Multi‐ancestry genome‐wide association study of asthma exacerbations

Abstract Background Asthma exacerbations are a serious public health concern due to high healthcare resource utilization, work/school productivity loss, impact on quality of life, and risk of mortality. The genetic basis of asthma exacerbations has been studied in several populations, but no prior study has performed a multi‐ancestry meta‐analysis of genome‐wide association studies (meta‐GWAS) for this trait. We aimed to identify common genetic loci associated with asthma exacerbations across diverse populations and to assess their functional role in regulating DNA methylation and gene expression. Methods A meta‐GWAS of asthma exacerbations in 4989 Europeans, 2181 Hispanics/Latinos, 1250 Singaporean Chinese, and 972 African Americans analyzed 9.6 million genetic variants. Suggestively associated variants (p ≤ 5 × 10−5) were assessed for replication in 36,477 European and 1078 non‐European asthma patients. Functional effects on DNA methylation were assessed in 595 Hispanic/Latino and African American asthma patients and in publicly available databases. The effect on gene expression was evaluated in silico. Results One hundred and twenty‐six independent variants were suggestively associated with asthma exacerbations in the discovery phase. Two variants independently replicated: rs12091010 located at vascular cell adhesion molecule‐1/exostosin like glycosyltransferase‐2 (VCAM1/EXTL2) (discovery: odds ratio (ORT allele) = 0.82, p = 9.05 × 10−6 and replication: ORT allele = 0.89, p = 5.35 × 10−3) and rs943126 from pantothenate kinase 1 (PANK1) (discovery: ORC allele = 0.85, p = 3.10 × 10−5 and replication: ORC allele = 0.89, p = 1.30 × 10−2). Both variants regulate gene expression of genes where they locate and DNA methylation levels of nearby genes in whole blood. Conclusions This multi‐ancestry study revealed novel suggestive regulatory loci for asthma exacerbations located in genomic regions participating in inflammation and host defense.


| INTRODUC TI ON
Asthma is a common chronic inflammatory airway disorder affecting over 300 million people worldwide. The disparities in asthma prevalence across populations reflect a complex interplay between environmental exposures (i.e., air pollution and viral infections), behavioral and socioeconomic factors (i.e., treatment adherence and healthcare access), and genetic ancestry, which is inferred from whole-genome variation and tracks geographic and historical factors and the aforementioned factors influencing asthma prevalence. 1,2 Asthma exacerbations are defined as worsening of respiratory symptoms requiring hospitalization, unscheduled/emergency asthma care, and/or use of systemic corticosteroids. 3 Prevention of asthma exacerbations is a major public health priority due to their associated consequences on health (i.e., decreased quality of life, accelerated decline in lung function, or mortality), school attendance, work productivity, and healthcare costs. 1,4,5 To date, the best predictor of future exacerbations is the occurrence of one in the previous year. 6 Thus, identifying potential biomarkers to guide the reduction and prevention of exacerbations is a priority for therapeutics development and for precision medicine of asthma. All funding agencies had no role in the study design, data collection, and analysis, decision to publish, or preparation of the manuscript. The views expressed are those of the authors and not necessarily those of the any funder, National Health Service (NHS), the National Institute for Health Research (NIHR) or the United Kingdom Department of Health.

Editor: Ömer Kalayci
With the advent of high-throughput sequencing and genotyping technologies, the study of the genetic contributions to asthma exacerbations has shifted from hypothesis-driven, limited candidategene strategies to genome-wide association studies (GWAS). [7][8][9][10][11][12][13][14] Pharmacogenomics studies of asthma exacerbations as an outcome of treatment response have identified five suggestive associations for asthma exacerbations despite inhaled corticosteroids (CMTR1, 9 APOBEC3B-APOBEC3C, 8 and CACNA2D3-WNT5A 11 ), or long-acting beta2-agonists (TBX3 and EPHA7). 10 Beyond pharmacogenomics, other studies have focused on asthma exacerbations independently of treatment. In European-descent populations, CDHR3, CTNNA3, and HLA-DQB1 have been associated with severe asthma exacerbations. 7,13 More recently, the representation of ethnically diverse populations has increased in GWAS of asthma exacerbations. A meta-analysis of GWAS in Hispanic/Latino children identified a single-nucleotide polymorphism (SNP) at FLJ22447 that modulated KCNJ2-AS1 expression in nasal epithelium through DNA methylation. 12 In Hispanic/Latinos and African Americans, a genome-wide significant locus for asthma with exacerbations regulated LINC01913 lung gene expression and DNA methylation levels of the PKDCC gene in whole blood. 14 However, none of those studies has approached the search for genetic determinants of asthma exacerbations independently of treatment from a multi-ancestry framework.
To improve our understanding on genetic and biological mechanisms of asthma exacerbations across multiple populations, we conducted the first multi-ancestry meta-analysis of GWAS of asthma exacerbations independently of treatment and attempted to validate previous associations. Then, we conducted in silico and in vivo downstream analyses to assess the potential functional effects of the associated SNPs over DNA methylation and gene expression.

