Risk loci for chronic obstructive pulmonary disease: a genome-wide association study and meta-analysis

Abstract

Background

The genetic risk factors for susceptibility to chronic obstructive pulmonary disease (COPD) are still largely unknown. Additional genetic variants are likely to be identified by genome-wide association studies in larger cohorts or specific subgroups.

Methods

Genome-wide association analysis in COPDGene (non-Hispanic whites and African-Americans) was combined with existing data from the ECLIPSE, NETT/NAS, and GenKOLS (Norway) studies. Analyses were performed both using all moderate-to-severe cases and the subset of severe cases. Top loci not previously described as genome-wide significant were genotyped in the ICGN study, and results combined in a joint meta-analysis.

Findings

Analysis of a total of 6,633 moderate-to-severe cases and 5,704 controls confirmed association at three known loci: CHRNA3/CHRNA5/IREB2, FAM13A, and HHIP (10⁻¹² < P < 10⁻¹⁴), and also showed significant evidence of association at a novel locus near RIN3 (overall P, including ICGN = 5•4×10⁻⁹). In the severe COPD analysis (n=3,497), the effects at two of three previously described loci were significantly stronger; we also identified two additional loci previously reported to affect gene expression of MMP12 and TGFB2 (overall P = 2•6x10⁻⁹ and 8•3×10⁻⁹). RIN3 and TGFB2 expression levels were reduced in a set of Lung Tissue Research Consortium COPD lung tissue samples compared with controls.

Interpretation

In a genome-wide study of COPD, we confirmed associations at three known loci and found additional genome-wide significant associations with moderate-to-severe COPD near RIN3 and with severe COPD near MMP12 and TGFB2. Genetic variants, apart from alpha-1 antitrypsin deficiency, increase the risk of COPD. Our analysis of severe COPD suggests additional genetic variants may be identified by focusing on this subgroup.

Funding

National Heart, Lung, and Blood Institute; the COPD Foundation through contributions from AstraZeneca, Boehringer Ingelheim, Novartis, and Sepracor; GlaxoSmithKline; Centers for Medicare and Medicaid Services; Agency for Healthcare Research and Quality; US Department of Veterans Affairs.

Introduction

Chronic obstructive pulmonary disease (COPD), characterized by persistent and usually progressive airflow obstruction, is one of the leading causes of morbidity and mortality worldwide. While cigarette smoking is the major environmental risk factor, the burden of COPD is increasing^1,2 despite many successful efforts at tobacco control, and the response to cigarette smoke is characterized by high inter-individual variability³. Genetic factors are a major contributor to this variability^4–6, but the specific genetic loci responsible for this variation remain largely unknown⁷. Genome-wide association studies have successfully identified loci which are often novel for a range of complex diseases, including COPD, that have subsequently replicated^8–17, but the majority of genetic susceptibility due to common variation remains unexplained¹⁸. Identifying genetic loci may lead to improved risk prediction and subtype identification¹⁹, and is arguably the most promising unbiased approach to understand disease mechanisms in humans and enable future specific and rational therapies²⁰.

We recently completed genome-wide genotyping in COPDGene, a large, genetic epidemiology study of over 10,000 non-Hispanic White and African American cigarette smokers (both current and ex-smokers) with and without COPD²¹. We sought to determine whether a genome-wide association study (GWAS), combining the results from COPDGene with previous association studies⁷, would reveal new genetic susceptibility loci.

Genome-wide association analyses in COPD to date have included subjects with mild or moderately severe airflow limitation^7,9,12,22. To our knowledge, a genome-wide association case-control study of severe COPD has not been previously reported. The severity of airflow limitation in COPD correlates with many other important disease characteristics, such as emphysema²³, functional limitation²⁴, and higher mortality²⁵. In addition to potentially identifying novel signals unique to severe disease, a genome-wide association study of severe COPD may have improved power compared with a study of moderate-to-severe COPD due to decreased phenotypic heterogeneity and misclassification in severe COPD cases, as well as enrichment for subjects with the highest genetic risk profile^26–30, despite the decreased sample size.

Methods

COPDGene (NCT00608764) is a large, multicenter study designed to investigate the genetic and epidemiologic characteristics of COPD and other smoking-related lung diseases²¹. COPDGene subjects were of self-described non-Hispanic white or African-American ancestry, and genotyped using the HumanOmniExpress (Illumina, San Diego, CA). Genotype imputation on the COPDGene cohorts was performed using MaCH and minimac^31,32 using 1000 Genomes³³ Phase I v3 European (EUR) and cosmopolitan reference panels for the non-Hispanic whites and African-Americans, respectively. Detailed descriptions of the ECLIPSE, NETT/NAS, and Norway (GenKOLS) cohorts, including genotyping quality control and imputation, have been previously published^{7,9,12,21,34–36}.

In all cohorts, ‘moderate-to-severe’ cases had GOLD Grade 2-4 COPD (moderate, severe, and very severe COPD; post-bronchodilator FEV₁ < 80% predicted with FEV₁/FVC < 0•7); individuals with severe alpha-1 antitrypsin deficiency were excluded.

