Genome wide interaction study of genetic variants associated with lung function decline

Abstract

Some genetic variants are associated with lung function decline and chronic obstructive pulmonary disease (COPD), but functional studies are necessary to confirm causality. We investigated the genetic susceptibility-associated lung function decline with or without COPD, using data from a community-based cohort (N = 8554). A genome-wide interaction study was conducted to identify the association between genetic variants and pulmonary function, and the way variants relate to lung impairment in accordance with smoking status and amount was examined. We further used a linear mixed model to examine the association and interaction to time effect. We found annual mean FEV₁ declines of 41.7 mL for men and 33.4 mL for women, and the annual rate of decline in FEV₁ was the fastest for current smokers. We also found a previously identified locus near FAM13A, the most significant SNPs from the results of two likelihood ratio tests for FEV₁/FVC (P = 1.56 × 10^-10). These selected SNPs were located in the upstream region of FAM13A on chromosome 4 and had similar minor allele frequencies (MAFs). Furthermore, we found that certain SNPs tended to have lower FEV₁/FVC values, and lung function decreased much faster with time interactions. The SNP most associated with lung function decline was the rs75679995 SNP on chromosome 7, and those SNPs located within the TAD of the DNAH11 region and the eQTL of rs9991425 revealed a higher expression of MFAP3L and AADAT genes (P = 2.28 × 10^-7 and 2.01 × 10^-6, respectively). This is the first study to investigate gene-time interactions in lung function decline as a risk factor for COPD in the Korean population. In addition to replicating previously known signals for FAM13A, we identified two genomic regions (DNAH11, AADAT) that are potentially involved in gene-environment interactions, warranting further investigation to confirm their roles.

Keywords: AADAT gene; DNAH11 gene; FAM13A gene; Airflow obstruction; Chronic obstructive pulmonary disease; Genome-wide association study; Single nucleotide polymorphism.

Fig. 1

Comparison of the annual mean decline of FEV₁. (a) FEV₁ for male (n = 4026) and female patients (n = 4528). (b) Comparison of the annual mean decline in FEV₁ among male healthy never smokers, former smokers, and current smokers. ^aP-value compared to healthy never smoker, ^bP < 0.001 compared to former smoker. *Data are presented as mean and 95% confidence intervals. FEV₁ forced expiratory volume in 1 s.

Fig. 2

QQ plot and MH plot from GWIS in KARE data. (a) Comparison between observed and expected P-values of quantiles under uniform distribution (null hypothesis). The VIF was 1.04. (b) Plot of the logarithms of the P-values of 3,333,374 SNPs against the physical chromosomal position. The red line represents genome-wide significance level (5 × 10^–8), which was met by several SNPs located at chromosome 4. QQ Quantile–quantile, MH Manhattan, VIF variance inflation factor, GWIS genome-wide interaction study, KARE Korea Association Resource Project.

Fig. 3

Regional plot for chromosome 4 SNPs considering SNP × time interaction. Regional association plots at the most significant loci on chromosome 4 show associations with lung function, as measured by FEV1/FVC and SNPs for SNP Analysis Incorporating SNP × Time Interaction.

Fig. 4

QQ plot and MH plot (interaction effects) (a) QQ plot. (b) MH plot of the results of interaction effect on FEV₁/FVC. The blue line indicating the genome-wide suggestive level (5 × 10⁻⁵). QQ Quantile–quantile, MH Manhattan, VIF variance inflation factor.

Fig. 5

LD among the four most significant SNPs and regional plots (chromosome 7, interaction effects). LD plot and regional association plots at the most significant loci showing association with lung function as measured by FEV₁/FVC and SNP-by-time interaction for chromosome 7. LD linkage disequilibrium, SNP single nucleotide polymorphism, FEV₁ forced expiratory volume in 1 s, FVC functional vital capacity.

Fig. 6

Regional plot (chromosome 4, interaction effects).

Fig. 7

Estimated FEV₁/FVC according to time interaction and rs75679995 The estimated FEV1/FVC ratio was generated by gene-by-time analysis using KARE data. Minor allele G has negative impact on increment of time interaction.

Fig. 8

Flow diagram for the KARE cohort. Flow chart showing how the individuals and SNPs were included and excluded in the study.