Keywords

Introduction

Personalized medicine or precision medicine aims to use a subject’s characteristics to design an individualized treatment plan. Personalized medicine is based on the premise that biomarkers (e.g., genetic variants) can be used to predict disease risk or response to medications, in order to prevent or treat a disease in a given individual.

Genetic and “omics” studies of respiratory diseases, both published and ongoing, will lead the way to predictive profiles for precision medicine. This chapter will focus on asthma and chronic obstructive pulmonary disease (COPD), not only because of their public health importance (see Chap. 10) but also because of the strength of the evidence to support personalized medicine to prevent and treat these common airway diseases.

The frequency and severity of asthma and COPD differ among racial and ethnic groups in the USA (see Chaps. 2 and 10). In this chapter, we discuss the basis for the variable population structure and genetic diversity of modern human genomes from different racial and ethnic groups. We will summarize how such diversity has impacted genetic studies, and how studies in diverse populations have led to the identification of susceptibility loci for respiratory diseases and response to treatment. Finally, we highlight how future genetics and “omics” research in diverse populations should lead to identification of biomarkers for personalized medicine, which would help eliminate existing respiratory health disparities.

Genetic Studies in Ethnically Diverse Populations

According the US Census Bureau, the non-Hispanic White population will peak in 2012, and then slowly decrease in size from 2024 to 2060. In contrast, Hispanic and Asian populations will grow over the next four decades, making non-whites surpass non-Hispanic whites as the majority of the US population by 2060 [1].

The diverse genomes of modern racial or ethnic groups in the USA (see Chap. 2) largely resulted from racial admixture during the European colonization of the Americas over the past 500 years. Thus, African Americans and Hispanics (e.g., Puerto Ricans and Mexican Americans) have varying proportions of European, Native American, and African ancestries. On average, African Americans and Puerto Ricans have a greater proportion of African ancestry but a lower proportion of Native American ancestry than Mexican Americans. However, ancestral proportions can vary between members of an ethnic group [2, 3].

Because of human origins in Africa, subjects of sub-Saharan African descent have had a greater number of recombination events over many generations, resulting in greater genetic diversity and fewer co-inherited polymorphisms within genomic regions (i.e., shorter regions of linkage disequilibrium [LD], Fig. 13.1). In contrast, Europeans had loss of genetic diversity during a “bottleneck” as the first modern humans migrated to Europe from sub-Saharan Africa ~40,000 years ago, resulting in high correlation or co-inheritance of polymorphisms within genomic regions (i.e., greater LD, Fig. 13.1) [4]. Because European ancestry leads to lower genetic diversity but greater LD than African ancestry, fewer markers need to be genotyped to “tag” genetic variants in populations of mostly European descent (i.e., European Americans or non-Hispanic whites) than in those of mostly African descent (i.e., African Americans). For the same reasons, rare variants (allele frequency <0.05) are more frequently found in African Americans than in European Americans [4, 5].

Fig. 13.1
figure 1

Consequences of ancestral admixture on genetic diversity. The recent admixture of an ancient ancestry (such as African ancestry or ancestry A, highlighted in green) with a more recent ancestry (such as European ancestry or ancestry B, highlighted in blue) affects genetic diversity in chromosomal (Chr.) regions throughout the genome. The more ancient ancestry (A) has had a greater number of recombination events over more generations, resulting in greater genetic diversity with fewer co-inherited genetic variants in genomic regions, highlighted with the red triangles signifying shorter regions of linkage disequilibrium (LD). In comparison, ancestry B resulted from a recent loss of genetic diversity or “bottleneck,” leading to gene variants that are highly co-inherited or correlated over longer genomic regions through LD (highlighted with larger red triangles). The more ancient ancestry A has also had more time for rare variants to occur (red “R”) and has a higher frequency of rare variants compared to ancestry B

High-throughput genotyping allows for the analysis of millions of single nucleotide polymorphisms (SNPs), which have been used in genetic association studies of airway diseases. Such studies have targeted biologically plausible candidate genes or the whole genome (genome-wide association studies or GWAS), most often—but not exclusively—in populations of European descent. More recently, next-generation DNA sequencing has expanded the catalogue of human genetic diversity, facilitating studies of ethnically diverse populations. We will next highlight salient findings from genetic studies of asthma and COPD.

