An admixture mapping meta-analysis implicates genetic variation at 18q21 with asthma susceptibility in Latinos

Abstract

Background: Asthma is a common but complex disease with racial/ethnic differences in prevalence, morbidity, and response to therapies.

Objective: We sought to perform an analysis of genetic ancestry to identify new loci that contribute to asthma susceptibility.

Methods: We leveraged the mixed ancestry of 3902 Latinos and performed an admixture mapping meta-analysis for asthma susceptibility. We replicated associations in an independent study of 3774 Latinos, performed targeted sequencing for fine mapping, and tested for disease correlations with gene expression in the whole blood of more than 500 subjects from 3 racial/ethnic groups.

Results: We identified a genome-wide significant admixture mapping peak at 18q21 in Latinos (P = 6.8 × 10^-6), where Native American ancestry was associated with increased risk of asthma (odds ratio [OR], 1.20; 95% CI, 1.07-1.34; P = .002) and European ancestry was associated with protection (OR, 0.86; 95% CI, 0.77-0.96; P = .008). Our findings were replicated in an independent childhood asthma study in Latinos (P = 5.3 × 10^-3, combined P = 2.6 × 10^-7). Fine mapping of 18q21 in 1978 Latinos identified a significant association with multiple variants 5' of SMAD family member 2 (SMAD2) in Mexicans, whereas a single rare variant in the same window was the top association in Puerto Ricans. Low versus high SMAD2 blood expression was correlated with case status (13.4% lower expression; OR, 3.93; 95% CI, 2.12-7.28; P < .001). In addition, lower expression of SMAD2 was associated with more frequent exacerbations among Puerto Ricans with asthma.

Conclusion: Ancestry at 18q21 was significantly associated with asthma in Latinos and implicated multiple ancestry-informative noncoding variants upstream of SMAD2 with asthma susceptibility. Furthermore, decreased SMAD2 expression in blood was strongly associated with increased asthma risk and increased exacerbations.

Keywords: Asthma; Latinos; SMAD2; admixture mapping; asthma exacerbations; gene expression; meta-analysis; rare variation; targeted sequencing.

FIG 1.

Outline of the study approach. Data generation steps are in green, and analysis steps are in blue. Numbers of subjects available for analyses can be found in Table I.

FIG 2.

Results of the admixture mapping meta-analysis. A, Admixture mapping Manhattan plot. B, Summary of admixture mapping findings across the 18q21 locus. C, Forest plots for each ancestry, including ORs and CIs, with the size of the square inversely proportional to the SE. Meta-analysis estimates through fixed-effects models are shown as diamonds, with values in Table II.

FIG 3.

Fine mapping of 18q21. A and B, pooled (SKAT-O) and individual (GWAS) variants in Mexicans (Fig 3, A) and Puerto Ricans (Fig 3, B). C, Regulatory elements from ENCODE, with darker boxes having stronger signals and numbers indicating hypersensitive cell lines. Also, transcription factor bindings site are shown through ChIP-seq in ENCODE. Green lines indicate the highest-scoring binding motif for the corresponding factor and most associated cell type from the UCSC Genome Browser.

FIG 4.

Analysis of SMAD2 expression in whole blood. A, Means and 95% CIs of whole blood SMAD2 expression, which was lower in cases (P = .002). Neither ZBTB7C nor SMAD3 showed any differential expression. B, Receiver operating characteristic (ROC) curves of high exacerbation scores in patients with asthma by using clinical variables, demographic data, and SMAD2 gene expression accounting for age, population, spirometry, and controller medication use. AUCs and 10-fold cross-validated prediction errors are shown in legends.