| Studydesignandstudypopulations
We performed a two-stage study to identify genetic variants associated with asthma exacerbations, defined as a binary variable based on the presence of emergency care, hospitalizations, or administration of systemic corticosteroids because of asthma. We also considered a definition of moderate exacerbations, 3 comprising unscheduled general practitioner or pulmonary specialist visits and school absence, as no information on the former variables was available for some studies. A period of 6-24 months or ever was considered depending on the data available for each study (Tables S1 and S2). In the discovery phase, we performed a multi-ancestry meta-analysis of GWAS of asthma exacerbations in 9392 patients with asthma from 12 studies, including 4989 European-descents from nine studies, 2181 Hispanics/Latinos, 1250 Singaporean Chinese, and 972 African Americans. We attempted to replicate the findings from the discovery phase in a total of 37,555 participants with asthma, including 36,477 Europeans from seven studies, 877 Latinos from two studies, and 201 Filipinos from one study (Table S2). A detailed description of each study is available in the Appendix S1. All studies included were approved by their respective Institutional review boards, and written informed consent was provided by participants or their parents/caregivers. All methods followed the Declaration of Project (1KGP). 16 Genotyping and imputation procedures for the discovery and replication studies are detailed in the Appendix S1 and Tables S1 and S2.

| Associationanalysis
Association between genetic variants and asthma exacerbations was tested using logistic regression models including age, sex, and principal components from the genotype matrix (if needed to correct for population stratification) (Table S1). Analyses were conducted separately for each study using PLINK 2.0, 17 EPACTS 3.2.6 18 or rvtests 2.1.0. 19 Results were filtered with the EasyQC software 20 to retain variants with a minor allele frequency (MAF) ≥ 1% and imputation quality R 2 ≥ .3, absolute value of the beta coefficient <10, standard error of the beta included in the interval [0,10], and minor allele cut-off ≥6.
In the discovery phase, genetic variants that were available in at least two ethnic-specific studies were meta-analyzed with METASOFT, 21 using fixed-effects or random-effects models based on the heterogeneity among studies (measured by the Cochran's Q test p-value). Ethnic-specific results were then combined in a multiancestry meta-analysis. Independent variants (r 2 ≤ .8) with suggestive association 22 at p ≤ 5 × 10 −5 within 1 Megabase were identified with GCTA-COJO v1.93.2 23 using the 1KGP reference. 16 These variants were evaluated in the replication stage, following the same procedures as in the discovery phase. Evidence of replication was

Key Message
A large multi-ancestry meta-analysis of GWAS of asthma exacerbations revealed two novel susceptibility loci located close to PANK1 and at the intergenic region of VCAM1 and EXTL2. These loci decreased PANK1 and EXTL2 gene expression in whole blood, respectively. Both genetic variants were associated with DNA methylation levels at CpG sites nearby. Our results identified two gene targets for asthma exacerbations that should be further explored to assess their specific role in asthma.
considered if the variants showed consistent direction of effects with the discovery stage at p ≤ .05.

| Assessmentofsharedgeneticbasisofasthma exacerbations with other traits
To identify groups of genes previously associated with other traits, we used a Gene-Set Enrichment Analysis (GSEA), as implemented in FUMA GWAS 24 via the GENE2FUNC algorithm, and queried the GWAS catalog. 25 SNPs with p ≤ 1 × 10 −4 in the discovery phase of the meta-analysis of GWAS were mapped to the closest gene using the UCSC Table Browser tool. 26 A false discovery rate (FDR) of 5% was used to declare significance.
To estimate the pairwise genome-wide genetic correlations (R g ) between asthma exacerbations and other traits, we compared our findings with publicly available GWAS summary statistics via LD score regression using LDHub. 27 As most of the GWAS have been conducted in European populations, the analysis was restricted to predominantly European-descent individuals to maximize the statistical power. A Bonferroni-corrected significance threshold of p < .05/711 traits = 6.48 × 10 −5 was applied.

