Resolving the Etiology of Atopic Disorders by Genetic Analysis of Racial Ancestry

Abstract

Atopic dermatitis (AD), food allergy (FA), allergic rhinitis (AR) and asthma are common atopic disorders of complex etiology. The frequently observed “atopic march” from early AD to asthma and/or AR later in life as well as the extensive comorbidity of atopic disorders, suggests common causal mechanisms in addition to distinct ones. Indeed, both disease-specific and shared genomic regions exist for atopic disorders. Their prevalence also varies among races; for example, AD and asthma have a higher prevalence in African-Americans when compared to European-Americans. Whether this disparity stems from true genetic or race-specific environmental risk factors or both is unknown. Thus far, the majority of the genetic studies on atopic diseases have utilized populations of European ancestry, limiting their generalizability. Large cohort initiatives and new analytic methods such as admixture mapping are currently being employed to address this knowledge gap. Here we discuss the unique and shared genetic risk factors for atopic disorders in the context of ancestry variations, and the promise of high-throughput “-omics” based systems biology approach in providing greater insight to deconstruct into their genetic and non-genetic etiologies. Future research will also focus on deep phenotyping and genotyping of diverse racial ancestry, gene-environment, and gene-gene interactions.

Introduction

The development of atopic disorders often follows an age-dependent pattern that starts early in life and progresses (marches) from one tissue to another in the same individual, a phenomenon referred to as the “atopic march”^1–3 (Figure 1). The International Study of Asthma and Allergies in Children (ISAAC), which examined the global prevalence of atopic dermatitis (AD), food allergy (FA), allergic rhinitis (AR), and asthma, found that the prevalence of these disorders differed among races.^4,5 Delineating genetic and environmental risk factors that relate more specifically to certain atopic disorders and ancestry than to other factors provides valuable information to address the following questions: Are atopic disorders genetically inherited and/or inter-related? Do all atopic disorders share the same loci or are there unique genetic loci associated with each atopic disorder? Are there ancestry differences in the risk of atopic disorders and what are their potential sources? Is an individual ancestry or race a good predictor of atopic disorders? In this review, we discuss the genetics of atopic disorders, racial ancestry risk differences, role of gene-environment interactions, and recent evidence of shared and unique etiologies of atopic disorders. As the world is becoming highly multiethnic and racially admixed, and, thus, genetically diverse, individuals who self-identified as a particular race may have widely varying genetic ancestry.^6,7,8 Thus, we discuss admixture mapping approach from the vantage point of disentangling ancestry-specific risk loci. Enormous progress has been made on developing high-throughput genotyping technology but focus on devising high-throughput methods to standardize phenotyping is still at its infancy. Our review emphasizes the need for high-throughput methodologies for constructing causative models that will help resolve the underlying clinical and ancestral complex architecture of atopic disorders to yield greater insights. These approaches have the potential to target atopic phenotypes with interventions relevant to specific disorders and ancestry group. In addition, we provide an overview of the complex components of atopic disorders, as well as future research strategies.

Figure 1. Schematic diagram illustrating the age-dependent progression “atopic march” of atopic disorders.

Food allergy and atopic dermatitis peak in the first years of life and decline after that time. Asthma and allergic rhinitis increase over time as sensitization develops further. Food sensitization can be used as an early indicator for identifying children at risk for subsequent allergic disease who may benefit from early intervention.

Atopic Disorders and Race

It has long been recognized that there are differences among racial and ethnic groups in the prevalence and severity of atopic disorders,^9–12 and these differences are believed to have both environmental and genetic components.^13–15 It is notable, however, that there are racial and ethnic minority populations in the U.S. that are under-represented in cohort studies and clinical trials and that the vast majority of genome-wide association studies (GWAS) of atopic disorders conducted thus far have used populations of European ancestry. Genomic studies often do not translate well from one race to another due to ancestral variation. Different populations have different linkage disequilibrium (LD) structures (otherwise known as haploblocks structures), allele frequencies and effect sizes¹⁶ and, thus, single-nucleotide polymorphism (SNP) associations may be population-specific.¹⁷ As a consequence, our current knowledge might be skewed toward the European ancestry–based etiology of atopic disorders, and important genetic and environmental clues to population differences may have been missed. This is because most of the risk alleles and index variants are derived using commercial genotyping arrays ascertained from populations of European ancestry. This drawback biases the generalization of results of these studies to other ancestral populations since not all atopic disorder loci found in Europeans are transferable to all populations. Identification of causal variants through trans-ancestry fine mapping and sequencing in different populations will help to understand the genetic contribution to the disparity in atopic disorder prevalence. In particular, the differential and lower level of LD in African ancestry will likely greatly improve the fine mapping of causal variants using trans-ancestry meta-analysis efforts. In addition, many of the diagnostic tools using in atopic diseases such as pulmonary function tests are developed mainly for Europe ancestry. However, marked differences in lung function and allergic sensitization between children of different racial backgrounds exist even when known confounders including socioeconomic status are taken into account.^18–20 For example, spirometric assessment of forced expiratory volume in 1 s (FEV₁) and forced vital capacity (FVC) are known to be reduced by 14% in individuals of African ancestry compared with those of European ancestry.^21–24 The use of race-specific reference equations can help to minimize such differences and improve accuracy.^20,25

Racial ancestry differences can have implications for clinical practice; an example is the use of long-acting β2-agonists (LABAs) in patients with asthma. Though combining corticosteroid treatment with a LABA can lead to better asthma control for most patients, a small proportion of patients, including African Americans, appear to be at increased risk for harmful effects and even fatal outcomes from the use of LABAs.²⁶ The reason for this observed difference is still controversial. Some studies have suggested that genetic differences in the β2-adrenergic receptors and their interaction with race may be responsible.^27,28 Several studies have shown that African American asthmatics have an increased likelihood for treatment failures and overall differential response to treatment that may be caused by genetic variants specific to their ancestry.^29,30 However, race is a crude proxy to genetic ancestry and falls short of explaining the variation in response to medication.⁶ Investigators trying to address the knowledge gaps in our understanding of the racial aspects of atopic disease are faced with several challenges. Assembling minority cohorts large enough to power the detection of genetic associations can be difficult and requires very active outreach, especially to racial/ethnic groups that may be underserved by the health care system. Another approach is to increase power by using newer statistical methods such as admixture mapping. In conventional association studies, the high degree of recent genetic admixture in many racial minorities in the U.S. can lead to population stratification—ancestry differences between cases and controls. It is well recognized that racial minorities in the U.S., including African-Americans and Latino groups, have a lower socio-economic status, which is often associated with environmental factors such as diet, presence of allergens, and pollution exposure.¹⁵ These factors can have a direct effect on the development of allergic disorders and need to be carefully adjusted for in statistical analyses.³¹

Different ancestral populations have different proportions of rare genetic variants, and it has been shown that populations of African ancestry have up to three times as many rare variants as populations of European and Asian origin.³² Although rare variants are believed to be enriched for functionality,^33,34 they present a challenge in association studies because of the large sample sizes needed for their detection.^35,36 Trans-ancestry studies have been invaluable in deconstructing the genetic architecture of complex diseases, such as asthma. They provide an opportunity to replicate signals in independent populations and for meta-analyses to boost statistical power. For example, African genome analysis enables prioritizing candidate variants in follow-up and fine-mapping functional variants due to the unique short-range LD in African ancestry. The use of a population with short-range LD will result in the greatest localization success rate in distinguishing the causal SNP from its neighbors.^37–39 The recent completion of whole genome sequencing data from the 1000 Genomes Project (http://www.1000genomes.org), encompassing over 88 million variants from 26 world-wide populations (2,504 individuals), will make it easier to optimize LD pattern, imputation and SNP selection for specific racial populations, thereby increasing power in GWAS.⁴⁰

Genetics of Atopic Disorders

The strongest evidence for the importance of genetic factors in atopic disease stems from twin studies. The approximately 80% concordance rate for AD among monozygotic (MZ) twins far exceeds the concordance rate of 20% observed among dizygotic (DZ) twins.^41,42 Heritability estimates for atopic diseases are as high as 81% for FA, 84% for AD, 91% for AR, and 95% for asthma.⁴³ Although FA, AD, AR, and asthma are clinically diagnosed as discrete diseases, current evidence suggests that they are genetically related conditions, with correlation estimates being 0.55 for asthma and AD, 0.47 for asthma and AR, and 0.62 for AD and AR.⁴⁴ The phenotypic similarity among atopic disorders suggests that elements of biological etiology may also be shared among these disorders. Risk genetic variants that are shared or influence multiple atopic disorders, also known as “pleiotropic” effects, can help explain common pathogenic features and provide suggestions for development of novel therapies.

Atopic disorders are considered a type I hypersensitivity immune response associated and/or mediated in part by immunoglobulin E (IgE) antibodies to environmental antigens (Table 1). Pathways that have been associated with atopic disorders fall into several broad categories, such as those involved in the epidermal barrier function and those involved in regulating the innate and adaptive immune response, including IgE sensitization^45,46 (Table 2). Lists of disorder-specific (Fig. 2A for FA; Fig. 2B for AD; Fig. 2C for AR; and Fig. 2D for asthma) enriched genes based on over 15.74 million PubMed abstracts are shown using Acumenta Biotech Literature Lab.⁴⁷ The genes are ranked by the relative intensity of their co-occurrence with each atopic disorder. The ranking is shown with the most intensely related genes starting at the 12 o’clock position in the inner ring and descending in a clockwise direction. Table E1 shows list of genes ranked in the order of co-occurrence with each atopic disorder. To understand the relationship among atopic disorders in terms of their genetic etiology, we investigated genetic overlap using ranked gene lists from each disorder. Among the top ranked genes listed in Figure 2, there were 43 gene overlaps between AR and asthma, 29 between AD and asthma, 28 between FA and asthma, 27 between AR and AD, 22 between FA and AD, 22 between FA and AR, and 16 among all disorders (Figure 3). AR and asthma commonly coexist and are regarded as “unified airways disease.”⁴⁸ Thus, there appeared to be more genetic commonalities between AR and asthma than between AR and FA or between AR and AD. To characterize the main functional networks/pathways underlying the 16 shared genes (IL5, IL4, TSLP, IL10, RNASE3, IGHG4, IL13, CCL11, IFNG, RNASE2, FCER2, CD4, FOXP3, IL4R, KCNE4, and CCL26) among atopic disorders, we carried out network analysis using Ingenuity Pathway Analysis (IPA). Figure 4 and Table E2 show the most overrepresented network and functional ontologies among these shared genes, respectively. There is 1 significant network which lies in regions of high network connectivity (“hubs”) centered at IL4, a key mediator of allergic airway diseases.⁴⁹

Table 1.