Genetic Studies of Asthma

Early family-based genome-wide linkage studies failed to identify susceptibility genes for asthma or related phenotypes, yet demonstrated that asthma is caused by multiple genes [617]. More than 100 genes have been examined for association with asthma, based on biologic plausibility (“functional candidate genes”) or location in genomic regions linked to asthma or related phenotypes in family-based studies (“positional candidate genes”) [18, 19]. Such studies were largely based on genotyping common SNPs (i.e., allelic frequency ≥0.05) in the genes of interest, which were then tested for association with asthma. Using this candidate-gene approach, the most highly replicated genes for asthma in subjects of European descent were in biologic pathways related to lung development (ADAM33), Th2 inflammation (IL4, IL13, IL4R), innate immunity (HLA-DRB1, HLA-DQB1, CD14), and cellular inflammation (TNF, FCER1B, DPP10) [1828]. Consistent with findings for many candidate genes, ADAM33 was associated with asthma or related phenotypes in African Americans and New Mexico Hispanics [29], but not in Puerto Ricans or Mexicans [30]. The inconsistent findings for most candidate-gene association studies of asthma could often be due to false positive results from chance or population stratification (confounding by population substructure). Alternatively, nonreplication across ethnic groups may have been due to limited statistical power because of small sample size or ethnic-specific genetic effects (due to differing allelic frequencies (Table 13.1 [3134]) or gene-by-environment interactions) [29, 35].

Table 13.1 Allele frequencies of asthma risk loci in different racial and ethnic groups

The first GWAS of asthma susceptibility (conducted in Europeans) identified a novel locus on chromosome 17q21 (containing ORMDL3 and GSDMB), which has been well replicated across multiple racial or ethnic groups [3644]. ORMDL3 encodes a transmembrane protein anchored to the endoplasmic reticulum, but its role (or that of GSDMB) in asthma is unclear. Subsequent GWAS have identified additional asthma-susceptibility loci, notably including genes (IL33, IL1RL1, and TSLP) conferring susceptibility to asthma in ethnically diverse North American populations (non-Hispanic whites, African Americans, Afro-Caribbeans, Puerto Ricans, and Mexicans) [28, 37, 42, 45]. Many of these genes are involved in pathways related to epithelial integrity and adaptive immune responses, suggesting that they promote TH2-mediated airway inflammation through altered production of cytokines (i.e., TSLP and IL33) and/or damage of the airway epithelium.

The first GWAS conducted primarily in subjects of African descent identified the genes for the α-1B-adrenergic receptor (ADRA1B) and prion-related protein (PRNP) as novel asthma-susceptibility loci, while also replicating findings for DPP10 from an earlier study of Europeans [11, 18, 4648]. A multiethnic GWAS in North America (see above) also identified a gene (PYHIN1) that appears to only confer susceptibility to asthma in subjects of African descent [42].

Severe asthma (characterized by baseline airflow obstruction, uncontrolled symptoms, or frequent exacerbations despite adequate treatment) occurs more frequently among African Americans and Puerto Ricans [49]. Emerging evidence suggests that genes that influence asthma severity differ from those that determine asthma susceptibility [50, 51]. Identifying such loci should thus help to discover mechanisms underlying interethnic differences in asthma severity [5255].

Lung function is an indicator of asthma severity. SNPs in the gene encoding the hedgehog-interacting protein (HHIP) are associated with reduced lung function in African American and non-Hispanic whites with asthma in the Severe Asthma Research Program (SARP) [56]. Moreover, variants in HHIP and genes previously associated with lung function in the general population (FAM13A and PTCH1) had additive effects on lung function and asthma severity in non-Hispanic whites and African Americans with asthma (Fig. 13.2a, b) [3, 56, 57].

