Final Report Summary - GENG*EED (GWAS, endophenotypes and gene environment interactions in eating disorders)
Objective 1:
Our first objective of the project was:
a.) to perform genome wide association (GWAS) analysis, followed by replication analysis and advanced statistical analysis, to identify main genetic effects on the disease.
b.) to examine various endophenotypes through a process of data replication by assessing candidate genes.
Objective 1 was part of the WTCCC3 funded project 'A Genome wide Association Study of Anorexia Nervosa' in which genotyping and preliminary analysis is funded by the Welcome Trust (PI Prof Collier).
During the fellowship two GWAS studies were carried out, one assessing the clinical data collected from the GCAN project and the other one using the St.Thomas twin data looking at a general population.
1.) As regards to the GCAN (GWAS on anorexia) the following results have been achieved (Poster presented at the Eating Disorder Research Society in Edinburgh 2011):
WTCCC3 and GCAN: A Genomewide scan for anorexia nervosa
CM Bulik, DA Collier, E Zeggini, PF Sullivan, GCAN Consortium4, WTCCC .
Abstract:
The genetic consortium for anorexia nervosa (GCAN, part of WTCCC3) conducted a GWAS on 2907 cases with anorexia nervosa (AN) and 14,860 ancestrally matched controls. 94 SNPs with p-value less than 10-4 derived from the fixed effects meta-analysis. Most of these SNPs are in the frequency and effect size classes expected for common diseases. Replication is underway.
Background and method:
Anorexia nervosa (AN) is a debilitating and potentially lethal mental illness influenced by both genetic and environmental factors (prevalence approximately 1%; heritability 22-88%). GCAN is a 17-country collaboration dedicated to identifying genes that influence AN. This is the largest GWAS for any eating disorder ever performed. As part of WTCCC3, we conducted a GWAS to identify associations between DNA sequence variation in 2907 AN cases and 14,860 geographically matched controls. Cases were female and met DSM IV criteria for AN (excluding amenorrhea). Genotyping was with the Illumina 660W-Quad array. Individual analyses were carried out across 15 strata. Eleven association analyses were conducted per stratum: one with no principal components (PC) adjustment and the others adjusted up to 10 PCs. On the basis of QQ plots, lambda values and 54 population stratification SNPs, we selected the best analysis per each stratum for meta-analysis.
Results:
The uncorrected lambda was 1.024. One SNP on chr5 exceeded genome-wide significance. 94 SNPs with p-value less than 10-4 derived from the fixed effects meta-analysis. Most of these SNPs are in the frequency and effect size classes expected for common diseases.
Future directions:
Replication:
SNPs for replication will be prioritized on the basis of directly genotyped data. Criteria for prioritization: independent signals will be followed-up to minimize redundancy; power to detect association for variants with lower frequencies and modest effect sizes will be taken into account; regional association plots will be created to examine loners; biological credentials and the number of studies/samples contributing to SNP association result will also be considered.
Secondary analyses:
Pre-planned secondary analyses will further explore CNVs, employ detailed phenotypic information, and apply novel methodologies.
2.)In terms of the GWAS study conducted with a normal population (St.Thomas twin data) the main results are as follows:
Genome-Wide Association Analysis of Eating Disorder-Related Symptoms, Behaviors, and Personality Traits
Boraska, V., Davis, O., Helder, S., Cherkas, L, F., Pei-Chi LIao, T., Harris, J., Krug, I., Treasure, J., Shin, S., Ntalla, I., Karhunen, L., Keski-Rahkonen, A., Raevuori, A., Kapiro, J., Dedoussis, G., Soranzo, N., Spector, T, D., Collier, D.A. Zeggini, E. (2012). American Journal of Medical Genetics Part B: Neuropsychiatric Genetics; 159B(7):803-11.