Common atopic disorders

Atopic disorder	Clinical symptoms	Allergens
Food allergy	Rash, especially around the mouth; sometimes diarrhea	Peanuts, tree nuts, milk, egg, fish
Atopic dermatitis	Chronic, relapsing, intense itching, particularly around skin flexures at the wrist, elbow, ankle, and knees	Egg, cow’s milk, wool
Allergic rhinitis	Blocked and runny nose, sneezing, and itching and streaming eyes Intermittent	Pollen, house dust mite, animal dander
Asthma	Episodic or recurrent breathlessness with cough, wheezing, and chest tightness	Pollen, house dust mite, animal dander, molds, food allergens, viral infections, drugs, smoke

Table 2.

Select pathways involved in atopic disorders

Disorder	Pathway	Functions	Reference
Atopic disorders	Innate immunity and immunoregulation	Triggering immune response Immunoregulatory Antigen presentation	306,307
	Th2-cell differentiation and effector functions	Th2-mediated responses	308,309
	Epithelial biology and mucosal immunity	Chemoattractants Epithelial barrier integrity Protection against proteases	310
	Lung function, airway remodeling, and disease severity	Formation of epithelial-mesenchymal trophic unit Negative regulation of the TLR and IL-1R pathway Bronchial smooth muscle relaxation Allergic airway inflammation and airway remodeling	311,312

Figure 2. Genes associated in the literature with each atopic disorder.

Based on mining of 15.73 million PubMed abstracts ((01/01/90 to the present) by Literature Lab™ from Acumenta Biotech,⁵⁷ genes enriched in FA, AD, AR, and asthma are shown. The rankings of relative strengths of the gene/disease associations in the literature are as follows. The genes with the strongest relationship to each atopic disorder revolve around the center as measured by LPF (log of the product of frequency) starting at the 12 o’clock position in the inner ring descending in a clockwise direction, as indicated by arrows, and outward. Thus IGHG4, FLG, RMASE3, and IL5 have the most co-occurrence and genes MIR604, HRH1, GBP5, and ALOX5AP have the least co-occurrence with FA, AD, AR, and asthma, respectively. The LPF measures relative intensity or strength of co-occurrences for pairs of items in a large corpus, in this case genes and diseases.⁴⁷ The LPFs are computed by Literature Lab™ for co-occurrences of all human genes in the NCBI Gene Database and each disease identified in the NCBI MeSH Diseases ontology (about 4,200 terms) in 15.73 million PubMed abstracts. The LPF is calculated as: LPF= Lg(x/G * x/T), where G is number of genes, T is the number of terms and x is co-occurrence between G and T. The greater the absolute magnitude of the LPF the weaker the association ; the closer the LPF value to zero, the stronger the association.

Figure 3. Venn diagram of unique and shared genes among atopic disorders (FA, AD, AR and asthma).

There are more genes shared between AR and asthma than between any other pair of atopic disorders. To understand the relationship among atopic disorders in terms of their genetic etiology, we investigated genetic overlap using ranked gene lists from each disorder. Among the top ranked genes listed in Figure 2, there were 43 gene overlaps between AR and asthma, 29 between AD and asthma, 28 between FA and asthma, 27 between AR and AD, 22 between FA and AD, 22 between FA and AR, and 16 among all disorders. AR and asthma commonly coexist and are regarded as “unified airways disease.”⁴⁸ Thus, there appeared to be more genetic commonalities between AR and asthma than between AR and FA or between AR and AD. However, it is to be noted that our overlap measure uses gene lists in each disorder based on studies in the literature so far, and these diseases are not equality studied (for example, asthma has 110,428 PubMed abstract compared with 6,348 for AR). As we generate more data for all diseases, we will be able to determine disease-specific or overlapping genes with more confidence.

Figure 4. IPA network for 16 genes shared among FA, AD, AR and asthma.

Genes with red nodes are focus genes in our analysis, the others are generated through the network analysis from the Ingenuity Pathways Knowledge Base (http://www.ingenuity.com). Edges are displayed with labels that describe the nature of the relationship between the nodes. The lines between genes represent known interactions, with solid lines representing direct interactions and dashed lines representing indirect interactions. Nodes are displayed using various shapes that represent the functional class of the gene product. Examination of the networks would move us toward a holistic understanding of atopic disorders.

As expected, the majority of the sixteen genes associated with all four atopic diseases are involved in fundamental inflammatory processes, and many of them, IL4, IL4R, IL13, IL5, IGHG4, FCER2 and TSLP, specifically function in the Th2 cell immune response characteristic of allergy. The eotaxins CCL11 (eotaxi-1) and CCL26 (eotaxin-3), and the RNases RNASE2 and RNASE3, as well as IL5, are important for eosinophil recruitment and function, also a hallmark of allergic disease. IL10 and FOXP3 are important for the activity of regulatory T cells. IL10 plays an essential role in inducing an immunoregulatory phenotype in B cells that exerts substantial anti-inflammatory and immunosuppressive functions.⁵⁰ IL5 is the most important interleukin responsible for eosinophilic airway inflammation in asthmatics, and has become a definite target for treatment.⁵¹ IL5 is produced by CD4+ Th2 cells, mast cells, eosinophils, basophils and innate lymphoid cells type 2 (ILC2) and is involved in several steps of eosinophil development and function.⁵² Others, such as IFNGγ and IL17A are involved in the functioning of Th1 cells and Th17 cells, respectively. Several of the sixteen genes, CCL11, IFNγ, IL17A, RNASE2, and RNASE3, also have antimicrobial activity, but it is unclear what role, if any, this activity has in the development of allergic responses. TSLP is produced by epithelial cells and links the innate immune response to Th2 cell-driven inflammation. Its association with all four allergic diseases suggests the important role of the epithelial barrier in allergic response. Despite these findings, overlapping genes are based on studies in the literature so far, and these diseases are not equality studied (for example, asthma has 110,428 PubMed abstracts compared with 6,348 for AR). As we generate more and more data to all diseases, we will be able to determine disease-specific or overlapping genes with more confidence.

Recent studies suggest that the most important factor that predicts atopic march is an impaired epidermal barrier mechanism of the epithelium with multiple organ-specific inflammation involvement (i.e., skin, nose, airway, gut/gastrointestinal tract)⁵³ (Figure 5). A better understanding of the role of epithelial dysfunction along with crosstalk with innate immunity in the lung, skin, and gut in ancestry-specific manner could provide a framework on which new therapeutic strategies can be planned to prevent exacerbations and to alter the natural course of these disorders. Figure 6 represents a conceptual model that deconstructs the molecular and genetic factors in the pathogenesis of atopic disorders. Major genes related to epidermal protease–inhibitor function, epidermal development, keratinocyte morphology/function, lipid metabolism, anti-oxidant function, cytokine/chemokine/growth factors, and innate immunity are shown. Genes related to protease function, epidermal development, and keratinocyte morphology can directly contribute to epidermal barrier dysfunction, whereas lipid genes can modulate barrier function, inflammation, and anti-oxidant function. Genes related to cytokine/chemokine function modulate lymphocyte activity, as well as the expression of innate immune genes. Therefore, atopic disorders can be viewed as a collection of related diseases, characterized by clinical manifestations in several different organs that are linked through similar pathophysiological processes including epidermal barrier dysfunction and cutaneous inflammation, environmental insults such as allergen challenge and microbial infections and “allergic” Th2 and Th17/Th22 cell responses.⁵⁴

Figure 5. The genetic and environmental interplay of atopic disorders.

The relationship between genetic and environmental factors determine the overall outcome of atopic disorders. Skin barrier dysfunctions are part of the spectrum of allergic disorders with immunoglobulin E (IgE)-mediated sensitization and T helper type 2 (Th2) immune dysregulation. The interactions between epidermal barrier dysfunction and dysregulation of innate and adaptive immunity along with environmental risk factors contribute to the pathogenesis of atopic disorders.

Figure 6. Deconstructing atopic disorders using genetic and molecular information.

Complex interplay between functional gene clusters and interpretation based on a model that integrates each atopic disorder in a central theme are presented. A mechanism by which epidermal barrier dysfunction may lead to inflammation and allergic sensitization in atopic disorders is shown. In parallel, genetic and molecular analysis of atopic disorders have evidenced the key role played in the disease mechanisms by epithelial skin cells, especially epidermal keratinocytes, with an abnormal pattern of production of cytokines and chemokines that could trigger sustained, chronic inflammation. Current data favor the paradigm shift in which the downstream systemic effect of allergen penetration through the impaired skin barrier causes immune cells to mount an exaggerated inflammatory response at any allergen-exposed epithelial surface.^304,305 A decreased skin barrier response allows for increased susceptibility of the skin to allergens. In addition, decreased skin barrier leads to an increase in skin pH, altered keratinocyte adhesion properties, and both increased serine protease activity and inflammation. Among multiple reported genes relevant to the cellular location of gene products, only the well-established ones are shown for clarity.

