The multifactorial nature of food allergy
van Ginkel, Cornelia Doriene
IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below.
Document Version
Publisher's PDF, also known as Version of record
Publication date: 2018
Link to publication in University of Groningen/UMCG research database
Citation for published version (APA):
van Ginkel, C. D. (2018). The multifactorial nature of food allergy. Rijksuniversiteit Groningen.
Copyright
Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).
Take-down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.
Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.
alternative 3’UTRs. Trends Cell Biol. 2016;26:227-237.
22. Potaczek DP, Harb H, Michel S, Alhamwe BA, Renz H, Tost J. Epige-netics and allergy: from basic mechanisms to clinical applications. Epigenomics. 2017;9:539-571.
23. Moffatt MF, Phil D, Gut IG, et al. A large-scale, consortium-based genomewide association study of asthma. N Engl J Med. 2010;363:1211-1221.
24. Paternoster L, Standl M, Chen C-M, et al. Meta-analysis of genome-wide association studies identifies three new risk loci for atopic der-matitis. Nat Genet. 2011;44:187-192. 25. Qian X, Gao Y, Ye X, Lu M. Association of STAT6
variants with asthma risk : a systematic review
and meta-analysis. Hum Immunol.
2014;75:847-853.
26. Tamura K, Suzuki M, Arakawa H, Tokuyama K, Morikawa A. Linkage and association studies of STAT6 gene polymorphisms and allergic diseases. Int Arch Allergy Immunol. 2003;131:33-38.
exposure, CD14, and allergic disease: an interaction between genes and the environment. Am J Respir Crit Care Med. 2006;174:386-392.
28. Wainstein BK, Saad RA. Repeat oral food challenges in peanut and tree nut allergic children with a history of mild/moderate reactions. Asia Pac Allergy. 2015;5:170. 29. Turner PJ, Baumert JL, Beyer K, et al. Can we
identify patients at risk of life-threatening allergic reactions to food? Allergy Eur J Allergy Clin Immunol. 2016;71:1241-1255.
30. Sleiman PMA, Wang M-L, Cianferoni A, et al. GWAS identifies four novel eosinophilic esophagitis loci. Nat Commun. 2014;5:5593. 31. Levin AM, Mathias RA, Huang L, et al. A
meta-analysis of genome-wide association studies for serum total IgE in diverse study popula-tions. J Allergy Clin Immunol. 2013;131:1176-1184.
32. Weidinger S, Gieger C, Rodriguez E, et al. Genome-wide scan on total serum IgE levels identifies FCER1A as novel susceptibility locus. PLoS Genet. 2008;4:e1000166.
GENOME WIDE ASSOCIATION STUDY OF FOOD
PEANUT ALLERGY
C.D. VAN GINKEL1 ,2, A.E.J. DUBOIS1 ,2, A.B. SPRIKKELMAN1 ,2, CHENG-JIAN XU1 ,2,
F.N. DIJK1 ,2, I.M. NOLTE3, U. GEHRING4, I. MARENHOLZ5 ,6, Y.-A. LEE5 ,6, J.M.
VONK2 ,3, G.H. KOPPELMAN1 ,2
1 University of Groningen, University Medical Center Groningen, Department of Paediatric
Pulmonology and Paediatric Allergy, Groningen, the Netherlands; 2 University of Groningen,
University Medical Center Groningen, GRIAC Research Institute, Groningen, the Netherlands;
3 University of Groningen, University Medical Center Groningen, Department of
Epidemiology, Groningen, the Netherlands; 4 Institute for Risk Assessment Sciences (IRAS),
Utrecht University, Utrecht, the Netherlands; 5 Max-Delbrück-Center (MDC) for Molecular
Medicine, Berlin, Germany; 6 Clinic for Pediatric Allergy, Experimental and Clinical Research
Center, Charité University Medical Center, Berlin, Germany
140
ABSTRACT
INTRODUCTIONFood allergy (FA) is partly caused by genetic factors. We aimed to identify single nucleotide polymorphisms (SNPs) associated with FA and Peanut allergy (PA) and to replicate previously published SNPs associated with (food) allergy.
METHODS
Associations of SNPs with FA and PA were identified by meta-analysis of genome wide genetic data from two Dutch cohorts (Lifelines and PIAMA) including adolescents and adults in which
FA and PA were assessed by a detailed questionnaire. Top hits (p-value <1*10-6) were
replicated in two additional cohorts including children phenotyped by food challenges. Previously identified (food) allergy SNPs were investigated in the discovery dataset.
RESULTS
In 392 FA cases, 47 PA cases and 9470 controls, 6 and 25 independent SNPs were associated with FA and PA, respectively, including SNPs in introns of RBFOX1 and JAZF1, genes previously associated with (food) allergy. None of these replicated in the replication cohorts (859 FA cases including 336 PA cases, 3358 controls and 283 FA case-parent trios including 183 PA case-parent trios). In addition, we replicated the associations of 12/31 and 10/121 SNPs previously associated with FA and any allergic disease, respectively.
DISCUSSION
In this GWAS on FA in European adults and adolescents, we replicated the association of 22 SNPs previously associated with (food) allergy, thereby obtaining a more robust SNP set for FA. In addition, we report associations of SNPs with FA and PA in the general population, which did not replicate in 2 independent clinical cohorts of children. Our results could indicate age
specific effects on food allergy, which warrant further study.
ABBREVIATIONS
CI - confidence interval
DBPCFC - double-blind placebo-controlled food challenge
eQTL - expression quantitative trait loci
FA - food allergy
FAQ - food allergy questionnaire
GENEVA - Dutch study on genetics of food allergy (genetica van
voedselallergie)
GOFA - German study on genetics of food Allergy
GWAS - genome-wide association study
LikelyFA - likely food allergy
LikelyPA - likely peanut allergy
MAF - minor allele frequency
NoFA - not having food allergy
OR - odds ratio
PIAMA - Prevention and Incidence of Asthma and Mite Allergy
PA - peanut allergy
Chap
ter 9
141
ABSTRACT
INTRODUCTION
Food allergy (FA) is partly caused by genetic factors. We aimed to identify single nucleotide polymorphisms (SNPs) associated with FA and Peanut allergy (PA) and to replicate previously published SNPs associated with (food) allergy.
METHODS
Associations of SNPs with FA and PA were identified by meta-analysis of genome wide genetic data from two Dutch cohorts (Lifelines and PIAMA) including adolescents and adults in which
FA and PA were assessed by a detailed questionnaire. Top hits (p-value <1*10-6) were
replicated in two additional cohorts including children phenotyped by food challenges. Previously identified (food) allergy SNPs were investigated in the discovery dataset.
RESULTS
In 392 FA cases, 47 PA cases and 9470 controls, 6 and 25 independent SNPs were associated with FA and PA, respectively, including SNPs in introns of RBFOX1 and JAZF1, genes previously associated with (food) allergy. None of these replicated in the replication cohorts (859 FA cases including 336 PA cases, 3358 controls and 283 FA case-parent trios including 183 PA case-parent trios). In addition, we replicated the associations of 12/31 and 10/121 SNPs previously associated with FA and any allergic disease, respectively.