Controls had normal spirometry with a history of cigarette smoking. For the analysis of ‘severe’ COPD, cases were limited to those with GOLD 3 and 4 disease (severe and very severe, post-bronchodilator FEV₁ < 50% predicted). Baseline characteristics of each of the genome-wide cohorts are shown in Table 1. Logistic regression was performed within each cohort and racial / ethnic group adjusting for age, pack-years of smoking, and ancestry-based principal components using plink (v1•07)³⁷, as previously described^7,12. Fixed-effects meta-analysis was performed using METAL (version 2010-08-01)³⁸. Heterogeneity was reported as both I^2,39 using the meta package in R (v2•3•0) (www.r-project.org) and P-values for Cochrane's Q. Markers were included for analysis if they passed genotyping or imputation quality control (as appropriate) in all genome-wide cohorts. Regional association plots were created using LocusZoom⁴⁰, using the 1000 Genomes EUR reference data for linkage disequilibrium (LD) calculations.

Table 1.

Baseline characteristics.

	COPDGene				ECLIPSE		NETT/NAS		GenKOLS
	NHW Case	NHW Control	AA Case	AA Control	Case	Control	Case	Control	Case	Control
n	2812	2534	821	1749	1764	178	373	435	863	808
Age	64•7 (8•2)	59•5 (8•7)	59•0 (8•2)	52•8 (6•0)	63•6 (7•1)	57•5 (9•4)	67•5 (5•8)	69•8 (7•5)	65•5 (10•0)	55•6 (9•7)
Pack-years	56•3 (28•0)	37•8 (20•3)	42•4 (23•0)	36•4 (20•1)	50•3 (27•4)	32•1 (24•8)	66•4 (30•7)	40•7 (27•9)	32•0 (18•5)	19•7 (13•6)
FEV1, % predicted	49•6 (18•0)	96•8 (11)	52•2 (17•8)	98•4 (12•2)	47•6 (15•6)	107•8 (13•6)	28•1 (7•4)	100•0 (13•2)	50•6 (17•4)	94•9 (9•2)
Sex (% male)	55•7	49•3	55•2	58•1	67	57•9	63•8	100	60•1	50•1

Values given as mean (SD) or percent, as appropriate. NHW: Non-hispanic white. AA: African-American.

Results yielding a P value threshold of < 5×10⁻⁷ at loci not previously described⁷ in the moderate-to-severe and severe COPD meta-analysis of COPDGene, ECLIPSE, NETT/NAS, and GenKOLS (Norway) were subsequently genotyped in 983 probands and 1876 siblings from the family-based International COPD Genetics Network study (ICGN)³⁴. Association analysis in ICGN was performed using PBAT (v3•61), under an additive model, adjusting for age and pack-years of smoking. Results from the family-based ICGN study were combined with case-control results using a joint meta-analysis⁴¹ weighted by sample size, using the number of informative transmissions in ICGN and the effective number of cases in each cohort. A joint meta-analysis P-value of < 5×10⁻⁸ was considered significant.

Differences in odds ratios between severe cases versus controls and between all cases (moderate-to-severe) versus controls were assessed by permutation. Region-based conditional analyses were performed using logistic regression, adjusting for the most significant (lead) single nucleotide polymorphism (SNP) in each region using genotyped or dosage data as appropriate, and testing all SNPs within a 250kb window on either side of the lead SNP for association with affection status. To estimate the combined effect of genetic risk variants, we constructed a genetic score based on the cumulative number of risk alleles in a logistic regression in the COPDGene non-Hispanic whites including age, pack-years, and ancestry-based principal components.

Additional analyses using the meta-analysis results included gene-based testing using VEGAS⁴² and the literature mining using GRAIL⁴³. Gene expression levels of TGFB2 and RIN3 were measured in lung tissue samples from 15 COPD patients – 8 with moderate (FEV₁ < 80% predicted), and 7 with severe (FEV₁ < 50% predicted) disease – and 15 control subjects with normal lung function, obtained from the NHLBI Lung Tissue Research Consortium (LTRC), as described previously⁴⁴.

Role of the funding source

GlaxoSmithKline was involved in study design and data collection for the ECLIPSE, GenKOLS (Norway), and ICGN studies. No other study sponsors had a role in study design or data collection, and none of the study sponsors had a role in data analysis, data interpretation, or writing of the report. The corresponding author had full access to all of the data and the final responsibility to submit for publication.

Results

Results from the GWAS of moderate-to-severe COPD in the COPDGene non-Hispanic whites and African-Americans are shown in Tables S6 and S7. The analysis in the non-Hispanic whites confirmed three previously known (CHRNA3/5/IREB2, HHIP, and FAM13A) COPD susceptibility loci, but neither study alone identified novel loci achieving conventional genome-wide significance (P < 5•0×10⁻⁸, Table S7 and S8).