Systematic reviews have revealed that several candidate genes are clustered in chromosomal regions that have been linked to all atopic disorders, such as chromosome 5q31-32, 6p21.3, 11q13, and 12q14-24.⁵⁵ Their close proximity in linked regions suggest increased potential for inheritance of all of these disorders simultaneously.⁵⁶ Multiple studies including our 15.73 million PubMed abstracts search by Literature Lab™ from Acumenta Biotech have shown that several genes, including IL13, IL4 and IL5, belong to pathways that are common to atopic diseases.⁵⁷ However, there are also disorder-specific genes. Examples of primary disorder-specific genes include ADAM33 and ORMDL3 for asthma,⁵⁸ FLG for AD, and IL6 for AR⁵⁹ and recently CAPN14 for the FA-related disorder eosinophilic esophagitis.⁶⁰ Using the results of recently published SNPs from the GWAS catalog (www.genome.gov/gwastudies) and PhenoGram visualization software,⁶¹ we showed shared and unique genetic etiologies among atopic disorders grouped by ancestry (Figure 7). Chromosome 6 shows much of the genomic overlap both at the SNP and ancestry level for atopic disorders. Genetic loci that are shared among atopic disorders, as well as ancestries, would indicate shared molecular pathways in the pathogenesis of atopic march. On the other hand, the lack of shared etiology among atopic disorders and ancestries would indicate distinct pathogenic mechanisms or pathways underlying each atopic disorder and ancestry. A partial overlap would suggest that the combination of disease-specific and disease-overlapping genetic risk factors drive the progression of atopic disorders. Thus, studying the genetic/molecular determinants of these disorders will eventually enable us to deconstruct atopic disorders into distinct genetic/molecular phenotypes, leading to more customized disease management options. The challenges in disentangling atopic disorders include the very large sample sizes required to delineate associations that are specific to each atopic disorder, which results from the over 80% overlap among atopic disorders.⁴³ Omics approaches such as DNA sequence variation (genomics), genetic variants in regulatory regions (regulomics), gene expression (transcriptomics), regulation of gene expression (epigenomics), protein modification (proteomics), microbes (microbiomics), metabolites (metabolomics), and environmental exposure (exposomics) could help to identify molecular signatures and unravel the interrelationships between these disorders and diseases pathogenesis. In addition, the application of a systems biology approach could help to examine shared and unique genetic risk factors and functional categories among atopic disorders to identify at-risk individuals in order to prioritize them for early preventive interventions.

Figure 7. Atopic disorder NHGRI GWAS catalog variants grouped by racial ancestry.

Atopic disorder–related variants identified through genome-wide association studies (GWAS) [NHGRI catalogue of published GWAS was searched using the following terms: food allergy, atopic dermatitis, asthma, and allergic rhinitis]. PhenoGram was used to plot and visualize NHGRI GWAS catalog association results for potentially pleiotropic single-nucleotide polymorphisms (SNPs) in atopic disorders.⁶¹ An Ideogram of all 22 chromosomes is shown, along with the X and Y chromosomes. Horizontal lines on the chromosomes correspond to the base-pair location of each SNP, while lines which project from the chromosomes connect the SNP to colored circles that represent the phenotype(s) associated with the SNP. Overlapping and allergy-specific susceptibility loci among ancestries from GWAS might help to understand the putative pathogenetic relationship between allergy-related phenotypes and racial ancestry. However, loci from GWAS should be interpreted with caution, due to potential issues such as limited sample size in minority populations. Longitudinal and deep phenotyping (measurements at different molecular levels), of individuals along with information on genetic and disease specific environmental risk factors are relevant to study atopic disorders across racial ancestry.

Gene-Environment Interactions in Atopic Disorders

The pathogenesis of atopic disorders usually involves the complex interplay of genetic and environmental factors, which can be assessed statistically as gene-environment interactions. For example, several studies implicate that in utero and early-life environmental tobacco smoke (ETS) exposure leads to impaired lung function and increased risk of asthma;⁶² the ETS exposure modifies the strength of the association between genetic variants and asthma.^63–69 Studies of gene-environment interactions can involve one or more specific candidate genes or comprehensive whole-genome scans. Since much larger sample sizes are needed to detect interactions than main effects, the cohort size required to power genome-wide environmental interaction (GWEI) studies can be prohibitive, and thus far most studies of gene-environment interactions have employed a candidate-gene approach.^67,70–72 Another challenge in conducting GWEI studies is the absence of well-characterized cohorts with regard to environmental variables. The best characterized, allergy-related gene-environment interaction to date involves SNPs in the microbial sensor CD14, which interacts with several toll-like receptors and plays a critical role in the innate immune response to microbial insult. Although most studies have focused on the impact of postnatal environmental exposures, there is increasing evidence that prenatal exposures may also be important.^73–75 Epidemiologic studies examining the effect of prenatal exposures on the development of allergic disease have shown that maternal exposure to farming environments during pregnancy is associated with decreased rates of AD, AR and asthma in their children.⁷⁶ Early exposure to endotoxin, a bacterial cell wall component, appears to offer protection against allergic sensitization and asthma.⁷⁷ Several reports have demonstrated a modifying effect of a SNP in the promoter region of the microbial sensor CD14 on the relationship between endotoxin exposure and atopy.^78–80 Polymorphisms in CD14 have also been shown to interact with microbial exposures⁸¹ and pet exposure,⁸² as well as environmental tobacco smoke,⁸³ on the outcome of allergic disease.

Environmental factors associated with racial and ethnic groups are major confounders in genetic association studies.⁸⁴ However, there are very few published reports on race-specific gene-environment interactions to date. An example of a gene-environment interaction modified by race/ethnicity is the impact of endotoxin exposure on the innate immune system. Using a birth cohort that had 53% African American and 38% European American ancestry children, Williams et al.⁸⁵ showed that interaction between the CD14 genotype and endotoxin exposure on the number of regulatory T cells was limited to the African American subjects. Furthermore, in a study of CD14 gene-environment interaction in the U.S. Latino population, Shoudhry et al.⁶⁸ found that a modifying effect of the CD14 genotype on the association between IgE levels and exposure to ETS was specific to the U.S. Puerto Rican subjects and not observed in the U.S. Mexican subjects. Importantly, Puerto Ricans have a higher admixture of African ancestry than do Mexicans.⁸⁶ ETS has also been shown to interact with the IL4 genotype on the outcome of wheezing in infants.⁸⁷ This effect was specific for the African American population, and neither ETS exposure nor IL4 genotype alone had a significant effect on wheezing. These studies show that race-specific gene-environment interactions can be demonstrated in relatively smaller cohorts using a hypothesis-driven, candidate-gene approach.

The major challenges in uncovering gene-environment interactions in population-based GWAS include inadequate sample size/power, extensive computational demand and the burden of multiple testing for the total numbers of SNPs.⁸⁸ A potential approach to reduce the burden of multiple testing in relatively small sample sizes is to use admixture mapping (AM) and limit association testing to regions that exhibit different risk allele frequencies among ancestral populations. AM performs many fewer independent tests (i.e., smaller penalty due to multiple tests) across the genome than current GWAS.^89–93 Modeling interaction between ancestry and environmental exposures could potentially reveal effect modification on asthma risk.⁸⁴ Large cohort initiatives such as the Genetics of Asthma in Latino Americans (GALA) study,^68,94 as well as new analytic strategies,⁹¹ are expected to facilitate more systematic genome-wide searches for gene-environment interactions in the future.⁹⁵ Furthermore, assessing high-throughput phenotype and exposure definitions with careful adjustment for potential confounders increases the chances of finding significant interactions. However, standardizing environmental exposure assessment methods for “environmental phenotyping” is needed. For example, cotinine measurement in biofluids may be a better and more objective way to assess exposure to environmental tobacco smoke than the use of questionnaires because of underreporting and/or exposure outside the home.^96,97 The progressive accumulation of phenotypic differences between genetically identical MZ twins living in different environments illustrates how pollution, smoking, mold, and diet can shape human susceptibility to diseases. MZ twins were epigenetically indistinguishable early in life but as they adopted different lifestyles, substantial differences in disease susceptibility arose over time.⁹⁸ Therefore, MZ twin discordance for atopic disorders could be interpreted as the result of environmental factors that modulate susceptibility through a change in the profile of epigenetic modifications that ultimately determine gene function. In addition to gene-environment interactions, gene-gene interactions, also known as epistasis, likely play an important role in the genetics of atopic disorders.^99,100

Genetics of Food Allergy

FA is frequently associated with AD, and together they are associated with the development of AR and asthma later in life.¹⁰¹ Many epidemiological studies have suggested that early life food sensitization is associated with an increased risk of AD, AR and asthma later in life.^102–106 FA has a strong familial aggregation.¹⁰⁷ A child with a parent or sibling with peanut allergy, one of the most common forms of FA, has a 7 times higher risk of developing the condition than do children without familial risk factors.¹⁰⁸ Twin studies have demonstrated that the concordance of FA in MZ twins is much higher than in DZ twins,^109,110 with concordance rates of 64.3% and 6.8% in MZ and DZ twins, respectively. Though the high concordance in MZ twins suggests a strong genetic component, environmental factors also contribute to the disease since the MZ twin concordance is not 100%. The same study estimated the heritability of FA to be as high as 81%, which also does not rule out roles of common family environmental factors. Thus far, multiple genes and their variants have been interrogated for their association with FA. Since FA, like AD, was primarily thought to be a disease of the immune system, all of the genes investigated and found to be associated with FA, aside from FLG, were immune-related and included the major histocompatibility complex cluster, CD14, FOXP3, STAT6, SPINK5, TNF, GSTP1, IL13, IL10,^111–113 and IL4RA.¹¹⁴ In general, these were small-scale, candidate-gene studies that lacked replication, and the findings were frequently inconsistent among studies.