DISCUSSION
In this GWAS on FA in European adults and adolescents, we replicated the association of 22 SNPs previously associated with (food) allergy, thereby obtaining a more robust SNP set for FA. In addition, we report associations of SNPs with FA and PA in the general population, which did not replicate in 2 independent clinical cohorts of children. Our results could indicate age
specific effects on food allergy, which warrant further study.
ABBREVIATIONS
CI - confidence interval
DBPCFC - double-blind placebo-controlled food challenge
eQTL - expression quantitative trait loci
FA - food allergy
FAQ - food allergy questionnaire
GENEVA - Dutch study on genetics of food allergy (genetica van
voedselallergie)
GOFA - German study on genetics of food Allergy
GWAS - genome-wide association study
LikelyFA - likely food allergy
LikelyPA - likely peanut allergy
MAF - minor allele frequency
NoFA - not having food allergy
OR - odds ratio
PIAMA - Prevention and Incidence of Asthma and Mite Allergy
PA - peanut allergy
142
INTRODUCTION
Food allergy (FA) significantly impairs quality of life and currently, there is no treatment for FA other than avoidance of the culprit food and treatment of a potentially lethal anaphylactic
shock1. FA, being an expression of atopy, is significantly more often present in patients with
other allergic diseases such as asthma, allergic rhinitis and eczema2–4, indicating shared origins
of allergic disease. The pathogenesis of FA remains poorly understood but is likely to be caused
by a combination of environmental and genetic factors5. One study reported a 5-fold
increased risk of peanut allergy (PA) for children with a sibling or parent with PA6. A twin study
showed a likelihood of 64% for PA for a child with a monozygotic twin with PA, compared to
6.8% for dizygotic twins7. Both studies indicate PA is at least partly heritable. Identifying
genetic risk factors for FA may allow targeting of preventive measures in a population at risk and may provide insight into novel targets of therapeutic interventions.
To date, different SNPs have been associated with FA and PA in genome-wide
association studies (GWAS). Four out of five GWAS focussed on FA, specifically PA, in children8–
11. In these studies, PA was repeatedly and specifically associated with single nucleotide
polymorphisms (SNPs) in the HLA-DQB1 gene8,9,11,12. The Human Leukocyte Antigen (HLA) class
II genes encode molecules which present antigens, such as food allergens, to naïve CD4+T
lymphocytes. These stimulate the allergic response by driving B-cell antibody production. Recently, a GWAS in FA in Japanese women identified rs28359884 and rs74995702 as associated with peach and shrimp allergy, respectively, which are also in LD with single nucleotide polymorphisms (SNPs) in HLA-DQB1. One of the GWAS in children studied any FA, hen’s egg and cow’s milk allergy next to PA. They identified 4 susceptibility loci associated with
FA in general, 1 with PA and 1 with hen’s egg allergy11. A SNP in the SERPINB gene cluster on
chromosome 18 was associated with FA in general. The SERPINB7 gene is highly expressed in
the epidermis and is important for the integrity of the skin barrier11. Together with the widely
replicated association of the FLG gene with FA, this indicates that an impaired skin barrier may
increase the risk of any FA11,13–15. This suggests that there are both food-specific and more
general genetic pathways involved in the etiology of food allergies. In addition, the SHARE consortium recently published a list of 136 SNPs to be significantly associated with any allergic disease (astma, rhinitis, eczema), yet FA was not addressed in that study and it remains
unknown if these SNPs also relate to FA16.
Compared to other allergic diseases, the genetics of FA has only been studied in a few cohorts. Since FA is likely be caused by many SNPs with small effects, large populations are needed. However, it is difficult to study FA in a large cohort study since defining the phenotype
is challenging, and self-reported phenotypes most likely overestimate the FA prevalence2,17.
Therefore, most studies used food challenges with limited numbers of included subjects. We therefore recently developed the Food Allergy Questionnaire (FAQ) aiming to improve the specificity of the assessment of FA using questionnaires, to be able to study FA genetics in
larger populations3. Using the FAQ, we defined likely food allergy (LikelyFA) based on the
report of a likely culprit allergenic food and a description of the allergic reaction consistent
with an IgE mediated allergic reaction3. We hypothesize that food allergy in the general
population is caused by the same genetic variants as more severe food allergy, as seen in the clinic and diagnosed by food challenges.
This study aimed to identify genes associated with questionnaire-based LikelyFA and likely peanut allergy (LikelyPA) in a meta-analysis of two large Dutch cohorts (Lifelines and PIAMA). Top hits are replicated in two independent cohorts (GENEVA and GOFA) that assessed FA by food challenges. In addition, we aimed to replicate SNPs previously identified with FA and allergic diseases to verify their role in FA in adolescents and adults and to identify a robust SNP set for FA.
METHODS
Discovery analyses were performed in two cohorts, the Lifelines Cohort Study and Prevention
and Incidence of Asthma and Mite Allergy (PIAMA) study 18,19. Subjects within both cohorts
were phenotyped as previously described3. In summary, subjects completed the FAQ (see
online supplement) and were classified as not having food allergy (NoFA), indeterminate or likely having food allergy (LikelyFA) based on symptoms, suspected food, and reported diagnostic procedures. In the classification, specificity was preferred over sensitivity to minimize the number of false-positives in the LikelyFA group. For this analysis, subjects classified as LikelyFA were cases and compared to the control group classified as NoFA. Subjects classified as Indeterminate were excluded from our study. We subsequently replicated top hits in two replication populations (GENEVA and GOFA) that assessed FA by
(placebo controlled) food challenges11,15,20. All cohorts are briefly described below, a full
description including genotyping and quality control is provided in the online supplement. DISCOVERY COHORTS: LIFELINES COHORT STUDY AND PIAMA
Lifelines is a multi-disciplinary prospective population-based cohort study examining the health and health-related behaviors of 167,729 persons living in the north of the Netherlands in a unique three-generation design. It employs a broad range of investigative procedures in assessing the biomedical, socio-demographic, behavioral, physical and psychological factors contributing to the health and disease of the general population, with a special focus on
multi-morbidity and complex genetics19,21. The study protocol was approved by the medical ethics
committee of the University Medical Center Groningen. Recruitment of participants took place between 2006 and 2013 and from September 2014 onwards, all subjects were invited to complete a second examination, which included the FAQ. All genotyped adults who completed this before 01-01-2017 were included in this study (supplementary figure 1). Genotyping was performed using the Illumina CytoSNP-12v2 array and samples were imputed
using Minimac version 2012.10.322 against the reference data set of the Genome of The
Netherlands release 523. The quality control is described in the supporting data. Genotyped
SNPs with a minor allele frequency (MAF) above 0.001 and imputed SNPs with an info score >0.3 were included in the meta-analysis.