The combined GWAS of moderate-to-severe COPD included COPDGene non-Hispanic whites, COPDGene African-Americans, ECLIPSE, NETT/NAS, and GenKOLS (Norway), for a total of 6,633 cases and 5,704 controls. Both individual and overall quantile-quantile (Q-Q) plots showed no evidence of significant population stratification (individual study λ_GC all ≤ 1.04; overall λ_GC =1•03; λ ⁴⁵_GC1000 = 1•01, Figure S1). The top results at each of the loci with P < 10⁻⁷ are shown in Table 3 and Figure 1. The three most significant SNPs in this meta-analysis were either identical to, or in strong LD – r² > 0•5 – with the top SNPs previously described at these three loci: 4q22 (FAM13A), 15q25 (CHRNA3/5/IREB2), and 4q31 (HHIP), confirming these previous association results ^9,10,12.

Table 3.

Top results for the genome-wide association analysis of moderate-to-severe COPD versus smoking controls in COPDGene non-Hispanic white and African-American, ECLIPSE, NETT/NAS, and GenKOLS (Norway) studies.

Locus	Nearest gene	SNP	Risk Allele	Frequency		Meta-analysis
				NHW	AA	OR (CI)	P	I2	Q
4q22	FAM13A	rs4416442	C	0•42	0•54	1•28 (1•2-1•36)	1•12×10−14	0•23	0•27
15q25	CHRNA3	rs12914385	T	0•42	0•19	1•28 (1•2-1•36)	6•38×10−14	0•26	0•25
4q31	HHIP	rs13141641	T	0•59	0•89	1•27 (1•19-1•36)	1•57×10−12	0•31	0•22
14q32	RIN3	rs754388	C	0•83	0•85	1•28 (1•18-1•39)	5•25×10−9	0	0•59

Allele coding represents + strand, hg19. Allele frequency is given for the risk allele. Nhw = Non-Hispanic white; AA = African-American.

Figure 1.

Local association plots for significant loci for the analysis of moderate-to-severe COPD in COPDGene non-Hispanic whites and African-Americans, ECLIPSE, NETT/NAS, and GenKOLS (Norway). The x-axis is chromosomal position, and the y-axis shows the –log₁₀ P-value. The most significant SNP at each locus is labeled in purple, with other SNPs colored by degree of linkage disequilibrium (r²).

We identified one novel additional locus with P < 5×10⁻ at 14q32; the top SNP at this locus was rs754388 (nearest gene – RIN3), with a P-value of 5•25×10⁻⁹. We genotyped this SNP in the ICGN Study, and tested for association with COPD in ICGN using a family-based test. While the evidence of association at this SNP did not achieve statistical significance (one-sided P=0•20), the overall meta-analysis P-value (including ICGN) for rs754388 remained genome-wide significant (5•4×10⁻⁹). An analysis of the effect of this SNP on FEV₁ as a quantative trait was not statistically significant.

The analysis of severe COPD reduced the number of cases to 3,497, while the number of controls remained the same. Baseline characteristics of the severe subsets of COPDGene, ECLIPSE, and GenKOLS (Norway) cases are shown in Table 2 (characteristics of NETT subjects are included in Table 1, as all NETT cases are severe). Similarly to the analysis of moderate-to-severe cases, we found no evidence of inflation due to population stratification (individual λ_GC ≤ 1.04; overall λ_GC = 1•04; λ_GC1000 = 1•01, Figure S2) among severe cases and controls. We again confirmed the three previously described COPD loci: 4q22 (FAM13A), 15q25 (CHRNA3/5/IREB2), and 4q31 (HHIP) – as genome-wide significant for severe COPD (Table 4 and Figure 2). We noted effect estimates for these loci tended to be larger in severe COPD than in moderate to severe COPD cases; these differences were statistically significant at two markers (P < 0•01 for rs13141641 (15q25), and rs12914385, (HHIP)), and just above statistical significance for a third (P = 0•08 for rs4416442 (FAM13A)).

Table 2.

Baseline characteristics of severe COPD subsets (COPDGene, ECLIPSE, and GenKOLS; all NETT subjects have severe COPD and were included in the severe COPD analysis).

	COPDGene		ECLIPSE	GenKOLS
	NHW	AA
n	1390	352	999	383
Age	65•2 (7•8)	60•6 (8•1)	63•5 (7•0)	66•7 (9•7)
Pack-years	58•7 (28•4)	43•9 (23•4)	50•7 (26•3)	33•0 (19•9)
FEV1, % predicted	34•0 (9•9)	34•8 (10•4)	36•5 (8•6)	34•4 (10•3)
Sex (% male)	57•8	58	69•9	61•5

Values given as mean (SD) or percent, as appropriate. NHW: Non-hispanic white. AA: African-American.

Table 4.

Top results for the genome-wide association analysis of severe COPD versus smoking controls in COPDGene non-Hispanic white and African-American, ECLIPSE, NETT/NAS, and GenKOLS (Norway) studies.

Locus	Nearest gene(s)	SNP	Risk Allele	Frequency		Meta-Analysis
				Nhw	Aa	OR (CI)	P	I2	Q
15q25	CHRNA3	rs12914385	T	0•42	0•19	1•39 (1•29-1•51)	2•70×10−16	0	0•76
4q31	HHIP	rs13141641	T	0•59	0•89	1•39 (1•28-1•51)	3•66×10−15	0	0•44
4q22	FAM13A	rs4416442	C	0•42	0•54	1•36 (1•26-1•47)	9•44×10−15	0	0•68
11q22	MMP3/12	rs626750	G	0•83	0•74	1•36 (1•23-1•51)	5•35×10−9	0	0•62
14q32	RIN3	rs754388	C	0•83	0•85	1•33 (1•2-1•48)	6•69×10−8	0	0•66
1q41	TGFB2	rs4846480	A	0•75	0•65	1•26 (1•16-1•37)	1•25×10−7	0	0•99

Allele coding represents + strand, hg19. Allele frequency is given for the risk allele. Nhw = Non-Hispanic white; AA = African-American.

