Asthma is a complex respiratory disease characterized by chronic airway inflammation leading to respiratory symptoms, such as wheezing, coughing, shortness of breath, and airflow limitation.1 According to the last global reports, asthma affected 300 million individuals worldwide in 2019, and caused 400,000 annual deaths.2 However, large differences in asthma prevalence across populations and regions are observed, even within the same country, as in Spain.3 Some patients may not respond appropriately to the treatment with controller medications, and their worsened symptoms can lead to life-threatening episodes known as asthma exacerbations (AEs).1 In recent years, several single-nucleotide polymorphisms (SNPs) associated with AEs have been identified through genome-wide association studies (GWAS).4 These studies uncovered genetic biomarkers of AE with population-specific effects or shared across populations, such as Europeans, African Americans, and Hispanic/Latin Americans.4 To the best of our knowledge, no prior study has focused on the association of genetic variants with AEs in the Spanish population. Therefore, this study aimed to identify specific genetic variants associated with AEs in the Spanish population.

A discovery and replication design was followed. The discovery included 345 adults from the Genomics and Metagenomics of Asthma Severity (GEMAS),5 369 from Characterizing Alpha-1-Antitrypsin Deficiency in patients with pulmonary diseases (CAATDPUL),6 and 138 from MEchanisms involved in the Genesis and evolution of Asthma (MEGA)7 studies (Table 1). The replication phase analyzed 90 children with asthma enrolled in the GEMAS study. Asthma patients were classified into cases and controls based on the presence or absence of severe AEs, respectively, defined by the presence of hospitalizations, emergency room visits, and/or oral corticosteroid use in the past 12 months. Genotyping data was available for the CAATDPUL and MEGA studies,6 and was newly generated for some of the GEMAS participants (Table E1, supplementary material). Association between genetic variants and AEs was assessed using logistic regression models adjusted by age, sex, and principal components capturing genetic ancestry estimated from the genotype matrix using PLINK 2.0. In the discovery phase, GWAS for AEs was conducted separately for each study, followed by a meta-analysis. From the suggestive variants (p-value <5×10−5), independent variants (r2≤0.8) within 1 Megabase were further explored for replication, which was declared for variants showing nominal association (p-value ≤0.05) and consistent effects with the ones found in the discovery phase. To ensure the robustness of the replicated associations, sensitivity analyses were performed by further adjusting the logistic regression models for additional potential confounders, such as obesity, body mass index (BMI) categories, smoking, and asthma severity. Additionally, we assessed the effects of the SNPs with evidence of replication on DNA methylation (DNAm) through a cis-methylation quantitative trait locus (meQTL) analysis. In silico confirmation of functional effects on DNAm (meQTLs) and gene expression (eQTLs) was evaluated using public data. Finally, a Gene-Set Enrichment Analysis (GSEA) was carried out to evaluate previous gene-trait and gene-drug target associations. Detailed methods are provided in the supplementary material.

From the 8.7 million SNPs included in the meta-analysis from the discovery stage, 193 SNPs showed suggestive association with AEs (Fig. 1A, Table E2, supplementary material), including 52 independent variants (Table E3). The most significant association was the SNP rs4759090 annotated to the PDE1B gene (odds ratio G allele=0.56, 95% confidence interval=0.43–0.71, p-value=2.90×10−6) (Fig. E2, supplementary material). From the 52 SNPs independently associated with AEs, three were nominally replicated (p-value <0.05) in the peditric population: rs847517 (BRMS1L), rs79017090 (ZNF780A), and rs11786009 (DUSP4), Table E4, supplementary material, Fig. 1B). After sensitivity analyses, the association of these three SNPs was robust to the adjustment for all potential confounders evaluated (Table E5, supplementary material).

Fig. 1.

(A) Manhattan plot of the meta-GWAS of asthma exacerbations in the discovery phase. The blue and red lines represent the suggestive (p-value <5×10−5) and the genome-wide (p-value <5×10−8) significance thresholds, respectively. Gene annotation is represented for the top-hit signal and those with evidence of replication. (B) Forest plot of the association results for the SNPs rs11786009 (DUSP4), rs79017090 (ZNF780A), and rs847517 (BRMS1L) from the meta-GWAS of asthma exacerbations in the discovery phase and replication phase. The x-axis represents the odds ratios (ORs) and the vertical line corresponds to OR=1, which means no significant effect. ORs and confidence intervals obtained from SNP-AEs association in each study and meta-analysis are represented as squares and lines or diamonds and lines, respectively.

(0.38MB).