In light of the very strong association of null mutations in the FLG gene with AD that was first discovered in 2006 and subsequently replicated in more than 30 studies, Brown et al. hypothesized that FLG would be a candidate gene for FA. Indeed, they found that FLG was associated with peanut allergy regardless of AD status.¹¹⁵ This was the first genetic evidence that FA, like AD, was at least in part due to a defect in the skin barrier. In a follow-up study, the same group demonstrated the association of FLG with peanut allergy regardless of asthma status.¹¹⁶ Though still a matter of debate, it has been hypothesized that in children with FA, sensitization to peanuts primarily occurs through a defective skin barrier rather than through ingestion.¹¹⁷ The first GWAS of well-defined FA in 2,197 participants of European ancestry from the Chicago Food Allergy Study, including specific subtypes (peanut, milk, and egg), identified peanut allergy–specific loci in the HLA-DR and HLA-DQ gene region at 6p21.32.¹¹⁸ This gene region accounted for approximately 20% of PA in the study population. These associations were replicated, and the variants were associated with differential DNA methylation levels at multiple CpG sites.¹¹⁸

Available evidence suggests that FA, like AD and asthma, is the result of a complex interaction between genetic and environmental factors that could include dietary habits (timing of food introduction and route of exposure, type of food (e.g., reduced consumption of specific fatty foods), strict hygiene, ETS exposure, prematurity, low birth-weight, and vitamin D deficiency among other factors.^119,120 Available data indicate that there has been a rise in the prevalence of FA in recent times.¹¹⁹ Indeed, data from the National Health Interview Survey showed that in the age group of 0–17 years, the proportion of children with FA increased from 3.4% in 1997–1999 to 5.1% in 2009–2011.¹²¹ Since widespread fundamental and heritable sequence changes in the human genome are very unlikely during the span of a few decades, it is assumed that environmental factors are responsible for the rise in prevalence of FA.

Epigenetics, which includes changes in gene expression due to environmental influence without alteration in the DNA sequence, has been implicated. It has been suggested that heritable epigenetic changes resulting from the varying rate of development of tolerance to inhaled and ingested food allergens could be the reason for the current epidemic of FA.¹²² Indeed, food can be a strong inducer of epigenetic modification, and maternal diet has been proposed as a risk factor behind the subsequent development of FA in the offspring.¹¹² A recent epigenome-wide association study that investigated the DNA methylation profile of CD4⁺ T cells, which are instrumental in regulating the allergic response, found differences in methylation in children with FA compared to normal controls, providing fresh evidence of epigenetic regulation of FA.¹²³

Food Allergy and Race

Despite FA affecting between 1%–10% of all American children,¹²⁴ many investigators have noted that the prevalence of FA varies among races. Specifically, studies have generally found African ancestry to be a risk factor for food sensitization and FA compared with European ancestry. In the multiethnic Boston Birth Cohort consisting chiefly of Black and Hispanic children, Kumar et al. noticed food sensitization to be associated with both self-reported Black race and African ancestry.¹²⁵ In a large, nationwide cross-sectional survey, Gupta et al. found both Black and Asian children to be at a higher risk and Hispanic children to be at a lower risk of developing FA compared with White children.¹²⁶ In a clinic serving low-income, minority populations in New York, Taylor-Black and Wang identified a higher prevalence of FA allergy among African American children compared with children who were of Hispanic or multi-racial origin.¹²⁷ Preliminary data from a longitudinal, general birth cohort study in which skin testing for food allergens was performed indicated higher rates of sensitization in African American children than in their White counterparts.¹²⁸ Data from the National Health Interview Survey, 1997–2011, demonstrated that the prevalence of FA was lower in Hispanic children than in non-Hispanic, White and non-Hispanic, Black children.¹²¹ A meta-analysis by Keet et al. found that despite the prevalence of self-reported FA among American children rising over the last 20 years, Black children as a group have experienced the greatest increase.¹²⁹ Though the data generally tend to support a higher prevalence of FA in African Americans, there have also been reports that contradict this claim. In a study that included 2 public schools and 2 charter schools from New York, Taylor-Black et al. did not find a higher rate of FA among Black children compared to White children.¹³⁰ A recent meta-analysis by Greenhawt et al. concluded that though there was some evidence of increased rates of food sensitization and FA among the Black population, the available data were still insufficient to label Black race as a risk factor for FA.¹³¹

Even if differences truly exist among racial groups in their susceptibility to FA, the reasons behind such differences are not readily apparent. Though it is tempting to point to genes, it is to be noted that self-reporting as “African American”, “Hispanic”, or “Latino” could also be a surrogate for environmental variables including diet, socioeconomic status, and other lifestyle factors.¹²⁵ For example, African American children of families with significantly lower income were found to be at the greatest risk for FA and food sensitivities, pointing to the probability that discrete environmental and genetic factors are involved.¹¹⁰ Though many of the studies did adjust for some of these variables, the possibility of residual confounding factors cannot be excluded. Thus, methodological variation precludes firm inter-racial comparison of FA. Another possibility is gene-environment interaction. In a recent study, Koplin et al. found increased prevalence of peanut allergy among infants with parents born in East Asia when compared with infants who had parents born in Australia or Europe.¹³² Notably, Asian-born parents did not exhibit higher prevalence of FA compared with Australian-born parents. The authors speculated that although genetically susceptible, Asians born in Asia do not develop allergies due to protective effects of the environment in their native countries and that since Asian infants born in Australia lack this putative protective umbrella, the deleterious effects of allergy genes would be less restrained. Thus, the relationship between specific IgE level and skin prick test sensitization to FA needs to be considered within the context of race.¹³³

Genetics of Atopic Dermatitis

AD is one of the most common skin disorders and affects 5–20% of infants worldwide with lesser prevalence in adulthood.¹³⁴ The pathogenesis of AD is complex and requires the interplay of genetic and environmental risk factors. It has long been known that AD clusters in families. About 56% of the children who had a parent affected with AD developed the disease.¹³⁵ Twin studies of AD have demonstrated concordance rates of up to 86% in MZ twins and up to 50% in DZ twins, respectively.¹³⁶ Candidate-gene studies in AD have reported associations with numerous genes, the majority of which belong to one of two broad biological pathways, the epidermal barrier maintenance pathway (FLG) or the immune response pathway (e.g., CD14, IL4, IL4RA, IL13, TLR2, SPINK5, and RANTES).¹³⁶ A total of 6 GWAS and GWAS meta-analyses have been performed on AD to date. These studies have identified 19 loci that reached the genome-wide significance level.¹³⁷ These loci include 1p21.3, which contains the FLG gene, and 5q11.1, which is adjacent to the TSLP gene. The FLG gene, discussed in detail below, was first identified as a strong candidate for AD in 2006¹³⁸ and has since been associated with AD in numerous studies conducted in people of European and Asian ancestry.¹³⁹ The TSLP gene, which acts as a promoter of T helper type 2 (Th2) cell differentiation and an activator of Th2 cytokine–mediated inflammation¹⁴⁰ has also been associated with AD in recent reports.^141,142

Multiple studies have shown a link between AD and filaggrin, a protein that maintains skin and mucosal barrier function. The gene encoding filaggrin, FLG, is located within the epidermal differentiation complex (EDC), a cluster of genes in chromosome 1q21 that are involved in maintaining an intact skin barrier. The association of loss-of-function mutations in FLG with AD is one of the strongest associations to have been uncovered in complex disease genetics to date.¹⁴³ Carriers of specific loss-of-function mutations in FLG can have an approximately 3-fold greater chance of developing AD than do non-carriers.¹³⁹ This discovery has led to a paradigm shift in the understanding of AD, strengthening the “outside-inside” hypothesis, which postulates that AD is primarily a disease of the skin barrier. This is in contrast to the long-prevailing “inside-outside” hypothesis, which maintains that AD is primarily a disease of immune dysregulation. However, a skin barrier defect alone may not be sufficient for the development of AD. Though allergen entry and sensitization occur through a defective epidermal barrier, a favorable Th2 milieu already present in genetically predisposed individuals is also required for typical lesions of AD to develop. Indeed, as discussed earlier, candidate-gene studies and GWAS results of AD have implicated several immune-response genes that have the capability to alter the Th2 milieu. However, there is still considerable debate about whether initial skin barrier defects lead to secondary sensitization through the impaired skin or whether immune dysregulation is the first pathogenic mechanism.¹⁴⁴ Thus, AD is considered as a disease of both immune dysfunction and skin barrier disruption.¹⁴⁵

Atopic Dermatitis and Race

Various studies have examined the effect of racial group on the prevalence of AD. Studies of different racial groups in the same environment are particularly helpful to assess the contribution of genetic factors. Results from a 2003 U.S. national survey suggest that the prevalence of AD among individuals of African ancestry is 6 times higher than individuals of European ancestry,^9,146–148 affirming that AD places a high burden on the African American population.¹⁴⁸ Similarly, a study in the United Kingdom showed that Black children with AD have approximately 6 times higher risk of having severe AD than do White children.¹⁴⁶ In Australia, the 12-month cumulative incidence of AD is much higher in Chinese babies (44%) than in Caucasian babies living under similar environmental conditions (21%).¹⁴⁹ Compared with infants born to White mothers, infants born to Black mothers have an adjusted odds ratio (OR) for risk of AD of 2.4 (95% confidence interval [CI]: 1.47 vs. 3.94).¹⁵⁰ Though differences in socioeconomic status and environmental exposures between racial groups may contribute to prevalence differences, population genetic differences in skin barrier functions between the ancestral African, European, and Asian populations may be partly responsible for the loss of epithelial barrier functions in AD.