Fig. 13.2
figure 2

(a, b): Additive genetic effects of major lung function loci from the general population. An additive effect of risk variants from GWAS of lung function measures in the general population on lung function was shown in 1441 asthma subjects from three asthma cohorts. An increase in the number of risk alleles for SNPs in HHIP, PTCH1, and FAM13A resulted in a significant stepwise decrease in FEV1, FVC, and FEV1/FVC in both non-Hispanic Whites and African Americans. Data for FEV1 is shown for (a) African Americans and (b) non-Hispanic Whites. Adapted from Li X et al. J Allergy Clin Immunol 2011;127(6):1457–65 [56]

SNPs from whole-genome genotyping can be used to estimate whole-genome or global genetic ancestry [58]. As reviewed in Chap. 2, global African genetic ancestry has been associated with an increased risk of asthma, lower FEV1, and lower FVC in African Americans and Puerto Ricans [3, 57, 5961]. Conversely, global Native American ancestry has been associated with reduced risk of asthma but higher FEV1 and FVC in Latinos [60]. Of interest, global African ancestry has also been shown to be associated with severe asthma exacerbations in African American males, further suggesting a role for genetic or environmental risk factors correlated with African ancestry on asthma severity [62].

Admixture mapping (AM) is a whole-genome scanning approach that can be used to identify susceptibility loci for complex diseases in racially admixed populations. AM tests for association between local ancestry at each SNP locus and phenotype, under the assumption of significant differences in both disease prevalence and allelic variation between ancestral groups for a population of interest (see Chap. 2) [63, 64]. An AM study identified an AM peak for asthma in African Americans on chromosome 6q14, which contained an SNP surrounded by a European ancestral background. This finding was replicated in Puerto Ricans, and an interaction between a susceptibility locus and local genetic ancestry was shown in both African Americans and Puerto Ricans [65]. In another study, six AM peaks containing known asthma risk loci and a novel locus in the LYN gene were identified in Puerto Ricans and Mexicans [66, 67]. More recently, AM was combined with allelic association testing to identify a potential asthma-susceptibility locus (PSORS1C1) in Latinos [44]. Like genome-wide linkage studies, AM has limited resolution, thus requiring subsequent fine-mapping studies, replication in external cohorts, and functional studies to confidently identify disease-susceptibility variants.

Failure to replicate findings from studies of non-Hispanic whites in minority populations (or vice versa) may be explained by differing allelic frequencies among ancestral populations (Table 13.1), insufficient genomic coverage in populations of African descent, small sample size, or true ethnic-specific effects. Current evidence suggests that most asthma-susceptibility loci identified to date are “cosmopolitan” (affecting all racial/ethnic groups), but that a few such loci may be “ethnic-specific” (affecting one or a few ethnic groups). Pending additional work, however, no ethnic-specific asthma-susceptibility variant has been confidently identified to date.

Genetic Studies of COPD

COPD is a multifactorial disease, caused by the interaction between genetic variants and environmental risk factors such as tobacco smoke [6872]. Candidate-gene association studies, largely conducted in non-Hispanic whites, have identified a small number of potential COPD-susceptibility loci in inflammatory (TGFB1), protease-anti-protease (ADAM33, MMP12), and oxidant-antioxidant (GSTM1, GSTP1) gene pathways, some of which (e.g., MMP12) have been confirmed in subsequent genome-wide scans [7278].

Airflow obstruction and lung function decline are key intermediate phenotypes of COPD. GWASs in non-Hispanic whites have identified susceptibility genes for lung function (GSTO2 and IL6R) [79] and airflow obstruction (seven loci, including HHIP) [8082]. Subsequent studies confirmed a role of HHIP in COPD susceptibility and showed an association with airflow obstruction in asthma [56, 8386]. Evidence from murine models suggests that HHIP variants alter lung development and baseline respiratory reserve, ultimately increasing the risk of lung function decline and COPD [8789]. Subsequent meta-analyses of GWASs have identified additional susceptibility loci for lung function in genes which regulate inflammation (HTR4, THS4D), lung development (ADAM19, GPR126), the antioxidant pathway (GSTCD), and tissue remodeling (ADAM19, HTR4, THSD4, AGER, TNS1). Of interest, alleles in these genes have differing frequencies across racial or ethnic groups (Table 13.2 [90]) [81, 82, 91, 92]. Cumulative risk scores combining risk SNPs from lung function genes (including HHIP, TNS1, GSTCD, HTR4, AGER, and THSD4) have been shown to be associated with a stepwise decrement in FEV1 and FEV1/FVC, both in non-Hispanic whites and African Americans with asthma, and in non-Hispanic whites from a general population cohort (Fig. 13.2a, b) [56, 85].