Abstract:
Eating disorders (EDs) are common, complex psychiatric disorders thought to be caused by both genetic and environmental factors. They share many symptoms, behaviors, and personality traits, which may have overlapping heritability. The aim of the present study is to perform a genome-wide association scan (GWAS) of six ED phenotypes comprising three symptom traits from the Eating Disorders Inventory 2 [drive for thinness (DT), body dissatisfaction (BD), and Bulimia], Weight Fluctuation symptom, Breakfast Skipping behavior and childhood obsessive-compulsive personality disorder trait (CHIRP). Investigated traits were derived from standardized self-report questionnaires completed by the TwinsUK population-based cohort. We tested 283,744 directly typed SNPs across six phenotypes of interest in the TwinsUK discovery dataset and followed-up signals from various strata using a two-stage replication strategy in two independent cohorts of European ancestry. We meta-analyzed a total of 2,698 individuals for DT, 2,680 for BD, 2,789 (821 cases/1,968 controls) for Bulimia, 1,360 (633 cases/727 controls) for Childhood Obsessive-Compulsive Personality Disorder trait, 2,773 (761 cases/2,012 controls) for Breakfast Skipping, and 2,967 (798 cases/2,169 controls) for Weight Fluctuation symptom. In this GWAS analysis of six ED-related phenotypes, we detected association of eight genetic variants with P less than 10-5. Genetic variants that showed suggestive evidence of association were previously associated with several psychiatric disorders and ED-related phenotypes. Our study indicates that larger-scale collaborative studies will be needed to achieve the necessary power to detect loci underlying ED-related traits.
Objective 2:
Our second objective was to:
a.) assess the interplay between genetic and environmental risk factors by assessing subtypes of EDs in relation to the genetic variants discovered in objective 1;
b.) to examine cross-cultural variation in the interplay between genetic and environmental risk factors across various European countries.
Objective 2 will be a continuation of the previously financed V EU Research Project- Healthy Eating Framework Project (Re. QLK1-1999-916). This was undertaken using a sister-pair design
-Three main studies were obtained from this sample. Due to space limitations only the main study will be detailed here. Please refer to the publication list for more details on the other studies
Dissecting childhood and adulthood obsessive-compulsive personality traits in eating disorders using a discordant sister-pair design: a multicenter European study
Krug I, Kanakam N, Anderluh M, Fernandez-Aranda F, Micali N, Taborelli E, Tchanturia K, Karwautz A, Wagner G, Collier D and Treasure J (2012).
Objective:
To refine and identify genetic, behavioral and cultural underpinnings of childhood and adulthood OCPD traits by performing integrated research across various ED subtypes. Method: In part A of the study 147 discordant sister pairs for EDs [Anorexia Nervosa-Restricting (AN-R)=37; AN-Binge Purging (AN-BP)=61; Bulimia Nervosa-Purging (BN-P)=42] were interviewed (using the EATATE) about their childhood and adulthood obsessive compulsive personality traits (OCPD). In part B the same OCPD traits were assessed in 320 ED patients (AN-R=89; AN-BP=136; BN=89) and in addition the EDI-2 and TCI-R were administered. DNA was also collected and four candidate genes (5-HT2A, BDNF, 5-HTTLPR, and DRD4) were genotyped. Whereas part A assessed differences between patients and sisters, part B compared ED patients with OCPD traits to ED individuals without these traits.
Results:
The ED patients scored significantly higher than their healthy sisters on most childhood and adulthood OCPD traits. As regards to ED subdiagnoses, AN-BP patients scored significantly higher than the AN-R and BN-P individuals. A longer duration of dieting and exercising, higher levels of body dissatisfaction and drive for thinness and a negative personality constellation were positively related to OCPD traits. Cultural variations were also observed with the UK displaying the highest values for all of the assessed traits. Finally no significant findings were obtained for the candidate genes assessed.
Discussion:
OCPD traits provide a robust characteristic of EDs, especially for AN-BP individuals. These traits along with certain ED symptoms and related maladaptive personality profiles may provide an improved understanding of the topography and etiology of EDs.