It is to be noted that the FLG mutations associated with AD in Europeans are different from those associated with AD in Chinese people, which again are different from those associated with AD in Japanese people. These findings suggest that different FLG mutations that essentially have the same effect are population specific and arose after the individual populations had separated from common ancestors. It has been speculated that the FLG mutations conferred survival advantage to these populations. In a recent report, Thyssen et al. hypothesized that FLG mutations are more prevalent in Northern Europeans, who receive relatively little amount of sunlight, in order to allow better penetration of ultraviolet-B rays through a defective skin barrier and facilitate the production of Vitamin D3 within the skin.¹⁵¹ Meta-analyses of studies involving thousands of European ancestry patients have confirmed FLG associations with an overall odds ratio for AD ranging from 3.12 to 4.78.¹⁵² Though FLG has been identified as a major locus causing skin barrier deficiency, not all patients with AD have mutations in this gene, so AD cannot be explained by FLG mutation alone. For example, though several common FLG mutations have been found to be associated with AD in Europeans and Asians in candidate-gene studies and GWAS, the search for common FLG mutations in native Africans has not been fruitful. The observation that filaggrin mutations play an important role in European American patients with AD but are absent in African patients, who also have reduced filaggrin protein levels in their skin lesions, suggests that both genetic and immune mechanisms can lead to skin barrier dysfunction.¹⁵⁶ Despite a prevalence of 3.2% for the FLG mutations common in Europeans with AD having also been demonstrated in African Americans with AD, this finding could be explained by European admixture, which can be as high as 20% in the African American population.^138,153,154 Interestingly, in a recent study, Margolis et al. have demonstrated the association of null mutations in FLG2, a gene belonging to the EDC and closely related to FLG, with persistent AD in African American subjects.¹⁵⁵ These mutations were not present (or were infrequently present) in persons of European origin. This finding suggests that in African Americans, FLG2 could be more influential than FLG in predisposing individuals to AD. Studies exploring the role of FLG2 in patients with AD need to be conducted in native African populations to confirm this finding.

Genetics of Allergic Rhinitis

Similar to asthma and other allergic disorders, AR is a disease with a complex etiology. Typical AR-related allergens include tree, grass, and weed pollens; animal dander; dust mites; mold; and cockroach.¹⁵⁷ The relationship between AR and asthma is very close, with AR often preceding the development of asthma. Although AR affects the upper airways and asthma the lower airways, the pathogenesis of both conditions involve airway epithelial cells, and the pattern of inflammatory response is very similar, giving rise to the concept of “one airway, one disease”.¹⁵⁸ Preschool children and young adults with untreated AR are at threefold increased risk of asthma later in life.^159,160 Evidence regarding the heritability of AR came from twin studies.¹⁶¹ Though MZ twins were 45–60% concordant, the concordance rate in DZ twins was not more than 25%, with the calculated heritability ranging between 33% and 91%.^44,161 Several genome-wide linkage scans have been performed for the AR phenotype in European and Asian populations, in which loci in chromosomes 2,3, 4, and 9 were implicated.¹⁶¹

Candidate-gene studies performed in Asian and European populations have also implicated numerous genes, many of them belonging to the inflammatory/immune response pathways. These include chemokines/chemokine receptors (SDAD1, CXCL9, CXCL10, CXCL11, RANTES), interleukins/interleukin receptors (IL-13, IL-18, IL-23R, IL-12RB1, IL-12RB2, IL-27), and CD14, among others.^161–164 Three GWAS have been performed specifically for the AR phenotype. The first one by Andiappan et al. identified suggestive associations with two SNPs on the MRPL4 and BCAP genes in a Chinese cohort; both genes have putative roles in inflammatory/immune response pathways.¹⁶⁵ However, none of these SNPs reached genome-wide significance. In the second GWAS meta-analysis of AR by Ramasamy et al., a SNP in a locus situated between chromosome 11 open reading frame 30 (C11orf30) and leucine-rich repeat containing 32 (LRRC32) attained genome-wide significance.¹⁶⁶ Interestingly, this locus was previously identified in both asthma and AD/eczema GWAS.^167,168 In a more recent study, Bunyavanich et al. integrated findings from GWAS, co-expression network, and expression SNP analysis in an attempt to test the biological context of GWAS findings and thus discover novel AR pathways.¹⁶⁹ Using the GWAS approach, they were able to identify 4 loci in Latinos and 1 locus in a combined meta-analysis involving European Americans, Latinos, and African Americans/African Caribbeans that reached genome-wide significance, and the integrative approach allowed them to demonstrate the importance of mitochondrial pathways in AR.

Allergic Rhinitis and Race

AR occurs in persons of all races. In the U.S., Asian, but not Black, subjects are more likely to report hay fever than are White subjects.¹⁷⁰ In addition, White children are more likely to have AR than are Black children, and white pregnant women are more likely to have AR than are black pregnant women.⁵ However, migrant studies suggest that environmental factors, rather than genetic ones, may be more important in explaining the apparent discrepancy in the prevalence of hay fever between races.¹⁷¹ In fact, studies that reported any difference in the prevalence of hay fever between races frequently acknowledged the possibility of bias due to unmeasured socio-economic factors, selective underreporting, or underdiagnoses as the driving factor behind the findings.^5,170 The major problem is defining AR consistently, as there are no consensus clinical/phenotypic/endotypic definitions to accurately apply for AR studies.^172,173

Genetics of Asthma

Although familial aggregation of asthma was demonstrated as early as the first half of the last century, twin studies performed in the later half showed a wide range in asthma heritability estimates, between 36% and 95%.¹⁷⁴ Starting from the 1990s, more than 25 genetic linkage studies of asthma have been undertaken to date, and the combined approach of linkage and positional cloning has been able to identify 9 asthma susceptibility genes (ADAM33, DPP10, PHF11, NPSR1, HLA-G, CYFIP2, IRAK3, COL6A5, and OPN3/CHML).⁴³ More than 100 loci for asthma and related phenotypes have also been identified through candidate-gene studies, with the majority of these genes being associated with the immune system (e.g., IL4RA, CD14, TLR2, and HLA class II genes).¹⁷⁴

Close to 40 GWAS in asthma have also been performed to date,¹⁷⁵ leading to the validation of several genes previously identified through positional cloning or candidate-gene studies (IL13, IL1RL1, DPP10, and HLA-DQ)¹⁷⁶ and to the discovery of several novel loci, the most replicated of which is the 17q21 locus. In 2007, Moffatt et al. were the first to identify the 17q21 region as the site of an asthma susceptibility locus.¹⁷⁷ The asthma-associated SNPs are part of a haplotype that extends across four genes—ORMDL3, GSDMB, ZPBP2, and IKZF3. Subsequent expression quantitative trait loci (eQTL) studies have shown that the haplotype has the ability to influence the expression of ORMDL3, GSDMB, and ZPBP2,¹⁷⁸ providing strong, functional evidence in favor of these genes in asthma pathogenesis. Most of these variants are in non-genic regions, and additional studies are required to determine whether any of the variants are directly causal through effects on regulatory elements or are in LD with causal variants.¹⁷⁹ Hence, it is still unclear which of these genes are responsible for the observed association with asthma and what role they play in asthma pathogenesis.

A major step in asthma gene discovery has been the formation of large consortia, which have brought together very large numbers of subjects in a meta-analysis. The Europe-based GABRIEL consortium comprised more than 26,000 subjects of European ancestry,¹⁸⁰ whereas the U.S.-based EVE consortium¹⁸¹ included more than 24,000 subjects of White, African American, and Hispanic ancestries. The most strongly associated genes in both studies and across different ethnicities were IL1RL1/IL18R1, TSLP, and IL33.⁴³ However, there are also inconsistent results between these studies, primarily due to clinical heterogeneity (different phenotype definitions), population stratification, and results without multiple-testing corrections. Over the last few years, there has been a strong focus on powering genetic studies with very large sample sizes; however, large cohorts might not help improve our understanding of the genetic underpinnings of atopic phenotypes as much as precise phenotyping.¹⁸² Thus, compiling large groups of patients with asthma without precise, careful phenotyping will not yield useful results, as demonstrated by the fact that asthma is a highly heterogeneous disease and the exact phenotype may be “lost in diagnosis”. Consequently, a shift has occurred from defining clinical asthma phenotypes to identifying mechanistically relevant endotypes.^183–185 For future studies, more efforts are needed to split these large resources into subgroups of patients sharing the same asthma phenotype. This strategy may impair statistical power due to decreased sample size but is likely to be enriched by patients sharing the same underlying molecular mechanisms. Thus, a small, homogeneous sample size is sufficient to detect enriched genetic effect in a more precisely selected group of patients.

Although the GWAS has identified several common risk variants for asthma, these variants are collectively responsible for only a small fraction of asthma heritability.¹⁸⁶ One source of this “missing heritability” (discrepancy between the predicted and observed genetic risk) could be rare or structural variants that contribute to disease risk but were not covered in early GWAS (newer GWAS platforms have a greater representation of these variants). Another reason for missing heritability could be gene-gene and gene-environment interactions. Though interactions between candidate genes have been demonstrated in asthma (e.g., interaction between IL13 and IL4RA among African American¹⁸⁷ and Dutch populations¹⁸⁸) gene-gene interaction studies of asthma, or any complex disease, at a genome-wide level will require very large sample sizes and highly sophisticated statistical methods that are still being developed. Environmental factors, such as the gut microbiome, diet, or exposure to traffic, cigarette smoke, and mold, might play a role in triggering allergic disorders by modulating gene expression via epigenetic pathways.