Among the replicated associations, the SNPs rs79017090 and rs847517 were significantly associated with DNAm levels at six and two CpGs, respectively (Table E6, supplementary material). In silico analyses revealed that the G allele of the SNP rs79017090 showed a significant association with increased ZNF780A gene expression in whole blood (coefficient=0.20; p-value=1.60×10−11) and lung tissue (coefficient=0.46; p-value=1.40×10−24). Finally, the GSEA showed an enrichment in previous trait-associations, including cholesterol levels, BMI, and life expectancy, and also in drugs acting as phosphodiesterase inhibitors (i.e., oxagrelate, and papaverine) (FDR <0.05) (Fig. E3, Table E7, supplementary material).

To the best of our knowledge, this is the first GWAS of AEs specifically focused on the Spanish population, reporting novel associations with AEs not revealed by the largest meta-GWAS of AEs performed to date.8 In the meta-analysis of 852 adults with asthma, the most significant association with AEs was the SNP rs4759090, located at the second intronic region of a member of the cyclic nucleotide phosphodiesterase family (PDE1B) gene. PDE1B acts by hydrolyzing second messengers regulating smooth muscle relaxation in asthma and participates in interleukin-13 (IL-13) production, which is involved in allergic lung inflammation.9 In fact, PDE inhibitors such as theophylline are effective drugs in asthma treatment due to their anti-inflammatory effect.9

Regarding the validated associations, the G allele of the SNP rs79017090, located at the intergenic region of the Zinc Finger Protein 780A (ZNF780A) and Mitogen-Activated Protein Kinase Kinase Kinase 10 (MAP3K10) genes, showed a risk effect for developing AEs and exhibited an association with DNAm at several nearby CpG sites in whole blood from adults with asthma. Moreover, in silico analyses revealed an association of the same allele with increased ZNF780A expression in whole blood and lung tissue. Epigenetic variation in the ZNF780A gene, has been previously associated with prenatal exposure to perfluorononanoic acid,10 a compound associated with childhood asthma risk in the Spanish population.11 On the other hand, MAP3K10 participates in the MAPK signaling pathway that contributes to the airway inflammation and remodeling processes underlying asthma.12 In addition, the T allele of SNP rs847517 (BRMS1L) and the G allele of SNP rs11786009 (DUSP4) showed a consistent protective effect against AEs in both adults and children with asthma. BRMS1L, which encodes a breast cancer metastasis-suppressor, is involved in Wnt signaling13 epigenetic regulation, a pathway previously linked to AEs despite corticosteroid use.14DUSP4 encodes a protein from the dual specificity phosphatase subfamily which deactivates kinases in the MAPK pathway, reducing inflammation and enhancing corticosteroids’ anti-inflammatory effects, thereby improving corticosteroid sensitivity.15 Notably, a previous study showed DUSP4 expression was lower in blood from patients with severe asthma, suggesting it as a potential therapeutic target.15

Finally, genetic variants related to AEs were enriched in previous associations reported for BMI, cholesterol, and life expectancy. Obesity and abnormal levels of lipids in blood (dyslipidemia), including cholesterol, have been shown to increase the frequency and severity of AEs.16,17 Moreover, previous studies have reported that AEs lead to a higher mortality risk in the next month.18 Focusing on drugs, we found evidence of enrichment in PDE inhibitors, such as oxagrelate and papaverine. Although no major studies have validated these enriched drugs as a primary asthma treatment, other PDE inhibitors as theophylline are commonly used as second-line drugs for asthma.19

This study has several strengths. First, the definition of AEs was consistent across all studies comprised in the discovery and replication phases. Second, despite the small sample size, three suggestive associations were replicated and showed consistent effects in the pediatric population. Moreover, these associations were robust to the adjustment for potential confounders of AEs. Third, we integrated epigenomic and genomic data from our studies, as well as in silico data from transcriptomics and genomics, to assess the potential functional role of the associated variants on DNAm and gene expression levels.

On the other hand, we acknowledge some study limitations. First, AEs definition was based on retrospective questionnaires in the discovery and replication phases, which could reduce the accuracy and reliability of the phenotype, compromising the ability to detect robust genetic associations. Second, the reduced statistical power given by the limited sample size analyzed could explain the inability of our associations to reach the genome-wide significance threshold. Moreover, AEs constitute a complex and heterogeneous phenotype, and it has been shown that large sample sizes are necessary for the detection of any genome-wide signal in such scenario.20 Third, our discovery was focused on adults and the replication in children, which limited our ability to fully replicate the results found in the initial phase. Fourth, the functional implication of our AEs’ suggestive associations on gene expression was explored using publicly available from European populations since gene expression data was not available in the analyzed cohorts.

In summary, the first GWAS of AEs conducted in the Spanish population revealed novel associations for genetic loci with AEs with consistent effects across age groups. We also reported that some of these loci regulate DNAm levels at nearby CpG sites and gene expression in whole blood and lung. The AEs-related genetic markers were enriched in BMI, cholesterol, and life expectancy. These findings provide novel insights into the molecular mechanisms of AEs and suggest potential alternative therapies. However, future investigation of the genetic basis underlying AEs in larger sample sizes is needed.