 35,000 individuals in the UK during the COVID-

19 pandemic. Int. J. Eat. Disord. (2022).

46. Bandelow, B. & Michaelis, S. Epidemiology of anxiety disorders in the 21st century.

Dialogues Clin. Neurosci. 17, 327–335 (2015).

47. Leray, E. et al. Prevalence, characteristics and comorbidities of anxiety disorders in

France: Results from the ‘Mental Health in General Population’ Survey (MHGP). Eur.

Psychiatry 26, 339–345 (2011).

48. Thorp, J. G. et al. Symptom-level modelling unravels the shared genetic architecture

of anxiety and depression. Nat Hum Behav 5, 1432–1442 (2021).

49. Howard, D. M. et al. Genome-wide meta-analysis of depression identifies 102

independent variants and highlights the importance of the prefrontal brain regions.

Nat. Neurosci. 22, 343–352 (2019).

50. Nagel, M. et al. Meta-analysis of genome-wide association studies for neuroticism in

449,484 individuals identifies novel genetic loci and pathways. Nat. Genet. 50, 920–

927 (2018).

51. Wray, N. R. et al. Genome-wide association analyses identify 44 risk variants and

refine the genetic architecture of major depression. Nat. Genet. 50, 668–681 (2018).

52. Okbay, A. et al. Polygenic prediction of educational attainment within and between

families from genome-wide association analyses in 3 million individuals. Nat. Genet.

54, 437–449 (2022).

53. Davies, G. et al. Study of 300,486 individuals identifies 148 independent genetic loci

influencing general cognitive function. Nat. Commun. 9, 2098 (2018).

54. Watanabe, K. et al. Genome-wide meta-analysis of insomnia prioritizes genes

associated with metabolic and psychiatric pathways. Nat. Genet. 54, 1125–1132

(2022).

55. Vieira, N., Rito, T., Correia-Neves, M. & Sousa, N. Sorting Out Sorting Nexins

Functions in the Nervous System in Health and Disease. Mol. Neurobiol. 58, 4070–

4106 (2021).

56. Breiderhoff, T. et al. Sortilin-related receptor SORCS3 is a postsynaptic modulator of

synaptic depression and fear extinction. PLoS One 8, e75006 (2013).

57. Jin, X. et al. Gut microbiome partially mediates and coordinates the effects of

genetics on anxiety-like behavior in Collaborative Cross mice. Sci. Rep. 11, 270

(2021).

58. Mosing, M. A. et al. Genetic and environmental influences on the co-morbidity

between depression, panic disorder, agoraphobia, and social phobia: a twin study.

Depress. Anxiety 26, 1004–1011 (2009).

59. Hettema, J. M., Neale, M. C., Myers, J. M., Prescott, C. A. & Kendler, K. S. A

population-based twin study of the relationship between neuroticism and internalizing

disorders. Am. J. Psychiatry 163, 857–864 (2006).

60. Kendler, K. S. & Myers, J. The boundaries of the internalizing and externalizing

genetic spectra in men and women. Psychol. Med. 44, 647–655 (2014).

61. Watson, D. Rethinking the mood and anxiety disorders: a quantitative hierarchical

model for DSM-V. J. Abnorm. Psychol. 114, 522–536 (2005).

79

62. Watkins, L. L. et al. Phobic anxiety and increased risk of mortality in coronary heart

disease. Psychosom. Med. 72, 664–671 (2010).

63. Albert, C. M., Chae, C. U., Rexrode, K. M., Manson, J. E. & Kawachi, I. Phobic

anxiety and risk of coronary heart disease and sudden cardiac death among women.

Circulation 111, 480–487 (2005).

64. Nabi, H. et al. Psychological and somatic symptoms of anxiety and risk of coronary

heart disease: the health and social support prospective cohort study. Biol.

Psychiatry 67, 378–385 (2010).

65. Roest, A. M., Heideveld, A., Martens, E. J., de Jonge, P. & Denollet, J. Symptom

dimensions of anxiety following myocardial infarction: associations with depressive

symptoms and prognosis. Health Psychol. 33, 1468–1476 (2014).