There have been several studies that have investigated gene-environment interactions in asthma, mostly in the candidate-gene context. The interaction between CD14 and levels of endotoxin exposure is an important example.¹⁸⁹ Ege et al. performed a genome-wide environment interaction study of asthma on a central European population for farm-related exposures. For example, the CD14(−260)C>T variant was associated with low total IgE in school children living in urban/suburban Tucson, Arizona, but the opposite association was reported in a farming community.^190,191 Although this study failed to show interaction of common SNPs or SNPs previously reported to have interaction effects, rare SNPs in potentially relevant pathways were identified.¹⁹² Other factors that may contribute to poor replication of GWAS results include poorly matched control groups, LD, genetic heterogeneity between populations, differences in phenotype/endotype definitions, inadequate evaluation of environmental factors, chance due to multiple testing, and publication bias of positive results.^193,194 The success of post-GWAS era studies in asthma will greatly depend on the ability to adequately quantify environmental exposures by standardizing sample collection, as well as generating more precise phenotype/endotype definitions. Novel genes of susceptibility to asthma were revealed only when relevant environmental exposures were considered. Therefore, it could be envisaged that detecting gene-environment interactions may help to target the preventive strategies for susceptible individuals. Finally, epigenetic mechanisms have been hypothesized to play an important role in the development and heritability of asthma and allergy.¹⁹⁵ The advent of next-generation sequencing holds great promise for asthma genetics, including identifying rare variants with large effect sizes that could account for some of the missing heritability.

Asthma and Race

Significant racial differences exist in asthma prevalence rates in the U.S. The National Center for Health Statistics (NCHS) reported average annual prevalence rates of asthma to be 19% in Puerto Ricans, 14% in Blacks, 9% in American Indians, 8% in Whites, 5% in Mexican Americans, and 5% in Asian Americans for the period of 2008–2010.¹⁹⁶ Although both Puerto Ricans and Mexican Americans are considered to be of Hispanic ethnicity, the former group has 12–15% Native American ancestry and 18–25% African ancestry, whereas the latter group has 35–64% Native American ancestry and only 3–5% African ancestry.¹⁹⁷ When taken in conjunction with the fact that African Americans have a high prevalence rate for asthma, it would appear that African ancestry is an important risk factor for the development of asthma.

Studies have also revealed that African Americans a higher rate of allergic sensitization than do Whites.¹¹ However, this association of African ancestry with asthma and allergic sensitization does not necessarily involve genetics factors since ancestry could also be a surrogate for unique environmental exposures. An analysis of NCHS data from the period of 2001–2010 by Akinbami et al.¹⁹⁸ showed that asthma prevalence substantially increased in Blacks (1.2 million in 2001 and 1.7 million in 2010), whereas it remained essentially the same in Whites (4.4 million in 2001 and 4.5 million in 2010). Since a change in genetic sequence at the population level is highly improbable in this short time-span, it is likely that environmental exposures, either acting alone or in combination with genes, are responsible for this selective rise in asthma prevalence among African Americans.

There is a dearth of well-designed genetic studies of asthma in ethnic minorities. A genome-wide linkage scan and candidate-gene studies performed by The Collaborative Study on the Genetics of Asthma (CSGA) included African American and Hispanic families, but the regions with the highest candidate genes and linkage signals in African Americans did not overlap with the regions of highest linkage in Whites and Hispanics.¹⁹⁹ Although the recent GWAS are robust compared to the many drawbacks of candidate-gene studies, the majority of the GWAS performed to date have focused on Whites, with relatively few including members of ethnic minority groups. Such studies have been limited by sample size, phenotypic and biologic heterogeneity, self-reported race, and inadequate data on environmental exposures and/or behavioral factors.

Results from the two largest meta-analyses of GWAS on asthma done to date, the GABRIEL and EVE consortia (discussed in the “Genetics of Asthma” section),^180,181 indicated that some loci are ancestry specific, including ORMDL3/GSDML in White and Hispanic ancestry and PYHIN1 in African American ancestry.⁴³ Recent, ethnic-specific associations of rare and low-frequency variants with asthma showed association of GRASP and GSDMB variants in the Latino ancestry and MTHFR variants in the African ancestry samples.²⁰⁰ Although GWAS have identified many candidate regions and genes, the underlying causes and genetic basis of asthma remains elusive. It is evident, however, that asthma is polygenetic and single-gene effects can account for only a small percent of the asthma phenotype.

Since GWAS chips typically provide good coverage for common variants and poor coverage for rare ones, a significant portion of the missing heritability in asthma could be due to rare and structural variants that are not included in a typical GWAS. It is also possible that rare and structural variants associated with asthma are more common in African Americans,³⁴ both qualitatively and quantitatively, leading to a higher disease rate. Gene-environment interaction is another potential explanation for the higher disease prevalence in African Americans. It is possible that disease-causing genes are expressed differentially depending on environmental exposures, which could in turn depend on specific ethnic backgrounds (i.e., ethnicity-specific epigenetic factors). Indeed, in a genome-wide transcriptional profiling study, it was demonstrated that children with asthma who belonged to lower socioeconomic strata overexpressed genes regulating inflammatory pathways compared with their counterparts who came from higher socioeconomic backgrounds.²⁰¹ Therefore, ethnicity can be a surrogate for environmental exposures that interact with the genome and lead to disease.

Patterns of ancestry differentiation and selection at loci associated with atopic disorders

Several population genetic studies have attempted to understand if population diferences of allergy and asthma risk alleles is present and identify the respective factors that may contribute to asthma disparity across racial ancestry. Most asthma risk alleles tend to show decreasing frequency along with human migration out of Africa, such that the African and East Asian populations carry the highest and lowest asthma genetic risk scores, respectively.²⁰² This frequency variation across present-day populations was likely shaped by random genetic drift, migration history, and natural selection, particularly influenced by pathogens and differing environments.^202–206 While in random genetic drift allele frequencies change over time purely due to chance, in natural selection, a deleterious allele is selected in preference to the normal allele since it confers resistance to another disease with a high fatality rate. This confers the heterozygotes with one copy of the harmful allele a survival advantage over the homozygotes with two copies of the normal allele. Several genes related to atopic disorders are thought to have been subject to this balanced selection process; here we discuss a few that show the strongest evidence of evolutionary footprints. The observed differences in allele frequency and population differentiation for disease- associated SNPs may partly explain the disparities in AD prevalence observed in AAs.⁶

Malaria is known to have been a dominant selective force in shaping evolution in the African people. Interleukin (IL)-4 is an anti-inflammatory cytokine produced by activated T cells (Th2) that regulates humoral as well as adaptive immunity. IL-4 589(C>T) (rs2243250), a SNP located in the promoter region of the IL-4 gene, regulates its activity and has been reported to be associated with elevated levels of total serum IgE (a marker for atopy),^207,208 and with asthma.^209,210 The Fulani people of West Africa have fewer attacks of malaria and a lower prevalence of P. falciparum infection when compared to their geographical neighbors.²¹¹ They also have higher levels of anti-malarial IgG and IgE levels.²¹² Luoni et al.²⁰¹ showed that the variant T allele of IL-4 589 (C>T) occurs at a much higher frequency in the Fulani when compared to the neighboring Mossi and Rimaibe people, and is also positively associated with anti-malarial antibodies in the Fulani population. In this case, positive selection of the variant T allele that is associated with allergy and asthma presumably gave the Fulani a survival advantage over malaria.^211–214

The variants Gln551Arg (rs1801275)²¹⁵ and Ile50Val (rs1805010)²¹⁶ of the IL-4 receptor (IL-4R) gene have been associated with atopy and asthma respectively, and the haplotype Ile50Arg551, which has been associated with “enhanced signaling” and increased IgE production, was more common in African-Americans than in European-Americans.²¹⁷ IL-13, another important Th2 cytokine gene has multiple polymorphisms, including C-1112T ((rs1800925, also known as C-1111T), associated with allergy. The frequency of this allele was 12% in Caucasians, 13% in Chinese and 48% in Africans.²¹⁸

It may be noted that the Th2 response—in which IL-4 plays a key role²⁰³—is likely to be advantageous in tropical regions, where the likelihood of helminthic infection is higher.²¹⁹ This exaggerated Th2 response, which evolved for parasite expulsion early in the development of mankind, predisposes individuals to asthma and allergy in modern societies of developed countries, where exposure to parasites is rare. This could be one of the reasons why people of African ancestry in western countries have higher prevalence and greater severity of allergic symptoms than natives of these host countries, indicating that African ancestry might be a risk factor for allergic disease.^219–221 Indeed, several candidate genes that encode for Th2-related molecules, such as IL-4 and IL-13, show a higher frequency in people with African ancestry of alleles that have a strong Th2- promoting activity,²¹⁹ suggesting that these alleles, which have been conserved to combat parasitic infections in Africa are leading to increased prevalence of atopic disorders in African origin people settled in western countries.