| Sensitivityanalysis
To assess the robustness of the genetic associations, we conducted sensitivity analyses for the time-dependent probability occurrence of exacerbations, the effect of Body Mass Index (BMI), obesity, asthma severity, and age group. Moreover, we evaluated the association of the variants with asthma susceptibility, as detailed in the Appendix S1. Studies from the discovery stage that had covariate data available were considered.

| Methylationprofilingandqualitycontrol
Whole blood DNA methylation from Hispanics/Latinos and African Americans was profiled using the Infinium HumanMethylation450 BeadChip or the Infinium Methylation EPIC BeadChip arrays. Briefly, low-quality probes and samples, outliers of DNA methylation, and samples with sex mismatch or mixed genotype distributions on the control SNP probes were excluded. Standard background correction, dye-bias correction, inter-array normalization, and probe-type bias adjustment were performed, and beta values were transformed to M-values for better statistical performance. Quality control is detailed in the Appendix S1.

| Validationofpreviousassociations
A literature search for all studies reporting genetic loci significantly associated with asthma exacerbations was conducted, as described in the Appendix S1. Association results in the discovery stage were extracted and significance threshold was defined as p = .05/ number of tested SNPs to adjust for multiple testing.

| Characteristicsofthepatients
In the discovery phase, we analyzed 2781 exacerbators and 6611 non-exacerbators; 53 were male participants in the discovery and replication phases, respectively.

| Discoveryphase
The quantile-quantile plots did not show major genomic inflation due to population stratification in each individual study ( Figure S1), the combined results from individuals of European-descent ( Figure S2), or the multi-ancestry meta-analysis ( Figure S3). In the multi-ancestry meta-analysis of 9,634,748 variants, 447 SNPs exhibited suggestive association (  (Figure 1, Figure S4).

| Replicationphase
Fifteen of the 126 independent variants identified in the discovery phase were not available for replication as they were mostly present in African Americans and Hispanics/Latinos ( Table S3). Two of the 106 variants present in more than one ethnic group were consistently associated with asthma exacerbations ( Table 1) (Table S4). Even though rs6888198 reached genome-wide significance in the discovery and showed consistent effects among Europeans in the replication phase, this SNP had opposite effects in Latinos and Filipinos, which resulted in the lack of replication in the multi-ancestry replication phase (Table 1, Figure S5).  (Table S5).

| Sensitivityanalysis
To assess the robustness of associations that replicated across stages to the time-dependent probability of occurrence of exacerbations, stratified analyses were performed in European-descents from the discovery stage that reported exacerbations for 6 vs.
12 months. Consistent effects per period were observed across periods ( Table 2).
As the post-GWAS analyses revealed significant enrichment/ correlation at p < .05 with fat mass/distribution, the association of rs12091010 and rs943126 after additional adjustment by BMI/ obesity was examined in individuals from the discovery phase with BMI data available. Moreover, the effect of asthma severity alone or combined with BMI/obesity on the genetic association exacerbations was evaluated. The effects sizes of the genetic association after additional adjustment by these variables remained consistent with the effects reported in the discovery stage (Table S7).
We next investigated if the observed effects could differ across age groups in those studies that analyzed exclusively children or adults, but the effect sizes remained consistent across age groups (Table S8). Moreover, we assessed if the effects could be driven by the underlying asthma syndrome, rather than asthma exacerbations, and no significant association with asthma was found in results from the UK Biobank or the Michigan Genomics Initiative (Table S9).