We believe that the present integrated projects advanced the state of the art by integrating genetic, environmental, social and clinical data coming from different countries. It is anticipated that the findings of our project will enable us to create an integrated model of the aetiology of disturbed eating behaviour by have a major impact on the contribution to standards in three areas: Diagnosis/Classification, Assessment/Treatment and Prevention. Our translational approach involves bringing basic scientific knowledge through to the application stage by linking endophenotypical variables to clinical symptoms and then to treatment development and outcome. Such new understanding will have the potential to highlight novel risk factors and endophenotypes for EDs, as well as to confirm the association of previously identified risk factors and endophenotypes that still have to be replicated or definitively confirmed. The proposed project therefore is a true multidisciplinary approach, in which a coordinated exchange of research findings between biologists, clinical researchers, healthcare providers, ethicists and social scientists has lead to a synergistic outcome. GWAS is the next logical step for the genetic study of AN.
The potential payoffs of this line of inquiry are many. The clear-cut identification of genomic variation that predisposes to AN would likely revolutionize the field by providing researchers and clinicians with a 'hard' finding upon which to base the next generation of research into aetiology, treatment, and prevention. Moreover, 'hard' findings on AN may help enhance understanding of related psychopathology (e.g. depression, anxiety disorders, obsessive compulsive disorder) and critical aspects of appetite and weight dysregulation. Moreover, this project involved highly innovative research designs (e.g. discordant sister pairs), aimed at creating an integrated aetiological model of the genes, environment and their interactions for EDs. Finally, we believe that the project has a powerful European dimension by integrating research into EDs across various countries to provide community added value.
This ensures that both national and European resources are used cost effectively, with a critical mass and synergy between centres and disciplines for optimal exploitation and implementation. It is anticipated that such cost effective solutions will increase European competiveness, strengthen the ERA, and will also benefit at a community level by decreasing the social cost and economic burden of weight-related disorders across Europe. The fellowship will strongly contribute to enhanced EU scientific excellence and reintegrate the researcher by providing a strong impact on various major stakeholders.
Our first objective of the project was:
a.) to perform genome wide association (GWAS) analysis, followed by replication analysis and advanced statistical analysis, to identify main genetic effects on the disease.
b.) to examine various endophenotypes through a process of data replication by assessing candidate genes.
Objective 1 was part of the WTCCC3 funded project 'A Genome wide Association Study of Anorexia Nervosa' in which genotyping and preliminary analysis is funded by the Welcome Trust (PI Prof Collier).
During the fellowship two GWAS studies were carried out, one assessing the clinical data collected from the GCAN project and the other one using the St.Thomas twin data looking at a general population.
1.) As regards to the GCAN (GWAS on anorexia) the following results have been achieved (Poster presented at the Eating Disorder Research Society in Edinburgh 2011):
WTCCC3 and GCAN: A Genomewide scan for anorexia nervosa
CM Bulik, DA Collier, E Zeggini, PF Sullivan, GCAN Consortium4, WTCCC .
Abstract:
The genetic consortium for anorexia nervosa (GCAN, part of WTCCC3) conducted a GWAS on 2907 cases with anorexia nervosa (AN) and 14,860 ancestrally matched controls. 94 SNPs with p-value less than 10-4 derived from the fixed effects meta-analysis. Most of these SNPs are in the frequency and effect size classes expected for common diseases. Replication is underway.
Background and method:
Anorexia nervosa (AN) is a debilitating and potentially lethal mental illness influenced by both genetic and environmental factors (prevalence approximately 1%; heritability 22-88%). GCAN is a 17-country collaboration dedicated to identifying genes that influence AN. This is the largest GWAS for any eating disorder ever performed. As part of WTCCC3, we conducted a GWAS to identify associations between DNA sequence variation in 2907 AN cases and 14,860 geographically matched controls. Cases were female and met DSM IV criteria for AN (excluding amenorrhea). Genotyping was with the Illumina 660W-Quad array. Individual analyses were carried out across 15 strata. Eleven association analyses were conducted per stratum: one with no principal components (PC) adjustment and the others adjusted up to 10 PCs. On the basis of QQ plots, lambda values and 54 population stratification SNPs, we selected the best analysis per each stratum for meta-analysis.