66. Roest, A. M. et al. Fear and distress disorders as predictors of heart disease: A

temporal perspective. J. Psychosom. Res. 96, 67–75 (2017).

67. Lijster, J. M. de et al. The Age of Onset of Anxiety Disorders. Can. J. Psychiatry 62,

237–246 (2017).

68. Celano, C. M., Daunis, D. J., Lokko, H. N., Campbell, K. A. & Huffman, J. C. Anxiety

Disorders and Cardiovascular Disease. Curr. Psychiatry Rep. 18, 101 (2016).

69. Mallard, T. T. et al. Multivariate GWAS of psychiatric disorders and their cardinal

symptoms reveal two dimensions of cross-cutting genetic liabilities. Cell Genom 2,

(2022).

70. Mullins, N. et al. Genome-wide association study of more than 40,000 bipolar

disorder cases provides new insights into the underlying biology. Nat. Genet. 53,

817–829 (2021).

71. Coleman, J. R. I. et al. The Genetics of the Mood Disorder Spectrum: Genome-wide

Association Analyses of More Than 185,000 Cases and 439,000 Controls. Biol.

Psychiatry (2019) doi:10.1016/j.biopsych.2019.10.015.

72. Meier, S. M. et al. Diagnosed Anxiety Disorders and the Risk of Subsequent

Anorexia Nervosa: A Danish Population Register Study. Eur. Eat. Disord. Rev. 23,

524–530 (2015).

73. Kaye, W. H., Bulik, C. M., Thornton, L., Barbarich, N. & Masters, K. Comorbidity of

anxiety disorders with anorexia and bulimia nervosa. Am. J. Psychiatry 161, 2215–

2221 (2004).

74. Godart, N. T., Flament, M. F., Perdereau, F. & Jeammet, P. Comorbidity between

eating disorders and anxiety disorders: a review. Int. J. Eat. Disord. 32, 253–270

(2002).

75. Dellava, J. E., Kendler, K. S. & Neale, M. C. Generalized anxiety disorder and

anorexia nervosa: evidence of shared genetic variation. Depress. Anxiety 28, 728–

733 (2011).

76. Kendler, K. S. et al. The structure of the genetic and environmental risk factors for six

major psychiatric disorders in women. Phobia, generalized anxiety disorder, panic

disorder, bulimia, major depression, and alcoholism. Arch. Gen. Psychiatry 52, 374–

383 (1995).

77. Kendler, K. S. et al. The structure of genetic and environmental risk factors for

syndromal and subsyndromal common DSM-IV axis I and all axis II disorders. Am. J.

Psychiatry 168, 29–39 (2011).

78. Werme, J., van der Sluis, S. & de Leeuw, C. A. LAVA: An integrated framework for

local genetic correlation analysis. bioRxiv 2020.12.31.424652 (2021)

doi:10.1101/2020.12.31.424652.

79. Ronald, A. & Pain, O. A Revolution Is Brewing in How We Understand the Shared

Genetic Causes of Psychiatric Disorders. The American journal of psychiatry vol. 179

791–793 (2022).

80. de Jonge, P., Roest, A. M., Lim, C. C. W., Levinson, D. & Scott, K. M. Panic Disorder

and Panic Attacks. in Mental Disorders Around the World: Facts and Figures from the

WHO World Mental Health Surveys 93–105 (Cambridge University Press, 2018).

80

81. Davies, M. R. et al. Factors associated with anxiety disorder comorbidity. J. Affect.

Disord. 323, 280–291 (2023).

82. Aschard, H., Vilhj�lmsson, B. J., Joshi, A. D., Price, A. L. & Kraft, P. Adjusting for

heritable covariates can bias effect estimates in genome-wide association studies.

Am. J. Hum. Genet. 96, 329–339 (2015).

83. Grotzinger, A. D. et al. Genomic structural equation modelling provides insights into

the multivariate genetic architecture of complex traits. Nat Hum Behav 3, 513–525

(2019).

84. de Hoyos, L. et al. Structural models of genome-wide covariance identify multiple

common dimensions in autism. bioRxiv (2022) doi:10.1101/2022.10.21.22281213.

