In addition to detecting signals of association previously reported by studies of quantitative lung function (appendix p 32–37), in our case-control analysis of FEV1 extremes we identified six novel signals of association (p<5 × 10−8) with low FEV1 versus high FEV1 (table 2; figure 3; appendix pp 38–53, 96–98, 102). The sentinel SNPs at five of these six signals, in or near TET2, NPNT, HLA-DQB1/HLA-DQA2, KANSL1, and TSEN54, were common (MAF ≥5%) and showed a stronger association with low FEV1 in never smokers than heavy smokers. The sentinel SNP at an intergenic signal between RBM19 and TBX5 was a rare variant (MAF=0·13%) that showed strongest association with low FEV1 in heavy smokers. The lead SNPs at each of these loci showed association with COPD (table 2; appendix p 55). The 26 previously reported SNPs (associated with FEV1, FEV1:FVC ratio, or both)5,7–9 explained 2·33% of the variance of FEV1 in our data; adding in the SNPs representing our six novel signals of association with FEV1 extremes, we explained 3·63% of the variance of FEV1 (appendix p 17).