Results:
The uncorrected lambda was 1.024. One SNP on chr5 exceeded genome-wide significance. 94 SNPs with p-value less than 10-4 derived from the fixed effects meta-analysis. Most of these SNPs are in the frequency and effect size classes expected for common diseases.
Future directions:
Replication:
SNPs for replication will be prioritized on the basis of directly genotyped data. Criteria for prioritization: independent signals will be followed-up to minimize redundancy; power to detect association for variants with lower frequencies and modest effect sizes will be taken into account; regional association plots will be created to examine loners; biological credentials and the number of studies/samples contributing to SNP association result will also be considered.
Secondary analyses:
Pre-planned secondary analyses will further explore CNVs, employ detailed phenotypic information, and apply novel methodologies.
2.)In terms of the GWAS study conducted with a normal population (St.Thomas twin data) the main results are as follows:
Genome-Wide Association Analysis of Eating Disorder-Related Symptoms, Behaviors, and Personality Traits
Boraska, V., Davis, O., Helder, S., Cherkas, L, F., Pei-Chi LIao, T., Harris, J., Krug, I., Treasure, J., Shin, S., Ntalla, I., Karhunen, L., Keski-Rahkonen, A., Raevuori, A., Kapiro, J., Dedoussis, G., Soranzo, N., Spector, T, D., Collier, D.A. Zeggini, E. (2012). American Journal of Medical Genetics Part B: Neuropsychiatric Genetics; 159B(7):803-11.
Abstract:
Eating disorders (EDs) are common, complex psychiatric disorders thought to be caused by both genetic and environmental factors. They share many symptoms, behaviors, and personality traits, which may have overlapping heritability. The aim of the present study is to perform a genome-wide association scan (GWAS) of six ED phenotypes comprising three symptom traits from the Eating Disorders Inventory 2 [drive for thinness (DT), body dissatisfaction (BD), and Bulimia], Weight Fluctuation symptom, Breakfast Skipping behavior and childhood obsessive-compulsive personality disorder trait (CHIRP). Investigated traits were derived from standardized self-report questionnaires completed by the TwinsUK population-based cohort. We tested 283,744 directly typed SNPs across six phenotypes of interest in the TwinsUK discovery dataset and followed-up signals from various strata using a two-stage replication strategy in two independent cohorts of European ancestry. We meta-analyzed a total of 2,698 individuals for DT, 2,680 for BD, 2,789 (821 cases/1,968 controls) for Bulimia, 1,360 (633 cases/727 controls) for Childhood Obsessive-Compulsive Personality Disorder trait, 2,773 (761 cases/2,012 controls) for Breakfast Skipping, and 2,967 (798 cases/2,169 controls) for Weight Fluctuation symptom. In this GWAS analysis of six ED-related phenotypes, we detected association of eight genetic variants with P less than 10-5. Genetic variants that showed suggestive evidence of association were previously associated with several psychiatric disorders and ED-related phenotypes. Our study indicates that larger-scale collaborative studies will be needed to achieve the necessary power to detect loci underlying ED-related traits.
Objective 2:
Our second objective was to:
a.) assess the interplay between genetic and environmental risk factors by assessing subtypes of EDs in relation to the genetic variants discovered in objective 1;
b.) to examine cross-cultural variation in the interplay between genetic and environmental risk factors across various European countries.
Objective 2 will be a continuation of the previously financed V EU Research Project- Healthy Eating Framework Project (Re. QLK1-1999-916). This was undertaken using a sister-pair design
-Three main studies were obtained from this sample. Due to space limitations only the main study will be detailed here. Please refer to the publication list for more details on the other studies
Dissecting childhood and adulthood obsessive-compulsive personality traits in eating disorders using a discordant sister-pair design: a multicenter European study
Krug I, Kanakam N, Anderluh M, Fernandez-Aranda F, Micali N, Taborelli E, Tchanturia K, Karwautz A, Wagner G, Collier D and Treasure J (2012).