Two common and independent mutations in the FLG gene, namely R501Xand 2282del4, which have similar population frequencies, are present in about 10% of people of Northern European ancestry.^151,222 These mutations lead to an absence of filaggrin protein in the skin, which causes a barrier defect allowing entry of external allergens and subsequent sensitization, believed to be the starting point of AD and the atopic march. As Irvine and McLean²¹² point out, such a high prevalence of two very similar mutations in the European population is unlikely to be a result of random genetic drift but could rather be a consequence of balanced selection. They speculate that heterozygote carriers of these mutations had a survival advantage due to greater immunity to infectious diseases such as tuberculosis, influenza and plague that occurred as pandemics throughout known history, due to “natural vaccination” resulting from low level exposure to microbial antigens through a disrupted skin barrier.²²² Asian populations carry their own population specific FLG mutations at a prevalence rate of approximately 5%,¹⁵¹ and similar evolutionary advantage could well have been the cause. While FLG mutations have an extremely low prevalence in people of African origin, skin barrier defects in this population might be the result of mutations in a related gene, FLG2, as proposed by Margolis et al.¹⁵⁵ Another hypothesis attributes the higher prevalence of FLG mutations in Northern Europeans when compared to Southern Europeans, Asians and Africans to better penetration of sunlight through a defective skin barrier in sunlight deprived northern latitudes, which helps in the synthesis of Vitamin D3.¹⁵¹ This hypothesis better explains the observed North-South gradient in prevalence of FLG mutations, with northern regions in Europe, Africa and Asia having a higher relative prevalence compared to the southern regions.

Mapping Atopic Disorder Genes Contributing to Racial Differences

Gene discovery started in the late 1970s with a functional approach, in which the gene product and its function were used to identify a gene.²²³ Since then, several analytic approaches, such as linkage analysis, candidate-gene, genome-wide association (GWA), and admixture mapping (AM) studies, have been used to map genes related to atopic disorders (Figure 8). One approach to localize genes in which risk alleles are distributed differentially between racial ancestry groups within an admixed population is using admixture mapping. Admixed individuals are those who inherit chromosomal segments of distinct continental ancestry, in which ancestral genomes have diverged over time due to genetic drift and/or natural selection.²²⁴ African Americans are an example of admixed populations, with ~80% African ancestry and ~20% European ancestry.²²⁵ Mapping atopic disorder susceptibility genes in an admixed population using AM involves screening the genome of individuals of mixed chromosomal regions that have a higher frequency of alleles from the parental population with higher disease risk.²²⁶ African ancestry individuals exhibit an increased prevalence of AD, asthma, and FA compared to European Americans and AM could be an idea approach to map variants associated with atopic disorders.²²⁷ AM takes advantage of the high level of admixture created in populations of mixed ancestry by searching for associations between the ancestry score of a marker and disease, as opposed to GWAS which is between the genotype score of the marker and disease.²²⁸ The theory behind AM is that chromosomal segments of affected individuals contain a significantly higher than average proportion of alleles from the high-risk parental population and thus are more likely to harbor a disease-gene allele(s).²²⁸ Criteria to evaluate the applicability of AM in atopic disorder-related gene mapping include:²²⁹ (1) the prevalence of the disease differences in ancestral populations from which the admixed population was formed; (2) a measurable difference in disease-causing alleles between the parental populations; (3) reduced linkage disequilibrium between unlinked loci across chromosomes and strong linkage disequilibrium between neighboring loci; (4) a set of markers with noticeable allele-frequency differences between parental populations that contributes to the admixed population. Among atopic disorders, asthma clinical variables are relatively well studied for ancestry-related variations. A lower responsiveness to bronchodilators has been observed among individuals of African descent with asthma compared to Whites with asthma.^230,231 In African Americans, a decrease in lung function is associated with an increase in African ancestry proportion; adding genetically measured ancestry to the standard lung function prediction equations, rather than relying on self-identified race, reduced misclassification and resulted in reclassifying asthma severity by 5%.⁷ It is important to note that although ancestry is associated with asthma clinical phenotypes, socioeconomic status and related environmental exposure risk factors were not considered in most previous studies; thus, it is not clear whether ancestry is a surrogate for existing socio-environmental differences (i.e., may not be directly causal) or is an independent risk factor (serving as surrogate for genetic differences) for asthma risk. For example, if there is greater global ancestry noted across the genome in patients with asthma relative to controls, but no significant rise in local ancestry at a particular locus, this may point to a stronger role for environmental factors (e.g., exposures to traffic, home, cigarettes) independent of genetic ancestry.^232–234 Thus, a more careful assessment of the degree of ancestry, which is not captured by self-reported race/ethnicity alone, in larger cohorts while controlling for environmental exposure and other social determinants of health is needed to better understand the role of genetics, if any, in the observed racial differences in clinical asthma/allergy phenotypes. These insights are critical to develop intervention policies to reduce the burden of asthma due to non-genetic factors.²³⁵ Next-generation sequencing technologies now make it possible to genotype and measure hundreds of thousands of ancestry-specific markers across the genome, which have facilitated AM studies for discovery of asthma-associated genomic regions in admixed populations.²²⁸ AM has successfully been used to identify genomic regions harboring disease susceptibility loci for cardiovascular disease,²³⁶ multiple sclerosis,²³⁷ prostate cancer,²³⁸ serum IL-6 levels,²³⁹ and asthma in Mexican, Puerto Rican, and African American ancestry populations.^86,240–247 These results indicate that the admixed population provides an excellent opportunity to harness the power of LD represented by ancestral markers transmitted together, thereby making use of disease prevalence variation in the ancestral (founder) populations to map risk loci.

Figure 8. Atopic disorder variant discovery and follow-up strategies.

Methods ranging from genome-wide linkage, candidate gene, genome-wide association and admixture studies are presented. These gene mapping approaches yielded promising association results in the field of allergic diseases. However, association does not necessarily imply biological functionality, and follow-up studies are needed to translate initial findings into the biological insights that ultimately will advance prognostics, diagnostics and therapeutics.

Omics-based system biology approach to atopic disorders

Recent progress in high-throughput technology, including next-generation sequencing, has made it possible to quantitatively examine the genome, epigenome, transcriptome, proteome, metabolome and microbiome in different phenotypes, with the resultant generation of massive amounts of data, including in the allergy field.²⁴⁸ Dissection of atopic disorders using these -omics approaches has been attempted at least to some extent. For example, transcriptome analyses have identified genes differentially expressed between asthmatic and control subjects²⁴⁹, between lesional and nonlesional skin in AD,²⁵⁰ and between airway smooth muscles derived from patients with fatal asthma and healthy controls at baseline and after treatment with Vitamin D.²⁵¹ A recent study of transcriptome-wide responses to glucocorticoid treatment provides an interesting example of the use of transcriptomics to investigate the causes of population differences in allergic disease.²⁵² Proteomics has been used in several studies to identify biomarkers and peptide patterns associated with specific asthma endotypes.²⁵³ These include a proteomic study that measured cytokine levels in bronchoalveolar lavage to categorize asthma patients²⁵⁴ and a protein-protein interaction analysis of asthma to confirm the role of established asthma pathways and uncovered new candidate genes.²⁵⁵ More recent approaches include shotgun proteomics, which has been used to characterize protein patterns in induced sputum from asthmatic patients, including patients with exercise-induced asthma.²⁵⁶ While we are not aware of any studies exploring the use of proteomics to characterize the relationship between race and allergic disease, peptide profiles and protein signatures that are specifically associated with several different cancer forms in patients of African ancestry have been identified.^257,258 Similarly, we can expect proteomics to play a future role in explaining racial disparities and identifying race-specific biomarkers associated with allergic disease. MicroRNAs (miRNAs) are small noncoding RNAs that act mainly as suppressors of gene expression at the posttranscriptional level,^259–261 which may act as key regulators of the development and activity of innate and adaptive immune systems and control inflammatory processes. Several miRNAs have recently been associated with allergic airway diseases.^262–265 The epigenome has been interrogated to reveal micro-RNAs specifically targeting HLA-G alleles in patients at risk for asthma²⁶⁶ and reducing the intensity of chronic skin inflammation in patients with AD.²⁶⁷ Environmental triggers including tobacco smoke²⁶⁸ and traffic-related pollutants have been shown to induce epigenetic and microRNA expression changes in patients with allergic disease.^269–271 The impact of race on the role of epigenetics in allergic disease is still largely unexplored. A recent study showed that DNA methylation profiles of European American, African American, and Chinese American populations can be partially traced back to genetic variation.²⁷² A few studies have explored the relationship between smoking and DNA methylation in African-American cohorts. Dogan et al. performed genome-wide DNA methylation analysis in a cohort of African-American women, and found 910 loci that were significantly associated with smoking status.²⁷² Because epigenetic marks are biologically related to both environmental exposure and genes, they may provide a new explanation of gene-environment interactions.^273,274 A recent study showed that DNA methylation profiles of European American, African American, and Chinese American populations can be partially traced back to genetic variation.²⁷⁵ Genetic variants that influence the gene expression or methylation state include expression quantitative trait loci (eQTL) or methylation quantitative trait loci (meQTLs), are efficient approaches to identify functional SNPs regulating expression and methylation of allergy-associated genes.^{177,178,276–278} Information obtained from metabolomic studies of asthma have proved useful in distinguishing asthmatic from non-asthmatic subjects as well as in differentiating between different asthma subphenotypes.²⁷⁹ Several studies have examined associations between changes in microbiota and the development of AD, AR, and asthma, but far less have evaluated the impact of the microbiome on these outcomes. The skin or gut is colonized by numerous species of bacteria, fungi, and viruses that together are known as the microbiome.^280,281 It has been estimated that the human gut is populated with up to 100 trillion microbes.²⁸² This microbiota roughly contain on the order of 150-fold more genes than are encoded in the human genome.²⁸³ The human microbiome (which might interact with human genome) is a source of genetic diversity, a modifier of disease and an essential component of immunity.²⁸⁴ These microbes might interact with human genome. The potential impact of the microbiome on human health including respiratory and allergic diseases was examined in children.²⁸⁵ Children living in farming environments had a significantly decreased frequency of AD, AR and asthma compared with children living in urban areas.^286,287 This relationship was further explored in the GABRIELA and PARSIFAL cohorts, which confirmed previous observations that children living on farms had decreased rates of allergic disease compared with urban children.^288,289 Changes in lifestyle such as clean environment have led to a decrease in the infectious burden (e.g., reduce H. pylori) and are associated with a rise in allergic diseases also referred as the hygiene hypothesis.^290–293 As we advance our understanding of the diversity of microbiomes across geographical regions, time, individuals, and tissues within individuals, we become better positioned to take advantage of this emerging technology to inform the practice of precision medicine in atopic disorders.