The PIAMA study protocol has been described elsewhere18. Mothers were recruited
Chap
ter 9
143
INTRODUCTION
Food allergy (FA) significantly impairs quality of life and currently, there is no treatment for FA other than avoidance of the culprit food and treatment of a potentially lethal anaphylactic
shock1. FA, being an expression of atopy, is significantly more often present in patients with
other allergic diseases such as asthma, allergic rhinitis and eczema2–4, indicating shared origins
of allergic disease. The pathogenesis of FA remains poorly understood but is likely to be caused
by a combination of environmental and genetic factors5. One study reported a 5-fold
increased risk of peanut allergy (PA) for children with a sibling or parent with PA6. A twin study
showed a likelihood of 64% for PA for a child with a monozygotic twin with PA, compared to
6.8% for dizygotic twins7. Both studies indicate PA is at least partly heritable. Identifying
genetic risk factors for FA may allow targeting of preventive measures in a population at risk and may provide insight into novel targets of therapeutic interventions.
To date, different SNPs have been associated with FA and PA in genome-wide
association studies (GWAS). Four out of five GWAS focussed on FA, specifically PA, in children8–
11. In these studies, PA was repeatedly and specifically associated with single nucleotide
polymorphisms (SNPs) in the HLA-DQB1 gene8,9,11,12. The Human Leukocyte Antigen (HLA) class
II genes encode molecules which present antigens, such as food allergens, to naïve CD4+T
lymphocytes. These stimulate the allergic response by driving B-cell antibody production. Recently, a GWAS in FA in Japanese women identified rs28359884 and rs74995702 as associated with peach and shrimp allergy, respectively, which are also in LD with single nucleotide polymorphisms (SNPs) in HLA-DQB1. One of the GWAS in children studied any FA, hen’s egg and cow’s milk allergy next to PA. They identified 4 susceptibility loci associated with
FA in general, 1 with PA and 1 with hen’s egg allergy11. A SNP in the SERPINB gene cluster on
chromosome 18 was associated with FA in general. The SERPINB7 gene is highly expressed in
the epidermis and is important for the integrity of the skin barrier11. Together with the widely
replicated association of the FLG gene with FA, this indicates that an impaired skin barrier may
increase the risk of any FA11,13–15. This suggests that there are both food-specific and more
general genetic pathways involved in the etiology of food allergies. In addition, the SHARE consortium recently published a list of 136 SNPs to be significantly associated with any allergic disease (astma, rhinitis, eczema), yet FA was not addressed in that study and it remains
unknown if these SNPs also relate to FA16.
Compared to other allergic diseases, the genetics of FA has only been studied in a few cohorts. Since FA is likely be caused by many SNPs with small effects, large populations are needed. However, it is difficult to study FA in a large cohort study since defining the phenotype
is challenging, and self-reported phenotypes most likely overestimate the FA prevalence2,17.
Therefore, most studies used food challenges with limited numbers of included subjects. We therefore recently developed the Food Allergy Questionnaire (FAQ) aiming to improve the specificity of the assessment of FA using questionnaires, to be able to study FA genetics in
larger populations3. Using the FAQ, we defined likely food allergy (LikelyFA) based on the
report of a likely culprit allergenic food and a description of the allergic reaction consistent
with an IgE mediated allergic reaction3. We hypothesize that food allergy in the general
population is caused by the same genetic variants as more severe food allergy, as seen in the clinic and diagnosed by food challenges.
This study aimed to identify genes associated with questionnaire-based LikelyFA and likely peanut allergy (LikelyPA) in a meta-analysis of two large Dutch cohorts (Lifelines and PIAMA). Top hits are replicated in two independent cohorts (GENEVA and GOFA) that assessed FA by food challenges. In addition, we aimed to replicate SNPs previously identified with FA and allergic diseases to verify their role in FA in adolescents and adults and to identify a robust SNP set for FA.
METHODS
Discovery analyses were performed in two cohorts, the Lifelines Cohort Study and Prevention
and Incidence of Asthma and Mite Allergy (PIAMA) study 18,19. Subjects within both cohorts
were phenotyped as previously described3. In summary, subjects completed the FAQ (see
online supplement) and were classified as not having food allergy (NoFA), indeterminate or likely having food allergy (LikelyFA) based on symptoms, suspected food, and reported diagnostic procedures. In the classification, specificity was preferred over sensitivity to minimize the number of false-positives in the LikelyFA group. For this analysis, subjects classified as LikelyFA were cases and compared to the control group classified as NoFA. Subjects classified as Indeterminate were excluded from our study. We subsequently replicated top hits in two replication populations (GENEVA and GOFA) that assessed FA by
(placebo controlled) food challenges11,15,20. All cohorts are briefly described below, a full
description including genotyping and quality control is provided in the online supplement. DISCOVERY COHORTS: LIFELINES COHORT STUDY AND PIAMA
Lifelines is a multi-disciplinary prospective population-based cohort study examining the health and health-related behaviors of 167,729 persons living in the north of the Netherlands in a unique three-generation design. It employs a broad range of investigative procedures in assessing the biomedical, socio-demographic, behavioral, physical and psychological factors contributing to the health and disease of the general population, with a special focus on
multi-morbidity and complex genetics19,21. The study protocol was approved by the medical ethics
committee of the University Medical Center Groningen. Recruitment of participants took place between 2006 and 2013 and from September 2014 onwards, all subjects were invited to complete a second examination, which included the FAQ. All genotyped adults who completed this before 01-01-2017 were included in this study (supplementary figure 1). Genotyping was performed using the Illumina CytoSNP-12v2 array and samples were imputed
using Minimac version 2012.10.322 against the reference data set of the Genome of The
Netherlands release 523. The quality control is described in the supporting data. Genotyped
SNPs with a minor allele frequency (MAF) above 0.001 and imputed SNPs with an info score >0.3 were included in the meta-analysis.
The PIAMA study protocol has been described elsewhere18. Mothers were recruited
144
Netherlands. In short, this cohort includes 3 963 children born in 1996-1997, with follow-up visits and questionnaires from birth onwards. The study protocol was approved by the medical ethics committees of the participating institutions and all parents gave written informed consent. All genotyped adolescents who completed the FAQ at the age of 17 years were included in this study (supplementary figure 2). Genotyping was performed using the Illumina Omni Express Exome Chip (n=1377), Illumina Omni Express chip (n=288) and the Illumina Human610 quad array (n=404, Illumina Inc, San Diego, CA). The quality control is described in the supporting data. Samples were imputed using IMPUTE 2.0 against the reference data set
of the ALL panel of 1000G (version 3, March 2012)22,24. Genotyped SNPs with a MAF above
0.05 and an imputed SNPs with an info score >0.7 were included in the meta-analysis. REPLICATION COHORTS: GENEVA AND GOFA
The GENEtics of Food Allergy (GENEVA) cohort includes 491 Dutch trios (parents and child) recruited from 2005 onwards. These children had a double-blind placebo-controlled food challenge (DBPCFC), the gold standard for FA, as part of regular tertiary pediatric allergy care because of a history consistent with an IgE-mediated reaction after ingestion of a food. A
subgroup of the GENEVA cohort was described previously10,12,15,20. The study ethics were
approved (METc 2004-146) and written informed (parental) consent was obtained. All DBPCFCs with positive results were included. The DBPCFCs were performed as previously
described25. After quality control (see supporting data), 283 FA case trios and 183 PA case trios
were studied (supplementary figure 3).
