Each significant cis-eQTL SNP was then tested for association with COPD in the combined GWAS dataset from ECLIPSE, Norway, and NETT-NAS [11]. The published combined GWAS analysis was a mega-analysis of individual-level genotype data, using logistic regression, adjusted for age, pack-years of smoking and principal components for genetic ancestry. In the published meta-analysis, stratified logistic regression was performed within each case-control study and results were combined using Z-scores for weighting by the inverse variance. SNPs associated with COPD at p<0.01 in either the combined GWAS analysis (mega-analysis) or the GWAS meta-analysis were genotyped for replication in ICGN and COPDGene. In the COPDGene study, case-control data were analyzed with linear regression models, adjusted for age, sex, and pack-years of smoking, using PLINK version 1.0.7 [31]. Family-based ICGN data were analyzed in PBAT version 3.6.1, adjusted for age, sex, and pack-years of smoking [32].