Objective:
To refine and identify genetic, behavioral and cultural underpinnings of childhood and adulthood OCPD traits by performing integrated research across various ED subtypes. Method: In part A of the study 147 discordant sister pairs for EDs [Anorexia Nervosa-Restricting (AN-R)=37; AN-Binge Purging (AN-BP)=61; Bulimia Nervosa-Purging (BN-P)=42] were interviewed (using the EATATE) about their childhood and adulthood obsessive compulsive personality traits (OCPD). In part B the same OCPD traits were assessed in 320 ED patients (AN-R=89; AN-BP=136; BN=89) and in addition the EDI-2 and TCI-R were administered. DNA was also collected and four candidate genes (5-HT2A, BDNF, 5-HTTLPR, and DRD4) were genotyped. Whereas part A assessed differences between patients and sisters, part B compared ED patients with OCPD traits to ED individuals without these traits.
Results:
The ED patients scored significantly higher than their healthy sisters on most childhood and adulthood OCPD traits. As regards to ED subdiagnoses, AN-BP patients scored significantly higher than the AN-R and BN-P individuals. A longer duration of dieting and exercising, higher levels of body dissatisfaction and drive for thinness and a negative personality constellation were positively related to OCPD traits. Cultural variations were also observed with the UK displaying the highest values for all of the assessed traits. Finally no significant findings were obtained for the candidate genes assessed.
Discussion:
OCPD traits provide a robust characteristic of EDs, especially for AN-BP individuals. These traits along with certain ED symptoms and related maladaptive personality profiles may provide an improved understanding of the topography and etiology of EDs.
We believe that the present integrated projects advanced the state of the art by integrating genetic, environmental, social and clinical data coming from different countries. It is anticipated that the findings of our project will enable us to create an integrated model of the aetiology of disturbed eating behaviour by have a major impact on the contribution to standards in three areas: Diagnosis/Classification, Assessment/Treatment and Prevention. Our translational approach involves bringing basic scientific knowledge through to the application stage by linking endophenotypical variables to clinical symptoms and then to treatment development and outcome. Such new understanding will have the potential to highlight novel risk factors and endophenotypes for EDs, as well as to confirm the association of previously identified risk factors and endophenotypes that still have to be replicated or definitively confirmed. The proposed project therefore is a true multidisciplinary approach, in which a coordinated exchange of research findings between biologists, clinical researchers, healthcare providers, ethicists and social scientists has lead to a synergistic outcome. GWAS is the next logical step for the genetic study of AN.
The potential payoffs of this line of inquiry are many. The clear-cut identification of genomic variation that predisposes to AN would likely revolutionize the field by providing researchers and clinicians with a 'hard' finding upon which to base the next generation of research into aetiology, treatment, and prevention. Moreover, 'hard' findings on AN may help enhance understanding of related psychopathology (e.g. depression, anxiety disorders, obsessive compulsive disorder) and critical aspects of appetite and weight dysregulation. Moreover, this project involved highly innovative research designs (e.g. discordant sister pairs), aimed at creating an integrated aetiological model of the genes, environment and their interactions for EDs. Finally, we believe that the project has a powerful European dimension by integrating research into EDs across various countries to provide community added value.
This ensures that both national and European resources are used cost effectively, with a critical mass and synergy between centres and disciplines for optimal exploitation and implementation. It is anticipated that such cost effective solutions will increase European competiveness, strengthen the ERA, and will also benefit at a community level by decreasing the social cost and economic burden of weight-related disorders across Europe. The fellowship will strongly contribute to enhanced EU scientific excellence and reintegrate the researcher by providing a strong impact on various major stakeholders.