The top 100 SNPs showing the lowest P values in this stage were selected for assessment in replication cohorts. For replication, we used a two stage strategy using three independent cohorts (Figure 1). In the first stage, family-based association analysis for COPD affection status was conducted in the ICGN data using PBAT version 3.6 [30]. Adjustments for age, gender, pack-years of smoking, current smoking status and center were performed in order to take into account the effect of smoking on the association results. Association with FEV1 was also tested using PBAT with age, gender, pack-years of smoking, current smoking status and height as co-variates. Gene-by-environment interaction analyses were also conducted using the PBAT program. Biallelic tests were conducted for SNPs using an additive genetic model. In the second stage the NETT case-control study was analyzed for the presence/absence of COPD using an additive genetic model. An unadjusted analysis and a logistic regression model adjusted for age and pack-years of smoking were conducted; sex was not included as a covariate because all NAS subjects were male, and current smoking was not