Cases from the Genetics Of Food Allergy (GOFA) study were recruited at four clinical centers in Germany and phenotyped by food challenges, most of which were conducted in a double-blind setting. Controls were derived from two population-based cohorts of blood donors, the Heinz Nixdorf RECALL (HNR) (n=2682) and the Study of Health in Pomerania (SHIP) cohort(n=986). There were no data available regarding age, food allergy, asthma, eczema and nasal allergy in the controls. The study was approved by the ethics committees of the local institutions. After quality control (see supporting data), 859 FA cases including 336 PA cases and 3358 controls were analyzed.
STATISTICAL ANALYSIS
Logistic regression was performed to analyze associations between SNPs (genotyped and
imputed dosages) and LikelyFA (as compared to NoFA) using PLINK (V1.07)26. In a second
analysis on LikelyPA, associations were studied in subjects classified as LikelyFA who reported PA (compared to NoFA). To study the effect of potential confounding by allergic disorders, three different models were performed adjusted for; 1) age and sex, 2) age, sex and eczema and 3) age, sex, eczema, asthma and nasal allergy/hayfever. Summary statistics of both
cohorts were meta-analyzed in a fixed-effects model using METAL (v2011-03-25)27.
Associations with LikelyFA or LikelyPA were considered to be of genome wide significance
when they had a p-value<5x10-8. This threshold is based on multiple testing correction for
approximately 1 million independent common SNPs for Europeans28.The arbitrary cut-off of
a p-value below 1x10-6 was considered the threshold for suggestive evidence for association.
In the replication analysis in GOFA, associations were studied using a logistic regression model with sex as covariate. In GENEVA, associations were tested by the family-based association
test (FBAT 2.0.4 using the additive model29) which is robust against population stratification
and tests for Mendelian errors. The FBAT is based on the transmission disequilibrium test, which compares the alleles transmitted to affected offspring with the expected distribution of
alleles among offspring29. All SNPs below the suggestive threshold from the discovery analysis
were included in the replication phase and meta-analyzed in all four cohorts using METAL. IDENTIFICATION OF SENTINEL SNPS
Sentinel SNPs with the same direction in both the Lifelines and PIAMA GWAS were identified based on the lowest p-value in the discovery analyses (with age and sex as covariates) within
a region of 500 kb. When there were multiple SNPs within the region with a p<1*10-6, we
confirmed that there were no other statistically independent associations using GCTA-CoJo30.
In these analyses, linkage disequilibrium (LD) was based on the large Lifelines dataset. Briefly, for each locus, we identified a top SNP and adjusted the summary statistics of all SNPs in that locus by the effect of the top SNP. We then checked whether there were any other
significantly associated SNPs that remained below the threshold of p<1*10-6.
REPLICATION OF PREVIOUS RESULTS: FOOD ALLERGY AND ALLERGIC DISEASE
Since these cohorts include adults and adolescents with LikelyFA and LikelyPA, we aimed to replicate previous associations with FA as identified in previous GWAS studies. We extracted
all SNPs with a p<1*10-6 from the 5 previous GWAS studies on FA (including studies on PA)
and studied these in our discovery analysis in the Lifelines and PIAMA cohorts8,9,11,31,32. In
addition, we replicated the 136 independent SNPs that showed a significant genome wide association with any allergic disease (including asthma, hayfever and eczema), since FA was
not included in this largest GWAS to date among 360,838 subjects16.
FOLLOW UP
To understand the biological consequences of SNPs, functional consequences and LD patterns
were checked in the European population in LDlink33 (r2 threshold 0.8). Expression
quantitative trait loci (eQTL) characteristics were studied by
genenetwork.nl/bloodeqtlbrowser and the online Haploreg and GTEx project databases34–36.
Gene expression and genomic annotation was determined by Haploreg and GTEx as well.
Information on genes was obtained in the Genecards database37 (queried in March 2018).
RESULTS
From the Lifelines and PIAMA cohorts, 8 724 and 1 138 subjects, respectively, met the inclusion criteria and quality control standards, see supplementary figure 1 and 2. As indicated in table 1, 3.8% (Lifelines) and 5.3% (PIAMA) of the subjects were classified as LikelyFA and 0.4% (Lifelines) and 1.2% (PIAMA) were classified as LikelyPA. The mean age of Lifelines cases and controls was 52.3 and 53.6 years, whereas the mean age of the PIAMA cases and controls
Chap
ter 9
145 Netherlands. In short, this cohort includes 3 963 children born in 1996-1997, with follow-up
visits and questionnaires from birth onwards. The study protocol was approved by the medical ethics committees of the participating institutions and all parents gave written informed consent. All genotyped adolescents who completed the FAQ at the age of 17 years were included in this study (supplementary figure 2). Genotyping was performed using the Illumina Omni Express Exome Chip (n=1377), Illumina Omni Express chip (n=288) and the Illumina Human610 quad array (n=404, Illumina Inc, San Diego, CA). The quality control is described in the supporting data. Samples were imputed using IMPUTE 2.0 against the reference data set
of the ALL panel of 1000G (version 3, March 2012)22,24. Genotyped SNPs with a MAF above
0.05 and an imputed SNPs with an info score >0.7 were included in the meta-analysis. REPLICATION COHORTS: GENEVA AND GOFA
The GENEtics of Food Allergy (GENEVA) cohort includes 491 Dutch trios (parents and child) recruited from 2005 onwards. These children had a double-blind placebo-controlled food challenge (DBPCFC), the gold standard for FA, as part of regular tertiary pediatric allergy care because of a history consistent with an IgE-mediated reaction after ingestion of a food. A
subgroup of the GENEVA cohort was described previously10,12,15,20. The study ethics were
approved (METc 2004-146) and written informed (parental) consent was obtained. All DBPCFCs with positive results were included. The DBPCFCs were performed as previously
described25. After quality control (see supporting data), 283 FA case trios and 183 PA case trios
were studied (supplementary figure 3).
Cases from the Genetics Of Food Allergy (GOFA) study were recruited at four clinical centers in Germany and phenotyped by food challenges, most of which were conducted in a double-blind setting. Controls were derived from two population-based cohorts of blood donors, the Heinz Nixdorf RECALL (HNR) (n=2682) and the Study of Health in Pomerania (SHIP) cohort(n=986). There were no data available regarding age, food allergy, asthma, eczema and nasal allergy in the controls. The study was approved by the ethics committees of the local institutions. After quality control (see supporting data), 859 FA cases including 336 PA cases and 3358 controls were analyzed.
STATISTICAL ANALYSIS
Logistic regression was performed to analyze associations between SNPs (genotyped and
imputed dosages) and LikelyFA (as compared to NoFA) using PLINK (V1.07)26. In a second
analysis on LikelyPA, associations were studied in subjects classified as LikelyFA who reported PA (compared to NoFA). To study the effect of potential confounding by allergic disorders, three different models were performed adjusted for; 1) age and sex, 2) age, sex and eczema and 3) age, sex, eczema, asthma and nasal allergy/hayfever. Summary statistics of both
cohorts were meta-analyzed in a fixed-effects model using METAL (v2011-03-25)27.