Figure 2.

Local association plots for significant loci for the analysis of severe COPD in COPDGene non-Hispanic whites and African-Americans, ECLIPSE, NETT/NAS, and GenKOLS (Norway). The x-axis is chromosomal position, and the y-axis shows the – log₁₀ P-value. The most significant SNP at each locus is labeled in purple, with other SNPs colored by degree of linkage disequilibrium (r²).

We also identified two new genome-wide significant loci in the analysis of severe COPD versus controls. The first was at 11q22; the top-ranked SNP was rs626750 (nearest genes, MMP3 and MMP12). We found supportive evidence for association with severe COPD in ICGN (P = 0•06) and a genome-wide significant result in the joint meta-analysis (P = 2•6×10⁻⁹). This locus was previously reported in an analysis including subjects from NAS and NETT⁴⁶. After excluding NETT/NAS subjects, the joint meta-analysis P-value remained significant (P = 7•0×10⁻⁹). The second locus was at 1q41, where the top-ranked SNP was rs4846480 (nearest gene, TGFB2). This locus was just below-genome wide significance in the genome-wide cohorts (1•3×10⁻). Including the results from ICGN (P = 0•007), brought the joint meta-analysis results to P = 8•3×10⁻⁹.

To determine whether gene expression levels of these two genes not previously described in association with COPD – RIN3 and TGFB2 – were different in lung tissue samples from COPD cases versus controls, we performed real-time quantitative reverse transcription PCR in 18 control samples and 15 COPD samples – 8 with moderate (FEV₁ < 80% predicted), and 7 with severe (FEV₁ < 50% predicted) disease) – from the NHLBI Lung Tissue Research Consortium. RIN3 expression was significantly lower in COPD cases versus controls (P = 0•003, Figure S3). Differences in TGFB2 expression were not significant when comparing all cases versus controls (P = 0•5), but were significant when the cases were limited to those with severe disease (P = 0•002, Figure S4).

While the definitions of cases and controls within each study – based on GOLD criteria – were similar, COPD is a highly heterogeneous disease, and differences exist between the studies ⁴⁷. To explore these considerations, we used alternative methods for meta-analysis based on modified random-effects and binary effects model that may be more powerful in the presence of heterogeneity among studies^48,49 (see Supplement). However, we were not able to identify new genome-wide significant results using these methods.

We next sought to determine whether there was evidence for secondary associations at each described locus. We performed analyses conditioning on the top (lead) SNP at each genome-wide significant locus reported in this analysis, examining all SNPs present in 250kb flanking regions around the top signal. We found evidence suggestive of secondary associations (P < 5×10⁻⁴) in the analysis of moderate-to-severe COPD at 15q25 (conditioning on rs12914385) for a SNP in strong LD (r²=0•92 in EUR) with the previously reported rs13180 in IREB2 (rs12903295, intronic in IREB2, P = 9•9 × 10⁻⁵). Suggestive evidence of a secondary association was also found near the 14q32 (RIN3) locus conditioning on rs754388 (rs11849228, P = 1•3×10⁻⁴). In severe COPD, evidence supporting a secondary association was found at 15q25 in another intronic SNP in CHRNA3 (rs3743073, P=3•3×10⁻⁴).

The number of loci identified as influencing risk to COPD to date is modest, despite the relatively large sample size of this study; these loci explain < 5% of the liability-scale variance. To explore whether additional true association signals of weaker effect – beyond the ability to detect in our current analysis – might be present, we examined the characteristics of the top results (P < 0•01) in a meta-analysis of three white cohorts (ECLIPSE, NETT/NAS, and GenKOLS) within the COPDGene non-Hispanic whites. We found the direction of effect in the first three cohorts was consistent with the direction of effect in COPDGene more often than expected by chance alone (P = 0•03). This result suggests additional signals of significance may be found in larger GWAS, and are consistent with a recent analysis of COPDGene data¹⁸. As with most GWAS studies, the effect sizes of these identified loci are relatively small; however one subject may carry multiple risk loci. Within the COPDGene non-Hispanic whites, each additional copy of a risk allele within a composite risk score resulted in an increase in odds for COPD of 1•24; this estimate was similar whether the model included only loci previously discovered in studies not including COPDGene (e.g. 15q25, HHIP, and FAM13A loci), or included the additional loci (RIN3, TGFB2, MMP3/12) described in this study.