Table 13.2 Allele frequencies of loci from major lung function genes from general population GWAS by biologic pathway

Although global African ancestry has been shown to be inversely associated with FEV1 and FVC in African Americans and Latinos (see Chap. 2), there has been no GWAS of lung function in these populations. An admixture-based genetic study of Mestizo individuals and Native Mexicans demonstrated a strong correlation between Native American ancestry and geographic location within Mexico, resulting in ancestral clusters and ancestry-specific principal components. An analysis of ancestry-specific principal components in Mexicans and Mexican Americans then demonstrated that increased regional variation in Native American ancestry was positively associated with lung function [93]. Native American ancestry was also shown to be positively associated with lung function in a Costa Rican cohort of adolescents and adults with and without COPD [94], as well as in a study of Hispanic adults from New Mexico [95]. In the latter study, Native American ancestry was also inversely associated with lung function decline and COPD [95].

To date, GWASs of COPD have been conducted mostly in subjects of European descent. The first such GWAS identified susceptibility SNPs for COPD chromosome 15q25, a genomic region encompassing genes encoding the α-nicotinic acetylcholine receptors (CHRNA3/5) and a gene in the antioxidant pathway (IREB2) [86, 9698]. A second GWAS identified susceptibility SNPs for COPD in FAM13A, a gene previously associated with lung function [81, 99]. A subsequent (and larger) GWAS identified a COPD-susceptibility locus containing CYP2A6, a key enzyme for nicotine metabolism in the nicotine dependence pathway [100]. The first GWAS to include both non-Hispanic whites and African Americans confirmed risk loci for COPD in CHRNA3, FAM13A, and HHIP, while also identifying a novel locus in RIN3 (distantly adjacent to SERPINA1, which encodes α-1 antitrypsin, the strongest known genetic risk factor for COPD) [101]. A subsequent genome-wide admixture mapping in African Americans identified a novel locus for airflow obstruction (FAM19A2) [102]. The first GWAS of COPD in Hispanics (a meta-analysis of three cohorts in Costa Rica, the US Multi-ethnic Study of Atherosclerosis, and New Mexico) identified potential novel COPD-susceptibility loci (adjacent to KLHL7/NUPL2, and DLG2) and confirmed FAM13A as a locus for COPD in Hispanics [103].

Consistent with results in asthma, ancestry influences lung function and COPD in racially admixed populations. Moreover, most COPD-susceptibility loci appear to be “cosmopolitan,” but a few may be truly ethnic-specific. Thus, inclusion of large cohorts of minorities in COPD studies may yield novel insights into the role of ancestry and genetics in ethnic differences in the prevalence and severity of COPD.

Pharmacogenetic Studies of Respiratory Diseases

Pharmacologic responses have been shown to have both interindividual variability and significant heritability [104, 105]. Pharmacogenetic studies, which analyze gene-by-drug interactions on clinical outcomes, have been highly successful in identifying targeted therapies in cystic fibrosis. Most pharmacogenetic studies in pulmonary medicine have been conducted in subjects with asthma [106, 107] and have included mostly non-Hispanic whites. However, pharmacogenetic studies of response to inhaled β2-adrenergic receptor agonists (inhaled β2-agonists) have included racial and ethnic minorities with asthma.

Inhaled β2-agonists include short-acting β2-agonists (SABA, used most often as rescue therapy) and long-acting β2-agonists (LABA, often used in combination with an inhaled corticosteroid (ICS) for chronic treatment). Findings from surveillance studies and meta-analyses suggest that LABA increase the risk of life-threatening asthma exacerbations and asthma-related deaths when administered as a monotherapy without ICS therapy [108110]. The largest and most cited of these surveillance studies included 26,355 subjects (4685 African American), who were randomized to salmeterol or placebo with “usual therapy.” An interval analysis of that trial (SMART) demonstrated increased risk of asthma or respiratory-related life-threatening exacerbations and death among African Americans randomized to salmeterol [109]. Although limited by lack of a requirement for ICS in all study subjects, such findings formed the basis for a LABA safety controversy, leading to two advisory panel meetings by the US Food and Drug Administration (FDA), public health advisories, and a boxed warning for all inhalers containing LABA [111]. This controversy, further fueled by findings contradicting those from SMART [112115], is now being evaluated in an international FDA-mandated LABA safety study of over 40,000 asthmatics [111, 116].