Associations with LikelyFA or LikelyPA were considered to be of genome wide significance
when they had a p-value<5x10-8. This threshold is based on multiple testing correction for
approximately 1 million independent common SNPs for Europeans28.The arbitrary cut-off of
a p-value below 1x10-6 was considered the threshold for suggestive evidence for association.
In the replication analysis in GOFA, associations were studied using a logistic regression model with sex as covariate. In GENEVA, associations were tested by the family-based association
test (FBAT 2.0.4 using the additive model29) which is robust against population stratification
and tests for Mendelian errors. The FBAT is based on the transmission disequilibrium test, which compares the alleles transmitted to affected offspring with the expected distribution of
alleles among offspring29. All SNPs below the suggestive threshold from the discovery analysis
were included in the replication phase and meta-analyzed in all four cohorts using METAL. IDENTIFICATION OF SENTINEL SNPS
Sentinel SNPs with the same direction in both the Lifelines and PIAMA GWAS were identified based on the lowest p-value in the discovery analyses (with age and sex as covariates) within
a region of 500 kb. When there were multiple SNPs within the region with a p<1*10-6, we
confirmed that there were no other statistically independent associations using GCTA-CoJo30.
In these analyses, linkage disequilibrium (LD) was based on the large Lifelines dataset. Briefly, for each locus, we identified a top SNP and adjusted the summary statistics of all SNPs in that locus by the effect of the top SNP. We then checked whether there were any other
significantly associated SNPs that remained below the threshold of p<1*10-6.
REPLICATION OF PREVIOUS RESULTS: FOOD ALLERGY AND ALLERGIC DISEASE
Since these cohorts include adults and adolescents with LikelyFA and LikelyPA, we aimed to replicate previous associations with FA as identified in previous GWAS studies. We extracted
all SNPs with a p<1*10-6 from the 5 previous GWAS studies on FA (including studies on PA)
and studied these in our discovery analysis in the Lifelines and PIAMA cohorts8,9,11,31,32. In
addition, we replicated the 136 independent SNPs that showed a significant genome wide association with any allergic disease (including asthma, hayfever and eczema), since FA was
not included in this largest GWAS to date among 360,838 subjects16.
FOLLOW UP
To understand the biological consequences of SNPs, functional consequences and LD patterns
were checked in the European population in LDlink33 (r2 threshold 0.8). Expression
quantitative trait loci (eQTL) characteristics were studied by
genenetwork.nl/bloodeqtlbrowser and the online Haploreg and GTEx project databases34–36.
Gene expression and genomic annotation was determined by Haploreg and GTEx as well.
Information on genes was obtained in the Genecards database37 (queried in March 2018).
RESULTS
From the Lifelines and PIAMA cohorts, 8 724 and 1 138 subjects, respectively, met the inclusion criteria and quality control standards, see supplementary figure 1 and 2. As indicated in table 1, 3.8% (Lifelines) and 5.3% (PIAMA) of the subjects were classified as LikelyFA and 0.4% (Lifelines) and 1.2% (PIAMA) were classified as LikelyPA. The mean age of Lifelines cases and controls was 52.3 and 53.6 years, whereas the mean age of the PIAMA cases and controls
146
was 17.2 and 17.3 years. Being female was associated with a higher risk of LikelyFA in the Lifelines, but not in the PIAMA cohort. Cases classified as LikelyFA reported more eczema, asthma and nasal allergy in both cohorts, compared to subjects classified as NoFA. Subjects in the replication cohorts are described in supplementary table 3. In total, 7 459 872 and 6 657 365 SNPs were present in the PIAMA and Lifelines cohort and subjected to meta-analysis for
LikelyFA and LikelyPA, respectively. There was no evidence of inflation of the test statistics of
this meta-analysis (LikelyFA: l=0.99, LikelyPA: l=0.99). Within the discovery analyses, we
identified 6 and 25 independent top hits associated with LikelyFA and LikelyPA, respectively. All were imputed SNPs. CGTA-CoJo analyses did not reveal any additional independent associated SNPs within 500 kb.
FOOD ALLERGY
In the discovery analyses for LikelyFA, six independent SNPs were identified with suggestive
evidence for association (p<1*10-6, see figure 1). None of the SNPs passed the significance
threshold for genome-wide significance (p<5x10-8). Two were located in the 3’ untranslated
region of RPP30 and an intron of RBFOX1, genes previously associated with eosinophilia and
FA, respectively37,38. In general, the effect sizes remained stable after adjusting for allergic
comorbidities. None of the SNPs was significantly replicated or reached the threshold suggestive for association with FA in the meta-analysis of all four cohorts (see table 2). PEANUT ALLERGY
In the discovery analyses for LikelyPA, 25 independent SNPs were identified with suggestive
evidence for association (p< 1*10-6) of which four reached the threshold for genome wide
significance (p<5*10-8, see figure 2). Of the latter, two were located in introns of MYLK3 and
CMTM8, previously associated with CCR3 signaling in eosinophils and IgG glycosylation,
respectively37. In addition, two SNPs were located in introns of JAZF1 and CAPB1, previously
associated with allergic disease and asthma, respectively16,39. The effect sizes remained mostly
stable after adjusting for allergic comorbidities (see table 3). In the replication analyses, only one intergenic SNP, rs114770246, was significantly associated with PA in the GOFA and the
GENEVA cohort but in the opposite direction of effect (p=6,37*10-3 in GOFA and p=5,55*10-3
in GENEVA). This SNP was not associated with gene expression. REPLICATION OF PREVIOUSLY IDENTIFIED SNPS
Of the 47 SNPs with suggestive evidence for association with FA in previous studies, 31 were genotyped or imputed in both PIAMA and Lifelines and met quality control standards. We replicated the associations of 2 and 10 SNPs with FA and PA, respectively (all p<0.05, same direction as previously reported, see table 4). Effect sizes increased after adjustment for allergic comorbidities. Of the 10 SNPs associated with PA, 9 represent one locus (+/-270kb) containing HLA-DRA and HLA-DQB1. Unfortunately, rs12964116, located in SERPINB7 and
previously associated with any FA11, was not genotyped or imputed according to quality
control standards. One SNP, rs71353401, in LD with this SNP (D’=0.98, R2=0.8533) did not show
a significant association with LikelyFA (p=0.76).