To further explore additional signals not reaching genome-wide significance, we additionally performed a gene-based analyses – under the hypothesis that a given gene or genic region may harbor multiple susceptibility variants with p-values larger than the traditional GWAS significance level – with VEGAS, and a SNP and text-mining based analyses – identifying and prioritizing genes based on functional relationships identified using literature – with GRAIL. The top genes from the VEGAS analysis, using a Bonferroni correction for 17,640 genes, included previously implicated loci (FAM13A; CHRNB4, IREB2, CHRNA3/5, HYKK, and PSMA4 at 15q25); as well as RIN3 andAPOBR (Table S9). For the GRAIL analysis (Table S10), the top individual genes were OSM and OSMR; in contrast, genes at or near well-validated loci – HHIP, IREB2, andFAM13A – did not give significant P-values in the GRAIL analysis.

Discussion

In a large, genome-wide association meta-analysis of moderate to severe and severe COPD (and the first genome-wide association analysis to include African-Americans), we confirmed three previously described genome-wide significant loci, and identified three additional loci achieving genome-wide significance in moderate-to-severe and severe COPD. Our findings provide further evidence for a role of common genetic variants in contributing to COPD susceptibility (panel).

The association at 11q22 is located in a cluster of matrix metalloproteinases including MMP12 (matrix metalloproteinase 12, also known as macrophage metalloelastase or matrix metallopeptidase 12). MMP12 is produced by macrophages and degrades elastin, and has been extensively characterized in COPD both in mouse models⁵⁰ and in human studies^51,52. Several studies have described genetic associations with COPD or lung function for a SNP in the promoter region of MMP12, rs2276109 [-82A→G], where the minor allele leads to decreased promoter activity through less efficient binding of AP-1^46,53–55. In a combined analysis of a total of 7 cohorts, including subjects with both asthma and COPD, the minor allele (G) of rs2276109 was associated with improved lung function⁴⁶. Of note, two of the COPD cohorts included in this study were enriched for severe disease. Similarly, in a study of 977 European cases and 876 controls, an association was identified for a haplotype including rs2276109 in MMP12 among severe cases (P = 0•0039) ⁵⁴. SNP rs626750 is in strong LD with rs2276109 (r² = 0•63). Our study thus confirms, with the same direction of effect, these previously described associations at genome-wide significance, and supports a role for MMP12 in severe COPD.

Meta-analyses across large population-based cohorts have previously reported an association at 1q41, near TGFB2, with FEV₁ /FVC ratio⁵⁶. However, the lead SNP for this association, rs993925, is not in strong LD (r²=0•027 in EUR) and lies over 250kb away from the SNPs reported here. Our top association is, however, in strong LD (r²=0•97 in EUR) with rs6684205, recently identified as an expression quantitative trait loci (eQTL) for TGFB2 in lung tissue⁵⁷. The COPD risk allele has been associated with decreased expression, consistent with our findings of decreased TGFB2 expression in lung tissue from severe COPD cases versus controls. These lines of evidence strongly suggest effects of this locus on COPD susceptibility operate via changes in lung TGFB2 expression. While genetic variants in or near TGFB1 have been studied in association with COPD^58–60, an association of variants near TGFB2 with COPD has not been previously described. Tgfb2 null mice have dilated conducting airways and collapsed terminal and respiratory bronchioles⁶¹, and loss-of-function mutations in TGFB2 have been associated with Loeys-Dietz syndrome, a disorder of connective tissue showing phenotypic overlap with Marfan syndrome, and has rarely been associated with emphysema⁶². TGF-β₂ is also the predominant isoform present in airway tissue in severe asthma^63,64; it is secreted in airway epithelial cells in response to injury or inflammatory cytokines (e.g. IL-13) and appears to play a major role in airway inflammation and remodeling ^65–68.

The association at the RIN3 locus, while genome-wide significant in the overall analysis, was not significant in ICGN. This finding thus may represent a false positive, but it also may be due to the lower power of the family-based analysis. In support of the latter explanation, an alternative analysis using generalized estimating equations (which allows calculation of effect sizes) resulted in an odds ratio of 1•14 (95% confidence interval, 0•92-1•41), consistent with the estimates from our other cohorts. In addition, a lookup of a SNP in strong LD (rs17184313, r² = 0•94 in EUR) in a recently published meta-analysis of COPD identified from population-based studies²² demonstrates nominal evidence of significance (P = 0•009), though the direction of association was not given. RIN3 is a Rab5 GTPase binding protein expressed in many tissues, including the lung, and is involved in transport from plasma membranes to early endosomes^69,70. High levels of expression of RIN3 have been found on human mast cells⁷¹, a cell type that may be of interest in COPD^72–74. Furthermore, we demonstrated, in a small number of lung tissue samples, that RIN3 expression differs between COPD cases versus controls. While RIN3 is the closest gene to the lead SNP, this locus is also approximately 1•7 megabases away from SERPINA1, the gene encoding alpha-1 antitrypsin, which could suggest an effect of distant rare variants⁷⁵. For the loci reported in this study, and for most loci reported for GWAS, the role of candidate SNP(s) on a particular gene and on protein function cannot be deduced with certainty from linkage disequilibrium patterns and simple measures of gene expression, and requires further functional investigation including SNP-based functional studies^76,77.