Results from recent clinical trials suggest that African Americans with asthma have a reduced response to LABA-containing combination therapies compared to non-Hispanic Whites and Hispanics [117, 118]. In a study of adults with asthma, African Americans were less likely to respond to LABA than non-Hispanic Whites [117]. Similarly, Puerto Ricans with asthma have been shown to have a lower response to SABA than Mexicans, a finding that could be explained by ethnic-specific differences in genetic variants or response to psychosocial stress [119, 120].

Pharmacogenetic studies of response to inhaled β2-agonists have focused on the gene encoding the β2-adrenergic receptor (ADRB2), the pharmacologic target for β2-agonists. Although the most extensively studied ADRB2 variant is a coding SNP which substitutes a glycine for an arginine in amino acid position 16 (Gly16Arg), up to 49 SNPs have been identified through DNA sequencing in multiethnic populations, including rare variants [121, 122]. In vitro studies have shown that beta agonist stimulation results in enhanced downregulation of the β2-adrenergic receptor with the Gly16 allele compared to Arg16 [123, 124]. In vitro studies of the rare ADRB2 variant, Thr164Ile, show that this variant causes a marked decrease in receptor ligand binding and coupling to Gs protein in response to different SABAs and LABAs, and impaired salmeterol binding to its receptor “exosite” [125, 126].

Early association studies of ADRB2 in non-Hispanic whites consistently demonstrated that Arg16 homozygotes show greater response to SABA than Gly16 homozygotes, a finding confirmed in some ethnic groups (i.e., Puerto Ricans) but not in others (i.e., Mexican Americans) [127, 128]. An SNP in a pathway-related gene (GSNOR, encoding S-nitroso-glutathione reductase) has been shown to alter the genetic effect of Gly16Arg on response to SABA in Puerto Ricans but not in Mexican Americans, and thus this gene-gene interaction requires further replication [129].

Two pharmacogenetic studies using data from previous clinical trials (which randomized non-Hispanic whites with asthma to long-term treatment with SABA) demonstrated that ADRB2 Arg16 homozygotes were more likely to have a decline in lung function during SABA treatment than ADRB2 Gly16 homozygotes [130, 131]. The genetic effects of the Gly16Arg locus were confirmed in a genotype-stratified, cross-over pharmacogenetic study, the Beta Agonist Response by Genotype (BARGE) trial. In that trial, 37 Arg16 homozygotes and 41 Gly16 homozygotes were randomized to regular albuterol or placebo for 16 weeks, with ipratropium provided as a rescue inhaler to minimize β2-agonist use. Whereas Gly16 homozygotes had improved lung function and symptom control during regular albuterol therapy, Arg16 homozygotes had no change in lung function and a loss of symptom control during regular albuterol therapy [132].

In the BARGE trial, the proportion of Arg16 homozygotes was higher in African Americans (22 %) than in non-Hispanic whites (17 %) (see ADRB2 in Fig. 13.3) [132]. This is likely because Gly16 is the ancestral allele of Gly16Arg, and chromosomes from ancient African ancestors have had more generations to distribute the more recent Arg16 variant than chromosomes from a European ancestor [133]. The frequency of the Arg16 allele is thus higher in African Americans and Puerto Ricans than in non-Hispanic Whites [122], potentially explaining ethnic differences in response to β2-agonists. However, findings from genotype-stratified clinical trials have largely failed to show an effect of the Arg16 allele on response to LABA, with or without concurrent therapy with ICS [134138]. Gly16Arg (which has a frequency between 40 % and 60 % in different ethnic groups [ADRB2, Fig. 13.3]) should not be used to stratify patients for LABA treatment. Rare genetic variants with strong effects could explain the severe adverse effects found in <1 % of the LABA-treated subjects in SMART, but this is highly speculative [109].