Of the 136 SNPs associated with any allergic disease (asthma, rhinitis, eczema), identified by the SHARE consortium, 121 were genotyped or imputed in both cohorts and met
quality control standards16. Of these, six were associated with LikelyFA, three with LikelyPA
and one was associated with both phenotypes in the same direction as previously published (see table 4). The effect size decreased after adjustment for allergic comorbidities in six of these SNPs. The most significantly replicated SNP was rs10174949, which was associated with FA (p=3.04*10-3) and is located in an intron of Long Intergenic Non-Protein Coding RNA 299(LINC00299). This SNP is in strong LD with multiple SNPs characterized by enhancer and promotor histone marks and transcription factor binding sites in multiple tissues but none of
the SNPs were associated with gene expression 34–36. Another SNP, rs6990534, was associated
with FA but not with PA (p=4.50*10-3 and p=0.94, respectively). This SNP is a significant eQTL
for PVT1 in thyroid tissue36 (6.2*10-7) and MYC in blood34 (2.83*10-5). PVT1 was previously
associated with eosinophil count40 and was reported to regulate both IL-6 release and
Chap
ter 9
147 was 17.2 and 17.3 years. Being female was associated with a higher risk of LikelyFA in the
Lifelines, but not in the PIAMA cohort. Cases classified as LikelyFA reported more eczema, asthma and nasal allergy in both cohorts, compared to subjects classified as NoFA. Subjects in the replication cohorts are described in supplementary table 3. In total, 7 459 872 and 6 657 365 SNPs were present in the PIAMA and Lifelines cohort and subjected to meta-analysis for
LikelyFA and LikelyPA, respectively. There was no evidence of inflation of the test statistics of
this meta-analysis (LikelyFA: l=0.99, LikelyPA: l=0.99). Within the discovery analyses, we
identified 6 and 25 independent top hits associated with LikelyFA and LikelyPA, respectively. All were imputed SNPs. CGTA-CoJo analyses did not reveal any additional independent associated SNPs within 500 kb.
FOOD ALLERGY
In the discovery analyses for LikelyFA, six independent SNPs were identified with suggestive
evidence for association (p<1*10-6, see figure 1). None of the SNPs passed the significance
threshold for genome-wide significance (p<5x10-8). Two were located in the 3’ untranslated
region of RPP30 and an intron of RBFOX1, genes previously associated with eosinophilia and
FA, respectively37,38. In general, the effect sizes remained stable after adjusting for allergic
comorbidities. None of the SNPs was significantly replicated or reached the threshold suggestive for association with FA in the meta-analysis of all four cohorts (see table 2). PEANUT ALLERGY
In the discovery analyses for LikelyPA, 25 independent SNPs were identified with suggestive
evidence for association (p< 1*10-6) of which four reached the threshold for genome wide
significance (p<5*10-8, see figure 2). Of the latter, two were located in introns of MYLK3 and
CMTM8, previously associated with CCR3 signaling in eosinophils and IgG glycosylation,
respectively37. In addition, two SNPs were located in introns of JAZF1 and CAPB1, previously
associated with allergic disease and asthma, respectively16,39. The effect sizes remained mostly
stable after adjusting for allergic comorbidities (see table 3). In the replication analyses, only one intergenic SNP, rs114770246, was significantly associated with PA in the GOFA and the
GENEVA cohort but in the opposite direction of effect (p=6,37*10-3 in GOFA and p=5,55*10-3
in GENEVA). This SNP was not associated with gene expression. REPLICATION OF PREVIOUSLY IDENTIFIED SNPS
Of the 47 SNPs with suggestive evidence for association with FA in previous studies, 31 were genotyped or imputed in both PIAMA and Lifelines and met quality control standards. We replicated the associations of 2 and 10 SNPs with FA and PA, respectively (all p<0.05, same direction as previously reported, see table 4). Effect sizes increased after adjustment for allergic comorbidities. Of the 10 SNPs associated with PA, 9 represent one locus (+/-270kb) containing HLA-DRA and HLA-DQB1. Unfortunately, rs12964116, located in SERPINB7 and
previously associated with any FA11, was not genotyped or imputed according to quality
control standards. One SNP, rs71353401, in LD with this SNP (D’=0.98, R2=0.8533) did not show
a significant association with LikelyFA (p=0.76).
Of the 136 SNPs associated with any allergic disease (asthma, rhinitis, eczema), identified by the SHARE consortium, 121 were genotyped or imputed in both cohorts and met
quality control standards16. Of these, six were associated with LikelyFA, three with LikelyPA
and one was associated with both phenotypes in the same direction as previously published (see table 4). The effect size decreased after adjustment for allergic comorbidities in six of these SNPs. The most significantly replicated SNP was rs10174949, which was associated with FA (p=3.04*10-3) and is located in an intron of Long Intergenic Non-Protein Coding RNA 299(LINC00299). This SNP is in strong LD with multiple SNPs characterized by enhancer and promotor histone marks and transcription factor binding sites in multiple tissues but none of
the SNPs were associated with gene expression 34–36. Another SNP, rs6990534, was associated
with FA but not with PA (p=4.50*10-3 and p=0.94, respectively). This SNP is a significant eQTL
for PVT1 in thyroid tissue36 (6.2*10-7) and MYC in blood34 (2.83*10-5). PVT1 was previously
associated with eosinophil count40 and was reported to regulate both IL-6 release and
148
DISCUSSION
This is the first genome-wide association study on food allergy in European adults and adolescents to date. In the discovery analysis in 392 LikelyFA cases, including 47 LikelyPA cases, and 9 470 controls without FA in two independent discovery populations, we identified SNPs suggestive for association with questionnaire-defined LikelyFA and LikelyPA and showed that effect sizes remained stable after adjustment for allergic comorbidities. Unfortunately, we did not replicate these findings in two independent cohorts of children tested by food challenges and they are therefore candidates for further replication in cohorts with food allergic adults. Interestingly, we replicated 10 SNPs previously associated with PA in children, including well-known HLA-DQB1 SNPs, indicating that this effect continues into adulthood. The effect of these SNPs increased after adjustment for allergic comorbidities. The SHARE consortium recently published a list of 136 SNPs found to be associated with any allergic
disease including astma, rhinitis, and eczema16. Of the 121 SNPs available in our discovery
analyses, we showed that 10 are associated with LikelyFA and LikelyPA as well. However, the effect sizes decreased by up to 35% after adjustment for allergic comorbidities compatible with a more general effect on allergy.
In our discovery analysis, we provided evidence suggestive for associations of LikelyFA and
LikelyPA with SNPS located in introns of genes previously associated with FA (RBFOX138), and
asthma (CABP142), respectively. Furthermore, rs73299490 in an intron of JAZF1 was identified
in the discovery analyses on PA, a gene which was previously associated with allergic rhinitis
in children and any allergic disease (asthma, eczema and hayfever)16,43. However, these were
not replicated in two independent cohorts with children phenotyped by food challenges and are therefore candidates for further replication.
STRENGTHS AND LIMITATIONS
Although this is the first GWAS on FA in European adolescents and adults to date, this study has some limitations. The sample size and particularly the number of cases is still relatively small for a GWAS on a complex genetic disease. Therefore, we only considered SNPs as sentinel SNPs when they had the same direction of effect in both discovery cohorts, to limit false-positive findings. The majority of SNPs associated with LikelyFA and LikelyPA in the discovery analyses have a MAF between 0.01-0.05. Although we studied over 10,000 subjects, the effects estimates were obtained in a relatively low number of cases in these analyses. However, some of the SNPs were identified were in genes previously associated with food allergy or related phenotypes, indicating that they may represent true-positive associations which require further replication. The GENEVA cohort was underpowered to replicate these low frequency risk alleles, because of the limited number of trios with heterozygote parents, which are relevant for the FBAT.