The 19q13 locus did not achieve genome-wide significant in this study, despite being identified in our prior meta-analysis in the ECLIPSE, NETT/NAS, Norway (GenKOLS), and the initial 1000 non-Hispanic White subjects from the COPDGene study⁷. In the current analysis of moderate-to-severe disease, rs7937 (nearest gene, RAB4B) was just below genome-wide significance (6•2×10⁻); however, the association was genome-wide significant (1•0×10⁻⁹) in a model adjusting only for principal components of genetic ancestry, and more significant when limited to non-Hispanic whites. A recent study in a Japanese population confirmed an association with smoking behavior with SNPs in this region⁷⁸, and additional analyses of nicotine addiction and lung eQTLs suggest effects at this locus may be mediated through several different variants in CYP2A6 as well as EGLN2 ^79–81. Together, these data suggest effects of the 19q13 locus on COPD act through a mechanism involving cigarette smoking, and are complex, potentially in the presence of locus heterogeneity across populations.

Our gene-based analysis using VEGAS identified an association with APOBR, the apolipoprotein B receptor. Lipoproteins may have pathophysiologic importance in lung disease ⁸²; differences in apolipoprotein B have been described in association with lung function and COPD^83,84, and a recent study identified an association between SNPs near APOM, lung function, and emphysema⁸⁵. Similarly, our GRAIL analysis suggested a role for oncostatin M and its receptor, which may be of interest in COPD and emphysema^86,87. Additional studies will be needed to confirm these findings.

Racial differences in COPD may exist^88,89. Thus, we also examined the case-control results only in the African-Americans. While underpowered, these results did not reveal any novel genome-wide significant loci (Table S8); furthermore, at the loci described in this work, there was no convincing evidence of heterogeneity (Tables 3 and 4) or differential effect sizes compared with those in non-Hispanic whites (Table S4 and S5). These data are consistent with a prior report finding little evidence that the relationship of smoking to lung function differed by genetic ancestry⁹⁰, as well as genetic studies of other traits that have demonstrated overall similarities of loci shared between ethnically diverse groups^91,92. While these results support our decision to combine the African-Americans and non-Hispanic whites to improve statistical power, our results should not be interpreted to imply including other ethnic groups is generally redundant; indeed, genetic studies in specific ethnic groups have led to discovery of novel loci⁹³ and provided important information for identifying specific variants at individual loci^91,94.

Our study does not address other genetic contributors to COPD susceptibility. We did not, for example, consider gene-gene interactions or gene-environment interaction. The genotyping and imputation in this study are not well-suited to address the role of rare variants, which may also be important in explaining COPD susceptibility^95–97. Our definition of cases and controls was based only on the presence of moderate, or moderate-to-severe airflow obstruction, yet COPD is highly heterogeneous. Analysis of individual characteristics (e.g. emphysema) or of specific subtypes (e.g. severe disease, as we demonstrate here; radiographically defined subsets; or separate GOLD categories^2,98) may provide greater insight into the development of this complex and heterogeneous disease^27–30. Well-powered studies of lung function in the general population, as well as COPD ascertained through population-based studies, have not identified several of the loci reported here^{22,56,99–101}. Additional studies will be helpful in determining whether heterogeneity in COPD definitions, including varying degrees of severity and case ascertainment, differential effects of genetic variants in disease versus lung function in the general population, or Type 1 or Type 2 error could account for these discrepancies. The number of loci achieving genome-wide significance described here for COPD is few compared to other complex diseases^102,103, and the markers described here account for a very small fraction of the estimated heritability¹⁸. For unknown reasons, the number of discovered loci confirmed by GWAS for any given sample size can vary widely^104,105.

However, despite differences in this ‘rate of return’, increasing sample size appears critical to discovering novel loci.

While the effect of an individual genetic variant may be small, discounting small effects in GWAS as unimportant would have ignored such critical effects as the insulin gene (INS) in diabetes and the HMG-CoA reductase gene (HMGCR) in cholesterol metabolism^106,107; more dramatic perturbations of these causal genes – through experimental disruption⁷⁶ or through identification of rare, more deleterious genetic variants¹⁰⁸ – can highlight the importance of pathophysiology identified by GWAS. In addition, cumulative effects of these loci may be substantial. Although more accurate risk prediction estimates will require assessments in independent populations, the increased odds of 1·24 with each COPD GWAS risk allele in the COPDGene population suggest that harboring three risk alleles could nearly double odds of moderate to severe COPD. By comparison, the population-based BOLD study¹⁰⁹ estimated an odds ratio for COPD per ten pack-years of smoking from1•16 to 1•28. These data, together with previous studies of familial aggregation and heritability of COPD, highlight the importance of genetic risk factors apart from alpha-1 antitrypsin deficiency in increasing risk of COPD.

Our work provides strong statistical support for association with moderate-to-severe or severe COPD susceptibility for three previously-described^9–17 (CHRNA3/5/IREB2, HHIP, and FAM13A) and three additional (RIN3, MMP3/MMP12, TGFB2) loci. We provide evidence that additional GWAS in larger samples are likely to identify additional genetic determinants of COPD, and suggest using subsets of COPD (such as severe disease) may provide additional insight to genetic risk factors. Our work also suggests further studies to elucidate biological mechanisms⁷⁷, which we hope will reveal new insights into COPD pathogenesis, and ultimately, treatment for this important disease.