Fig. 13.3
figure 3

Allele frequencies of different pharmacogenetic loci by race or ethnic group. These pharmacogenetic loci have been identified through candidate-gene studies and GWAS in asthma clinical trials [141, 143153]. Loci with SNPs identified through GWAS are highlighted in red. Allele frequencies are provided for each ethnic group and ancestral population based on the International HapMap Project Genome Browser release 28, phases 1–3 [32]. ASW African Americans from the southwest United States, CEU Utah residents with ancestry from northern and western Europe, CHB Han Chinese from Beijing, China, JPT Japanese from Tokyo, Japan, MEX Mexican Americans from Los Angeles, CA, YRI Individuals from Yoruba in Ibadan, Nigeria [33]

Sequencing of ADRB2 in different ethnic groups has identified Thr164Ile, a rare ADRB2 variant primarily found in non-Hispanic Whites, and a rare 25 base-pair insertion variant at nucleotide position −376 relative to the start codon in the ADRB2 promoter (−376 In-Del) in African Americans and Puerto Ricans [121, 122, 139]. In a recent study, these rare variants were both associated with asthma-related hospitalizations, asthma-related urgent outpatient visits, and regular use of systemic corticosteroids among non-Hispanic whites and African Americans with asthma treated with LABA (Fig. 13.4a, b) [122]. In another study, an analysis combining results from AM and a GWAS showed an association between rare variants in two solute carrier genes (SLC24A4 and SLC22A15) and response to SABA in Puerto Ricans and Mexican American with asthma [140]. In that study, rare variants in two genes previously associated with response to SABA in non-Hispanic whites, ADCY9 and CRHR2, were shown to be associated with response to SABA in Latinos [140142].

Fig. 13.4
figure 4

(a, b): Rare ADRB2 variants and asthma-related hospitalization with long-acting beta agonist treatment. Two rare ADRB2 variants, Thr164Ile and a 25 base-pair promoter insertion-deletion (−376 In-Del), are shown with odds ratio (OR) for hospitalization in those treated with a long-acting beta agonist (LABA). Reproduced from Ortega VE et al. Lancet Respir Med 2014;2(3):204–13 [122]

Large candidate-gene and GWAS of the pharmacogenetics of asthma have been mostly conducted in non-Hispanic whites. Such studies identified loci for therapeutic responsiveness to SABA, ICS, and leukotriene modifiers, each of which shows varying allele frequencies among different racial and ethnic groups (Fig. 13.3) [141, 143153]. More recently, large-scale whole-genome sequencing studies have found rare ethnic-specific variants in populations of African descent [5], suggesting that rare variants (such as those in ADRB2 and solute carrier genes) could be biomarkers for personalized treatment approaches in racial or ethnic minorities (e.g., avoiding inhaled LABA in nonresponders or in subjects likely to experience severe adverse effects).

Future Directions

Whereas most susceptibility alleles for respiratory diseases (or response to treatment for such diseases) are “cosmopolitan,” a small but non-negligible proportion of such susceptibility alleles are likely to be ethnic-specific (particularly rare variants). Moreover, differences in environmental and behavioral exposures across racial or ethnic groups are likely to affect gene expression through gene-by-environment interactions or epigenetic mechanisms that remain largely unexplored as potential contributors to respiratory health disparities.

A potential short-term implication of the studies summarized above (including studies of African ancestry and lung function summarized in Chap. 2) is that global ancestry (determined by genetic markers) could replace self-reported race or ethnicity when developing predicted (or reference) values for lung function. For instance, traditional race-based calculations of reference values for lung function can misclassify disease severity in up to 5 % of African Americans with asthma, as African Americans have different proportions of African ancestry [3].

In the medium to long term, predicted values of lung function could be personalized on the basis of whole-genome profiling, as all rare and common susceptibility genes for lung function (and other determinants, see below) become known as a result of large-scale studies of multiethnic populations. Similarly, such studies should lead to the identification of both common and rare (e.g., ethnic-specific) variants associated with response to, or severe adverse effects from specific therapies for pulmonary, critical care, and sleep disorders. In fact, ongoing whole-exome and whole-genome sequencing projects such as the NHLBI GO Exome Sequencing Program, the 1000 Human Genomes Project, and the Consortium on Asthma in African Ancestry Populations (CAAPA) have identified, and will continue to identify, common and rare genetic variation for future genetic studies in racial and ethnic minorities [5, 154].

The path to personalized medicine for all members of society requires enrollment of sufficiently large numbers of subjects from racial or ethnic minorities in studies of gene-by-environment interactions and “omics” (genetics, epigenetics, transcriptomics, proteomics, and metabolomics) of respiratory diseases, as such integrated approaches are more likely to yield novel insights into disease pathogenesis or pharmacogenetics than traditional genetic studies. Such inclusive and diverse studies should lead to personalized medicine for all people, a key step toward eliminating respiratory health disparities and achieving respiratory health equality in the USA.