We did not apply correction for multiple testing in the analysis regarding SNPs previously associated with FA or any allergic disease since this data reflects a validation of these variants in adults and adolescents of previously reported findings. However, we tested
in a two-tailed approach and only considered SNPs replicated when they reported the same direction of effect.
The phenotype of subjects in the discovery analysis is based on questionnaire reported
LikelyFA with a culprit food, symptoms and characteristics consistent with immediate allergic
reactions to food. The questionnaire is based on previously published questionnaires, an
extensive literature search and expert opinion as previously discussed3 and might represent
the best questionnaire yet available to identify food allergic patients on this scale. However, we realize that we may still misclassify some cases by this questionnaire which may reduce the power of our analysis. Ideally, subjects within the discovery analyses should be phenotyped by the time-consuming DBPCFC as well, just as the subjects from the GENEVA and GOFA cohorts. Unfortunately, this was unfeasible within both cohort studies.
HETEROGENEITY IN STUDIES
Our study design was based on our hypothesis of shared genetic mechanisms of food allergy in children and adults, irrespective of the type and severity of food allergy. The Lifelines and PIAMA cohorts included adults with a mean age of 53 and adolescents with a mean age of 17 years, respectively. Both replication cohorts, included children with food challenge diagnosed FA. The reported culprit foods among children and adults are generally different since the majority of childhood allergies to milk, egg and wheat resolve during childhood, whereas allergies to peanut, tree nuts and shellfish tend to persist more frequently and are therefore
more common in adults, together with allergies to foods cross-reacting with tree-pollen44,45.
Both discovery cohorts included cases and controls from a general population and the most frequently reported allergenic foods were apple, hazelnut, walnut, shellfish and almond whereas the subjects within the replication cohorts from clinical populations reported hen’s egg, peanut, cow’s milk and cashew allergy. These differences in allergenic foods may explain the lack of replication. Unfortunately, there was no data available regarding the age of onset of the food allergies reported in Lifelines and PIAMA. In addition, the GENEVA study used the expected genotypes for the transmission disequilibrium test instead of imputed dosages. There was no reliable information on FA available for the controls used to compare to the children of the GOFA cohort, which might have deflated the p-values in this cohort.
An interesting finding was that we were able to observe associations of LikelyFA and LikelyPA with 12 out the 31 previously reported associations for FA and PA. Our study therefore adds to the body of evidence describing these genes for FA, including HLA-DQB1 and ZNF652. We also show for the first time that some of the SNPs described for allergic disease by the SHARE consortium relate to FA, such as SNPs located in PVT1 and JAK2.
Nevertheless, the replication of the associations of previously identified SNPs associated with FA and PA indicates that after six GWAS studies on FA, we are closer to identifying a robust gene set associated with FA and the effect of these SNPs continues into adulthood. More genetic studies in these study populations are therefore required to expand the knowledge on the genetic basis of FA.
Chap
ter 9
149
DISCUSSION
This is the first genome-wide association study on food allergy in European adults and adolescents to date. In the discovery analysis in 392 LikelyFA cases, including 47 LikelyPA cases, and 9 470 controls without FA in two independent discovery populations, we identified SNPs suggestive for association with questionnaire-defined LikelyFA and LikelyPA and showed that effect sizes remained stable after adjustment for allergic comorbidities. Unfortunately, we did not replicate these findings in two independent cohorts of children tested by food challenges and they are therefore candidates for further replication in cohorts with food allergic adults. Interestingly, we replicated 10 SNPs previously associated with PA in children, including well-known HLA-DQB1 SNPs, indicating that this effect continues into adulthood. The effect of these SNPs increased after adjustment for allergic comorbidities. The SHARE consortium recently published a list of 136 SNPs found to be associated with any allergic
disease including astma, rhinitis, and eczema16. Of the 121 SNPs available in our discovery
analyses, we showed that 10 are associated with LikelyFA and LikelyPA as well. However, the effect sizes decreased by up to 35% after adjustment for allergic comorbidities compatible with a more general effect on allergy.
In our discovery analysis, we provided evidence suggestive for associations of LikelyFA and
LikelyPA with SNPS located in introns of genes previously associated with FA (RBFOX138), and
asthma (CABP142), respectively. Furthermore, rs73299490 in an intron of JAZF1 was identified
in the discovery analyses on PA, a gene which was previously associated with allergic rhinitis
in children and any allergic disease (asthma, eczema and hayfever)16,43. However, these were
not replicated in two independent cohorts with children phenotyped by food challenges and are therefore candidates for further replication.
STRENGTHS AND LIMITATIONS
Although this is the first GWAS on FA in European adolescents and adults to date, this study has some limitations. The sample size and particularly the number of cases is still relatively small for a GWAS on a complex genetic disease. Therefore, we only considered SNPs as sentinel SNPs when they had the same direction of effect in both discovery cohorts, to limit false-positive findings. The majority of SNPs associated with LikelyFA and LikelyPA in the discovery analyses have a MAF between 0.01-0.05. Although we studied over 10,000 subjects, the effects estimates were obtained in a relatively low number of cases in these analyses. However, some of the SNPs were identified were in genes previously associated with food allergy or related phenotypes, indicating that they may represent true-positive associations which require further replication. The GENEVA cohort was underpowered to replicate these low frequency risk alleles, because of the limited number of trios with heterozygote parents, which are relevant for the FBAT.
We did not apply correction for multiple testing in the analysis regarding SNPs previously associated with FA or any allergic disease since this data reflects a validation of these variants in adults and adolescents of previously reported findings. However, we tested
in a two-tailed approach and only considered SNPs replicated when they reported the same direction of effect.
The phenotype of subjects in the discovery analysis is based on questionnaire reported
LikelyFA with a culprit food, symptoms and characteristics consistent with immediate allergic
reactions to food. The questionnaire is based on previously published questionnaires, an
extensive literature search and expert opinion as previously discussed3 and might represent
the best questionnaire yet available to identify food allergic patients on this scale. However, we realize that we may still misclassify some cases by this questionnaire which may reduce the power of our analysis. Ideally, subjects within the discovery analyses should be phenotyped by the time-consuming DBPCFC as well, just as the subjects from the GENEVA and GOFA cohorts. Unfortunately, this was unfeasible within both cohort studies.
HETEROGENEITY IN STUDIES
Our study design was based on our hypothesis of shared genetic mechanisms of food allergy in children and adults, irrespective of the type and severity of food allergy. The Lifelines and PIAMA cohorts included adults with a mean age of 53 and adolescents with a mean age of 17 years, respectively. Both replication cohorts, included children with food challenge diagnosed FA. The reported culprit foods among children and adults are generally different since the majority of childhood allergies to milk, egg and wheat resolve during childhood, whereas allergies to peanut, tree nuts and shellfish tend to persist more frequently and are therefore
more common in adults, together with allergies to foods cross-reacting with tree-pollen44,45.