Panel – Research in context.

Systematic Review

The genetic risk factors for COPD are still largely unknown. We searched PubMed with the search terms “genome-wide association” and “COPD” or “airflow”, as well as the genome-wide association study (GWAS) catalog (genome.gov/26525384). At the time of our search, the largest studies to date included approximately 3,500 cases. Evidence from GWAS in other diseases suggests larger sample sizes or analysis of specific subtypes could increase power and identify new genetic determinants of COPD.

Interpretation

Our study in moderate-to-severe and severe COPD confirms genome-wide associations near FAM13A, HHIP, and CHRNA3/CHRNA5/IREB2, and provides evidence in support of new associations near RIN3, MMP12 and TGFB2. GWAS continues to have potential to identify new genetic risk factors that could implicate novel disease mechanisms in COPD. Genetic variants, apart from alpha-1 antitrypsin deficiency, increase the risk of COPD; this burden may be higher in those with severe disease.

Appendix

Author Contributions

Study design: M.H.C., M.N.M., M.M., P.J.C., N.L., C.L., E.K.S., T.H.B. Data acquisition and quality control: M.H.C., X.Z., P.J.C., C.P.H, D.L.D., J.S.S., J. Z., A.A.L., D.S., R.C., R.G.B., E.A.R., B.J.M., J.E.H., T.M., H.F., J.B.H., R.T-S., D.A.L., P.B., A.G., J.D.C., E.K.S., T.H.B. Data Analysis: M.H.C., M.N.M, M.M., T.M., J.B.H., E.K.S., T.H.B. Critical revision of manuscript: all authors.

COPDGene^® Investigators - Core Units

Administrative Core: James Crapo, MD (PI), Edwin Silverman, MD, PhD (PI), Barry Make, MD, Elizabeth Regan, MD, Sara Penchev, Rochelle Lantz, Lori Stepp, Sandra Melanson,

Genetic Analysis Core: Terri Beaty, PhD, Barbara Klanderman, PhD, Nan Laird, PhD, Christoph Lange, PhD, Michael Cho, MD, Stephanie Santorico, PhD, John Hokanson, MPH, PhD, Dawn DeMeo, MD, MPH, Nadia Hansel, MD, MPH, Craig Hersh, MD, MPH, Peter Castaldi, MD, MSc, Merry-Lynn McDonald, PhD, Jin Zhou, MD, PhD, Manuel Mattheisen, MD, Emily Wan, MD, Megan Hardin, MD, Jacqueline Hetmanski, MS, Margaret Parker, MS, Tanda Murray, MS

Imaging Core: David Lynch, MB, Joyce Schroeder, MD, John Newell, Jr., MD, John Reilly, MD, Harvey Coxson, PhD, Philip Judy, PhD, Eric Hoffman, PhD, George Washko, MD, Raul San Jose Estepar, PhD, James Ross, MSc, Mustafa Al Qaisi, MD, Jordan Zach, Alex Kluiber, Jered Sieren, Tanya Mann, Deanna Richert, Alexander McKenzie, Jaleh Akhavan, Douglas Stinson

PFT QA Core, National Jewish Health: Robert Jensen, PhD

Biological Repository, Johns Hopkins University, Baltimore, MD: Homayoon Farzadegan, PhD, Stacey Meyerer, Shivam Chandan, Samantha Bragan

Data Coordinating Center and Biostatistics, National Jewish Health, Denver, CO: Douglas Everett, PhD, Andre Williams, PhD, Carla Wilson, MS, Anna Forssen, MS, Amber Powell, Joe Piccoli

Epidemiology Core, University of Colorado School of Public Health, Denver, CO: John Hokanson, MPH, PhD, Marci Sontag, PhD, Jennifer Black-Shinn, MPH, Gregory Kinney, MPH, PhDc, Sharon Lutz, MPH, PhD

COPDGene^® Investigators: Clinical Centers Ann Arbor VA: Jeffrey Curtis, MD, Ella Kazerooni, MD

Baylor College of Medicine, Houston, TX: Nicola Hanania, MD, MS, Philip Alapat, MD, Venkata Bandi, MD, Kalpalatha Guntupalli, MD, Elizabeth Guy, MD, Antara Mallampalli, MD, Charles Trinh, MD, Mustafa Atik, MD, Hasan Al-Azzawi, MD, Marc Willis, DO, Susan Pinero, MD, Linda Fahr, MD, Arun Nachiappan, MD, Collin Bray, MD, L. Alexander Frigini, MD, Carlos Farinas, MD, David Katz, MD, Jose Freytes, MD, Anne Marie Marciel, MD

Brigham and Women's Hospital, Boston, MA: Dawn DeMeo, MD, MPH, Craig Hersh, MD, MPH, George Washko, MD, Francine Jacobson, MD, MPH, Hiroto Hatabu, MD, PhD, Peter Clarke, MD, Ritu Gill, MD, Andetta Hunsaker, MD, Beatrice Trotman-Dickenson, MBBS, Rachna Madan, MD