85. Pritikin, J. N., Neale, M. C., Prom-Wormley, E. C., Clark, S. L. & Verhulst, B. GW-

SEM 2.0: Efficient, Flexible, and Accessible Multivariate GWAS. Behav. Genet. 51,

343–357 (2021).

81

Chapter 3. Genetic decomposition of the heritable

component of reported childhood maltreatment

This chapter is presented as a manuscript of an accepted paper in Biological Psychiatry:

Global Open Science. It is an exact copy of this manuscript. Supplementary materials for this

chapter, as detailed in the text, are included in Appendix B.

In press citation: ter Kuile, A. R., H�bel, C., Cheesman, R., Coleman, J. R. I., Peel, A. J.,

Levey, D. F., Stein, M. B., Gelernter, J., Rayner, C., Eley, T. C., & Breen, G. (In press). Genetic

decomposition of the heritable component of reported childhood maltreatment. Biological

Psychiatry Global Open Science. https://doi.org/10.1016/j.bpsgos.2023.03.003

Authors: Abigail R. ter Kuile1,2, Christopher H�bel1,2,3, Rosa Cheesman8, Jonathan R. I.

Coleman1,2, Alicia J. Peel1, Daniel F. Levey4,5, Murray B. Stein6,7, Joel Gelernter4,5,

Christopher Rayner1, Thalia C. Eley1,2, Gerome Breen1,2

1. Social, Genetic & Developmental Psychiatry Centre, Institute of Psychiatry,

Psychology & Neuroscience, King’s College London, UK

2. National Institute for Health and Care Research (NIHR) Maudsley Biomedical

Research Centre at South London and Maudsley NHS Foundation Trust, London, UK

3. National Centre for Register-based Research, Aarhus Business and Social Sciences,

Aarhus University, Aarhus, Denmark

4. Department of Psychiatry, Yale School of Medicine, New Haven, CT, USA

5. Veterans Affairs Connecticut Healthcare System, West Haven, CT, USA

6. Department of Psychiatry and School of Public Health, University of California San

Diego, La Jolla, CA, USA

7. VA San Diego Healthcare System, San DIego, CA, USA

8. PROMENTA Research Center, Department of Psychology, University of Oslo, Oslo,

Norway

82

Abstract

Background: Decades of research have shown that environmental exposures, including self-

reports of trauma, are partly heritable. Heritable characteristics may influence exposure to and

interpretations of environmental factors. Identifying heritable factors associated with self-

reported trauma could improve our understanding of vulnerability to exposure and the

interpretation of life events.

Methods: We used genome-wide association study summary statistics of childhood

maltreatment, defined as reporting of abuse (emotional, sexual and physical) and neglect

(emotional and physical) (N=185,414). We calculated genetic correlations (rg) between

reported childhood maltreatment and 576 traits to identify phenotypes that might explain the

heritability of reported childhood maltreatment, retaining those with |rg|>0.25. We specified

multiple regression models using genomic structural equation modelling to detect residual

genetic variance in childhood maltreatment after accounting for genetically correlated traits.

Results: In two separate models, the shared genetic component of twelve health and

behavioural traits and seven psychiatric disorders accounted for 59% and 56% of heritability

due to common genetic variants (h2

SNP) of childhood maltreatment, respectively. Genetic

influences on the h2

SNP of childhood maltreatment were generally accounted for by a shared

genetic component across traits. The exceptions to this were general risk tolerance, subjective

well-being, post-traumatic stress disorder and autism spectrum disorder, identified as

independent contributors to its h2

SNP. These four traits alone were sufficient to explain 58% of

the h2

SNP of childhood maltreatment.

Conclusions: We identified putative traits that reflect the h2

SNP of childhood maltreatment.

Elucidating the mechanisms underlying these associations may improve trauma prevention

and posttraumatic intervention strategies.

83

Introduction

Traumatic events, namely those perceived as physically or emotionally threatening and

violating, are associated with various adverse outcomes, including psychopathology1–4.