Both discovery cohorts included cases and controls from a general population and the most frequently reported allergenic foods were apple, hazelnut, walnut, shellfish and almond whereas the subjects within the replication cohorts from clinical populations reported hen’s egg, peanut, cow’s milk and cashew allergy. These differences in allergenic foods may explain the lack of replication. Unfortunately, there was no data available regarding the age of onset of the food allergies reported in Lifelines and PIAMA. In addition, the GENEVA study used the expected genotypes for the transmission disequilibrium test instead of imputed dosages. There was no reliable information on FA available for the controls used to compare to the children of the GOFA cohort, which might have deflated the p-values in this cohort.
An interesting finding was that we were able to observe associations of LikelyFA and LikelyPA with 12 out the 31 previously reported associations for FA and PA. Our study therefore adds to the body of evidence describing these genes for FA, including HLA-DQB1 and ZNF652. We also show for the first time that some of the SNPs described for allergic disease by the SHARE consortium relate to FA, such as SNPs located in PVT1 and JAK2.
Nevertheless, the replication of the associations of previously identified SNPs associated with FA and PA indicates that after six GWAS studies on FA, we are closer to identifying a robust gene set associated with FA and the effect of these SNPs continues into adulthood. More genetic studies in these study populations are therefore required to expand the knowledge on the genetic basis of FA.
150
TABLES AND FIGURES
Lifelines, n=8 724 PIAMA, n=1138 Likely food allergic
n=332
Not food allergic controls n=8 392
Likely food allergic n=60 Not food allergic controls n=1 078 Age in years mean, SD 52.4, 9.9 53.6, 10.6 17.2, 0.4 17.3, 0.4 Male % (n) 26.2 (87) 43.0 (3 607) * 48.3 (29) 50.3 (542) Asthma % (n) 20.2 (67) 6.7 (559) * 21.7 (13) 4.9 (53) * Eczema % (n) 26.8 (89) 14.0 (1 174) * 31.7 (19, m=3) 8.5 (88, m=38) * Nasal allergy^ % (n) 40.4 (134) 21.3 (1 789) * 56.7 (34) 19.1 (206) * Top 5 reported allergenic foods % (n) Apple: 31.9 (106) Apple: 43.3 (26) Hazelnut: 30.4 (101) Hazelnut: 23.3 (14) Walnut: 26.2 (87) Peanut: 23.3 (14) Shellfish: 20.5 (68) Walnut: 21.7 (13) Almond: 14.8 (49) Cashew: 21.7 (13)
TABLE 1. Descriptive statistics of the Lifelines and PIAMA study population *P<0.05*10-7. ^
for PIAMA, the definition includes hayfever, nasal allergy or allergic rhinitis, m=missing.
A. FO OD AL LE RG Y - A : D IS CO VE RY LIF EL IN ES PI AMA LIF EL IN ES +PI AMA SNP Ch r BP (GRC h37) Lo ca tio n (e ns em bl e. or g) , av ai lab le e QT L d at a M AF (E ns em bl. or g - E UR) EA F (LL) P (L L, Y, G) EA F (Pi ) P (Pi , Y , G) Effe ct alle le Be ta (Y , S ) P (Y , S ) Be ta (Y , S, E ) Be ta (Y , S , E, A, NA ) rs 11 75 84 66 6 170406892 In te rg en ic, e QT L RP1 1-302L 19. 3 C: 0. 30 0. 27 3. 36* 10 -5 0, 29 6. 13* 10 -3 C( m ) 0. 42 8. 41* 10 -7 0. 43 0. 42 rs 14 02 55 93 8 10 92670832 3’ U TR o f A NK RD 1 & R PP 30 *1 T: 0. 04 0. 06 4. 38* 10 -6 0, 06 0. 07 T(m ) 0. 70 8. 46* 10 -7 0. 71 0. 69 rs 11 79 01 03 3 10 121947575 Int er ge ni c T: 0. 01 0. 01 1. 38* 10 -5 0, 01 0. 02 T(m ) 1. 23 9. 64* 10 -7 1. 25 1. 22 rs 72 77 49 27 16 7621357 In tr on RB FO X1 *2 T: 0. 02 0. 02 5. 88* 10 -6 0, 02 0. 02 T(m ) 0. 97 4. 17* 10 -7 1. 06 0. 98 rs 60 33 59 2 20 13023388 eQ TL SP TLC 3, in tr on SP TL C3 C: 0. 08 0. 07 2. 46* 10 -6 0, 07 0. 01 C( m ) 0. 63 8. 71* 10 -8 0. 63 0. 66 rs 19 00 44 44 7 21 33343003 Int ro n HUN K G: 0. 01 0. 02 0. 03 0, 01 1. 36* 10 -7 G( m) 1. 32 7. 71* 10 -7 1. 09 1. 21 FO OD AL LE RG Y - B : R EP LICA TI ON GOFA GE NE VA M ET A A NA LY SI S ( al l 4 co ho rt s) SNP EAF Be ta P ca se s co nt ro ls EAF Z-sc or e P #fa m Ef fe ct a lle le We ig ht Zsc or e P-va lue Di re ct ion (L ife lin es, P IA M A, G OF A, G EN EV A) rs 11 75 84 66 0. 27 -0. 02 0. 72 859 3358 0. 26 -1. 24 0. 22 253 C( m ) 14332 3. 73 1. 94* 10 -4 ++ -- rs 14 02 55 93 8 0. 04 -0. 17 0. 28 859 3358 0. 02 0. 13 0. 90 20 T(m ) 14099 3. 53 4. 11* 10 -4 ++ -+ rs 11 79 01 03 3 0. 01 0. 10 0. 78 859 3358 0. 00 -1. 53 0. 13 9^ T(m ) 14088 4. 21 2. 52* 10 -5 +++ - rs 72 77 49 27 0. 02 0. 01 0. 96 859 3358 0. 01 -0. 80 0. 42 11 T(m ) 14090 4. 24 2. 24* 10 -5 +++ - Rs 6033592 0. 07 -0. 01 0. 90 859 3358 0. 06 -1. 59 0. 11 84 C( m ) 14165 4. 28 1. 88* 10 -5 ++ -- rs 19 00 44 44 7 0. 01 -0. 37 0. 29 859 3358 0. 01 -0. 95 0. 34 16 G( m) 14095 3. 52 4. 26* 10 -4 ++ -- TA BL E 2. T op h its fo r f oo d a lle rg y: a ll s en tin el S NP s w ith a p < 1x 10 -6. P an el A : di sc ov er y ana ly se s, pa ne l B : r ep lic at io n an aly se s. Ch r= ch rom os om e, B P= bas e pai r po sit io n, M AF = m in or al le le fr eque nc y, E AF = ef fe ct al le le fr eque nc y, LL= Li fe lin es , Pi =PI AM A, m =m in or a lle le , Y = adj us te d fo r age in ye ar s, S= adj us te d fo r s ex , E = adj us te d fo r e cz em a, A = adj us te d fo r as thm a, N A= adj us te d fo r nas al a lle rg y, *1ge ne pr ev io us ly as so ci at ed w ith the Om enn sy ndr om e w ith eo sino phi lia 37, *2 P re vi ou sly a sso ci at ed w ith fo od a lle rg y (L i, 2015 38), ^ b ec au se < 10 fa m ili es co ntr ib ute d to th is te st, te st re su lts a re le ss re lia bl e.