| Functionalexplorationofvariantsassociated with asthma exacerbations
We next assessed the association between DNA methylation levels in whole blood at 525 and 538 CpG sites with rs12091010 and rs943126, respectively. A total of 7 and 1 SNP-CpG pairs for rs943126 and rs12091010 exhibited Storey q < .05, respectively (Table 3 and Table S10). Two of these replicated consistently in F I G U R E 1 Manhattan plot of the results of the discovery stage of the multi-ancestry meta-analysis of GWAS of asthma exacerbations (represented as -log 10 p-value on the y-axis) along the chromosome position of the variants analyzed (x-axis). The suggestive (p = 5 × 10 −5 ) and genome-wide (p = 5 × 10 −8 ) significance thresholds are indicated by the black line and dark gray lines Europeans for rs943126 (cg25770176 and cg00475140). In silico analyses revealed 10 SNP-CpGs pairs, 3 of which showed consistent effects in Hispanics/Latinos and African Americans at Storey q < .05 (Tables S11 and S12) including the previous two pairs and rs943126-cg03948048. The eight significant CpG sites in minority children showed significant enrichment (q < .001) in transcription factor (TF) motifs in lung (Table S13). Besides, the T allele of rs12091010 was associated with decreased EXTL2 expression in whole blood from Europeans, according to PhenoScanner. 31 The C allele of rs943126 was associated with increased expression of PANK1 in whole blood from Europeans (Table S14). Both variants showed evidence of longrange chromatin interaction with several genes in lymphoblastoid cells, including VCAM1 and EXTL2 for rs12091010 and PANK1 for rs943126 (Table S15).

| Validationofpreviousassociations
We next examined 47 previous genetic loci for asthma exacerb ations 7,8,12,13,34-36 and moderate-to-severe asthma 37 for association with asthma exacerbations in the discovery phase. A total of 5 variants had p < .05 in Europeans, 2 in Hispanics/Latinos, 5 in African Americans, and 1 in Singaporean Chinese (Table S16).

| DISCUSS ION
To our knowledge, this is the first multi-ancestry meta-analysis of GWAS of asthma exacerbations independently of treatment including European, Hispanic/Latino, Asian, and African American patients with asthma. In our combined analysis of 46,947 individuals with asthma, two regulatory SNPs were significantly and consistently as-   N-acetylglucosamine to the glycosaminoglycan-protein linkage region. 41 Decreased EXTL2 causes an over-accumulation of GAGs 42 that can promote inflammation in injured areas. 43,44 Moreover, in bone marrow-derived macrophages from EXTL2 −/− mice, there is overproduction of key molecules involved in inflammation and extracellular matrix remodeling, including tumor necrosis factor α (TNFα) and several matrix metalloproteinases. 43 In a scenario of overaccumulation of GAGs under the loss of EXTL2 in macrophages, GAGs act as inflammatory mediators with strong Toll-like receptor 4 (TLR4) agonist capacity. 44 Interestingly, genetic variation in both VCAM1 and EXTL2 is associated with blood cell counts and multiple sclerosis, according to the GWAS catalog. 25 PANK1 catalyzes coenzyme A biosynthesis, regulated by the transcription factor peroxisome proliferator-activating receptor α (PPARα), 45 a key anti-inflammatory factor in asthma. 46  The T allele is more common among Latinos/Admixed Americans, followed by Europeans, Africans, and East Asians ( Figure S6).
The T allele of rs943126 at PANK1, which is less common among Europeans than the rest of populations ( Figure S7), was associated with a higher risk of asthma exacerbations in the combined analysis of the discovery and replication phases and with decreased gene expression of PANK1 in whole blood from Europeans according to PhenoScanner. 31 However, these eQTL effects were not validated in the GTEx data. 30 In the discovery phase, the most significant association was located at the intronic SNP rs6888198 (CDH12), but no evidence of replication was found in the second stage (p > .05) despite the consistency of the direction of the effect across study phases.
Interestingly, rs6888198 showed variable MAF among populations, with the largest MAF among Africans and Latinos ( Figure S8).
CDH12 has been associated with angiogenesis and progression of several types of cancers. [52][53][54] Specifically, in colorectal cancer, it has been suggested that CDH12 increases cancer cell migration by promoting epithelial-mesenchymal transition via activation of the Snail transcription factor pathway. CDH12 expression is positively modulated by the chemotactic factor CCL2, 53,54 whose levels increases in blood and airway smooth muscle from asthma patients compared to healthy controls. 55 We also attempted to assess previously associated loci for asthma  Maria Pino-Yanes: Resources (equal); conceptualization (lead); supervision (lead); funding acquisition (lead); methodology (equal); writing original draft (supporting); writing-review and editing (lead).

ACK N OWLED G M ENTS
The authors thank the patients, families, recruiters, healthcare providers, and community clinics for their participation in the studies analyzed in this manuscript. The authors also acknowledge the contribution of the high-performance compute cluster Wynton HPC underlying UCSF's Research Computing Capability to the results of this research and Sandra Salazar for her support as GALA II and SAGE study coordinator. The authors thank the Centro Nacional de