Decades of behavioural genetics research has shown that reported trauma exposures, like

many environmental measures and behavioural traits, are partly heritable5,6. Twin studies

estimate 6-62% of the variance in reporting different types of trauma is attributable to

genetics7–12. Interpersonal assaultive traumas (e.g., physical and sexual assault) have higher

heritability than non-interpersonal or non-assaultive traumas (e.g., accidents)7,9,13. In relation

to these observations, stressful life events (SLEs) dependent on one’s behaviour (e.g., fights)

are more heritable than those that are independent (e.g., natural disasters), with the latter

occurring more often due to chance14. However, being at higher genetic risk for reported

trauma does not signify that an individual is genetically predestined to experience trauma.

Furthermore, a large proportion of the total phenotypic variability of reported trauma is not

attributable to genetics. The environment itself may be harmful, or a perpetrator may exploit

those in vulnerable circumstances11,15. However, environmental risk factors are generally

unstable, idiosyncratic, and thus, unpredictable and challenging to examine16. Exploring traits

genetically related to reported trauma in different environmental contexts may provide a

framework for social research to help determine trauma risk factors and protect vulnerable

individuals6.

Heritable behavioural characteristics may contribute to the likelihood of experiencing certain

events. Personality traits, such as openness to experience and antisocial behaviour, are

phenotypically and genetically correlated with reporting interpersonal assaultive trauma17.

Such partially heritable characteristics may contribute to the heritability of reported trauma

through gene-environment correlation (rGE), whereby the environment reflects an individual’s

genetic propensities via three different processes18. Passive rGE occurs when a relative’s

genotype, such as parental genetic variation contributing to risk-taking behaviours, shapes the

child’s environment and potentially creates an unsafe home19,20. The environment that the

parent creates and the parental genotype are correlated as the child receives both from their

biological parents. Thus, parental environmental effects may be captured in genetic analyses

of offspring traits18. Evocative rGE arises when an individual’s genotype shapes how others

engage with them. For example, a child’s behavioural difficulties may evoke verbal and

physical discipline due to the carer’s expectations of how a child should behave18. Active rGE

involves an individual’s genetic disposition to, for example, risk-taking modifying and selecting

their environment18,21, leading to differing risks of exposure to adverse environments.

84

Correlations between genetic factors and retrospective reports of trauma may also, in part, be

driven by heritable characteristics influencing the subjective interpretation, willingness to

disclose and recollection of events22,23. Genetic research has largely relied on retrospective

self-reports of trauma exposure which may be more susceptible to genetically influenced

perceptions and recollection of events, as opposed to more objective measures prospectively

recorded closer to the time of exposure (e.g., court records, caregiver reports)23. Memory,

emotional regulation and interpretation biases are partly heritable24–27 and are associated with

retrospective reporting of trauma in early life28. Individual differences in subjective experiences

are partly influenced by genetics29,30. Subjective appraisal of trauma is important for

posttraumatic psychopathology, which is more strongly associated with retrospective self-

reports of trauma than objective court records31. Individual, partially heritable differences in

personality traits such as neuroticism and agreeableness may explain the discrepancy

between retrospective and prospective measures of trauma31–33. Furthermore, the consistency

and frequency of self-reports are impacted by individual factors involved in the willingness to

disclose a traumatic event, such as perception of stigma, fear of negative consequences, or

pre-existing relationships with the perpetrator34–37. Lack of disclosure is a barrier to therapeutic

and legal interventions36. Thus, a better understanding of the heritable factors that impact the

retrospective report of trauma experiences could help improve posttraumatic support.

In sum, the influences on retrospectively reported trauma are complex and difficult to

disentangle. A range of heritable traits may be involved. Heritability and genetic correlations

between traits can be estimated using genome-wide association study (GWAS) summary

statistics38. The proportion of heritability explained by common genetic variants (h2

SNP) ranges

from 6-9% for reported interpersonal trauma during childhood21,39 to 18% during childhood and

adulthood combined40. This accounts for a large proportion of the reported twin heritabilities

estimated at 20-62%7,9–12. Reported trauma shows genetic correlations with psychiatric

disorders, current mental state, personality traits, lifestyle factors, and sociodemographic

traits21,39,40. However, these studies did not analytically explain the extent to which the h2

SNP of

reported traumas reflects genetic correlations with these complex traits. Identifying specific

traits that explain a large proportion of h2