Application of the systems biology approach, which is the integration of high-throughput data for the comprehensive profiling of disease phenotypes, to allergic diseases will provide greater insight into disease pathology and pathogenesis, provide a better understanding of cell to cell signaling networks for more efficient treatment, and also help to better predict disease outcome.²⁹⁴ To avoid getting lost in the sea of “big data”, advanced computational and mathematical based-information theory models such as machine learning, dynamic Bayesian network models, and differential equation–based models are required to provide a deep insight and understanding of the complex and non-linear relationships within and between biological systems. The growing abundance of high volume genetic/molecular data in the field of allergy has prepared the ground for information theory to be used for a better understanding of the flow of biological information from the genotype to the eventual allergic phenotype through mathematical models of information exchange.^295,296 For example, the machine learning approach uses computer algorithms to identify patterns from the systematically collected large molecular profile data, and along with clinical metadata, can assist personalized treatments for effective management of atopic disorders with similar molecular subtypes.²⁹⁷ Thus, research toward a systems-level understanding of allergic diseases will provide clinically valuable insight into atopic disorders pathogenesis.²⁹⁸

In critically evaluating existing literature, compared to other specialties, notably the field of cancer research, the application of systems biology to asthma and allergy has been limited, and there are even fewer reports describing the systems biology approach to characterize racial differences and elucidate their role in allergic disease. The increasing focus on endotyping of allergic diseases, however, can be expected to prompt efforts to harness the power of systems biology as an approach to improve endotypic stratification. Systems biology can also be expected to play an important role in the development of personalized medicine approaches that takes racial ancestry into account. It needs to be stressed, however, that any racial differences identified using proteomics, transciptomics, or epigenetics can be expected to be modified or confounded by environmental factors associated with ancestry, given the profound influence of the environment on epigenetics and gene regulation. For an in-depth coverage of the systems biology approach in asthma and other allergic conditions, we refer the reader to a few excellent review articles published in the recent past.^{279,294,297,299,300}

Conclusion

Atopic disorders form a cluster of phenotypes that include AD, FA, AR, and asthma. The global prevalence of these disorders, which have strong genetic and environmental components, has increased steadily in recent decades. Increased awareness of atopic diseases, changes in environmental exposures, epigenetic/genetic factors, and possible immune imbalance could contribute to this substantial increase in prevalence. Indeed, environmental exposures such as smoking, air pollution, and stress have been shown to cause changes in epigenetic modifications of genes, as well as altered microRNA expression.³⁰¹ Although racial disparities have been noted in the prevalence of atopic conditions, determining the relative contribution of ancestry-specific genetic risk factors from environmental factors has proved to be challenging because of the limited number of studies performed on minority populations.¹⁴ Shared and unique genes/pathways involved in the “atopic march” also need to be better defined. The delineation and deconstruction of shared and unique biologic and genetic pathways among atopic disorders and ancestry-specific gene-environment interactions can help resolve the clinical complexity and better inform the development of novel therapies.

Although the concept of the atopic march has led to the study of overlapping genetic susceptibility loci and pathways as possible causal links among atopic disorders, the co-existence and excess comorbidity of several atopic disorders in a single individual also make it harder to distinguish between shared and disorder-specific variants.³⁰² As a result, many questions remain regarding the underlying mechanisms and causative models that attempt to connect atopic disorders and ancestry. As we move forward, multi-ethnic birth cohorts with deep phenotyping and longitudinal phenotypes are needed to investigate the course of atopic march progression (i.e., whether early life food sensitization leads to other allergic disease later in life) as well as to identify individuals (or ancestries) at high risk of these comorbidities. In this comprehensive review, we have provided up-to-date information on atopic disorders with particular emphasis on genetic components and racial variations. Future research should include deep phenotyping of diverse ancestral populations, better characterization of environmental determinants, and the application of new technologies utilizing “-omics” tools. These include next-generation sequencing, epigenetics, and eQTL approaches in appropriate tissues/cells (Figure 5) along with publicly available bioinformatics and ancestry tools (Table 3). For example, recent findings investigating eosinophilic esophagitis revealed that tissue specificity was important in the expression of genes harboring the eosinophilic esophagitis–associated variants.⁶⁰ This provides a paradigm shift for focusing on tissue–specific predisposition that may apply to other disorders, which may help to explain gaps in our understanding of the “atopic march” theory.^60,303 Systematic integration of “big data” coming from providers (e.g., EMR), from omics (e.g., genomic, proteomic, epigenomic, metabolomic), and from multi-ethnic patients and non-providers (e.g., smart phone, monitoring tools for environmental triggers) can thus provide valuable insights to resolve the clinical complexity and ancestry-specific (or shared) etiology of atopic disorders.

Table 3.

Web-based, publicly available and easily accessible genomic/epigenomic resources relevant to atopic disorders

Database/Tool	URL address
Population databases
HapMap	http://hapmap.ncbi.nlm.nih.gov/
1000 Genomes Project	http://www.1000genomes.org
ENCODE project	http://genome.ucsc.edu/ENCODE
NHGRI GWAS catalogue	www.genome.gov/gwastudies
dbGaP	http://www.ncbi.nlm.nih.gov/gap
Exome Variant Server	http://evs.gs.washington.edu/EVS/
Genevar	http://www.sanger.ac.uk/resources/software/genevar/
ClinVar	http://www.ncbi.nlm.nih.gov/clinvar/
dbSNP	http://www.ncbi.nlm.nih.gov/snp
dbVar	http://www.ncbi.nlm.nih.gov/dbvar
OMIM	http://www.omim.org
RefSeqGene	http://www.ncbi.nlm.nih.gov/refseq/rsg
MitoMap	http://www.mitomap.org/MITOMAP/HumanMitoSeq
SNPedia	http://www.snpedia.com
IGV for iPad	http://www.broadinstitute.org/igv/iPadLaunch
PhenX	https://www.phenxtoolkit.org
Genome/Epigenome browsers
UCSC Genome Browser	http://genome.ucsc.edu/
WashU Epigenome Browser	http://epigenomegateway.wustl.edu/
The Ensembl Genome Browser	http://www.ensembl.org/index.html
VEGA Genome Browser	http://vega.sanger.ac.uk/index.html
Blood eQTL browser	http://genenetwork.nl/bloodeqtlbrowser/
Functional genomics/Gene expression
HaploReg	http://www.broadinstitute.org/haploreg
RegulomeDB	http://regulome.stanford.edu
FunSeq2	http://funseq2.gersteinlab.org/
GWAS3D	http://jjwanglab.org/gwas3d
Gene Ontology	www.geneontology.org
GATK	https://www.broadinstitute.org/gatk/
miRWalk	http://zmf.umm.uni-heidelberg.de
Sherlock	http://sherlock.ucsf.edu/submit.html
UniProt	www.uniprot.org
GEO	http://www.ncbi.nlm.nih.gov/geo
ArrayExpress	http://www.ebi.ac.uk/microarray-as/ae/
Microbiome	http://www.hmpdacc.org/
Metabolome	http://www.hmdb.ca/
Human Gene Mutation Database	http://www.hgmd.org
In silico functional predictive tools
PFAM	http://pfam.sanger.ac.uk
PolyPhen-2	http://genetics.bwh.harvard.edu/pph2/
SIFT	http://sift.jcvi.org
SIFT BLink	http://sift.jcvi.org/www/SIFT_BLink_submit.html
SMART	http://smart.embl.de
SNPeffect	http://snpeffect.switchlab.org
Single-gene/variant browser
AncestrySNPminer	www.cchmc.org/mershalab/AncestrySNPminer
SPSmart	http://spsmart.cesga.es/hapmap.php?dataSet=hapmap
PupaSuite	http://pupasuite.bioinfo.cipf.es
SNPper	http://snpper.chip.org/bio/snpper-enter
GVS	http://gvs.gs.washington.edu/GVS
SNAP	https://www.broadinstitute.org/mpg/snap/ldsearchpw.php

Supplementary Material

1

Table E1: Genes ranked by level of association with each atopic disorder based on mining of 15.73 million PubMed abstracts using Literature Lab™ from Acumenta Biotech Software.

Table E2: IPA summary of associated networks, molecular and cellular functions, diseases and disorders and canonical pathways for the 16 genes shared among FA, AD, AR and asthma.

Abbreviations

AD

atopic dermatitis

AM

admixture mapping

AIMs

ancestry informative markers

AR

allergic rhinitis

DZ

dizygotic

EDC

epidermal differentiation complex

EMR

Electronic medical record

ETS

environnemental tobacco smoke

FA

food allergy

FLG

Filaggrin

GABRIELA

Multidisciplinary study to identify the genetic and environmental causes of asthma in the european community

GALA

genetics of asthma in latin Americans

GWEI

genome-wide environmental interaction

GWAS

genome wide association study

HapMap

haplotype map

IgE

Immunoglobulin E

ISAAC

international study of asthma and allergies in children

LABAs

long-acting β2-agonists

LD

linkage disequilibrium

MAF

minor allele frequency

MZ

monozygotic

NCHS

national center for health statistics

PARSIFAL

Prevention of allergy – risk factors for sensitization related to farming and anthroposophic lifestyle

PUBMED

search engine for accessing the MEDLINE database of citations

SNPs

single nucleotide polymorphisms

tagSNPs

haplotype-tagging SNPs