Columbia University, New York, NY: R. Graham Barr, MD, DrPH, Byron Thomashow, MD, John Austin, MD, Belinda D'souza, MD

Duke University Medical Center, Durham, NC: Neil MacIntyre, Jr., MD, Lacey Washington, MD, H Page McAdams, MD

Reliant Medical Group, Worcester, MA: Richard Rosiello, MD, Timothy Bresnahan, MD, Joseph Bradley, MD, Sharon Kuong, MD, Steven Meller, MD, Suzanne Roland, MD

Health Partners Research Foundation, Minneapolis, MN: Charlene McEvoy, MD, MPH, Joseph Tashjian, MD

Johns Hopkins University, Baltimore, MD: Robert Wise, MD, Nadia Hansel, MD, MPH, Robert Brown, MD, Gregory Diette, MD, Karen Horton, MD

Los Angeles Biomedical Research Institute at Harbor UCLA Medical Center, Torrance, CA: Richard Casaburi, MD, PhD, Janos Porszasz, MD, PhD, Hans Fischer, MD, Matt Budoff, MD

Michael E. DeBakey VAMC, Houston, TX: Amir Sharafkhaneh, MD, Charles Trinh, MD, Hirani Kamal, MD, Roham Darvishi, MD, Marc Willis, DO, Susan Pinero, MD, Linda Fahr, MD, Arun Nachiappan, MD, Collin Bray, MD, L. Alexander Frigini, MD, Carlos Farinas, MD, David Katz, MD, Jose Freytes, MD, Anne Marie Marciel, MD

Minneapolis VA: Dennis Niewoehner, MD, Quentin Anderson, MD, Kathryn Rice, MD, Audrey Caine, MD

Morehouse School of Medicine, Atlanta, GA: Marilyn Foreman, MD, MS, Gloria Westney, MD, MS, Eugene Berkowitz, MD, PhD

National Jewish Health, Denver, CO: Russell Bowler, MD, PhD, David Lynch, MB, Joyce Schroeder, MD, Valerie Hale, MD, John Armstrong, II, MD, Debra Dyer, MD, Jonathan Chung, MD, Christian Cox, MD

Temple University, Philadelphia, PA: Gerard Criner, MD, Victor Kim, MD, Nathaniel Marchetti, DO, Aditi Satti, MD, A. James Mamary, MD, Robert Steiner, MD, Chandra Dass, MD, Libby Cone, MD

University of Alabama, Birmingham, AL: William Bailey, MD, Mark Dransfield, MD, Michael Wells, MD, Surya Bhatt, MD, Hrudaya Nath, MD, Satinder Singh, MD

University of California, San Diego, CA: Joe Ramsdell, MD, Paul Friedman, MD

University of Iowa, Iowa City, IA: Alejandro Cornellas, MD, John Newell, Jr., MD, Edwin JR van Beek, MD, PhD

University of Michigan, Ann Arbor, MI: Fernando Martinez, MD, MeiLan Han, MD, Ella Kazerooni, MD

University of Minnesota, Minneapolis, MN: Christine Wendt, MD, Tadashi Allen, MD

University of Pittsburgh, Pittsburgh, PA: Frank Sciurba, MD, Joel Weissfeld, MD, MPH, Carl Fuhrman, MD, Jessica Bon, MD, Danielle Hooper, MD

University of Texas Health Science Center at San Antonio, San Antonio, TX: Antonio Anzueto, MD, Sandra Adams, MD, Carlos Orozco, MD, Mario Ruiz, MD, Amy Mumbower, MD, Ariel Kruger, MD, Carlos Restrepo, MD, Michael Lane, MD Principal investigators and centers participating in ECLIPSE (NCT00292552, SCO104960) include: Bulgaria: Y. Ivanov, Pleven; K. Kostov, Sofia. Canada: J. Bourbeau, Montreal; M. Fitzgerald, Vancouver; P. Hernández, Halifax; K. Killian, Hamilton; R. Levy, Vancouver; F. Maltais, Montreal; D. O'Donnell, Kingston. Czech Republic: J. Krepelka, Praha. Denmark: J. Vestbo, Hvidovre. The Netherlands: E. Wouters, Horn. New Zealand: D. Quinn, Wellington. Norway: P. Bakke, Bergen, Slovenia: M. Kosnik, Golnik. Spain: A. Agusti, Jaume Sauleda, Palma de Mallorca. Ukraine: Y. Feschenko, Kiev; V. Gavrisyuk, Kiev; L. Yashina, Kiev. UK: L. Yashina, W. MacNee, Edinburgh; D. Singh, Manchester; J. Wedzicha, London. USA: A. Anzueto, San Antonio, TX; S. Braman, Providence. RI; R. Casaburi, Torrance CA; B. Celli, Boston, MA; G. Giessel, Richmond, VA; M.

Gotfried, Phoenix, AZ; G. Greenwald, Rancho Mirage, CA; N. Hanania, Houston, TX; D. Mahler, Lebanon, NH; B. Make, Denver, CO; S. Rennard, Omaha, NE; C. Rochester, New Haven, CT; P. Scanlon, Rochester, MN; D. Schuller, Omaha, NE; F. Sciurba, Pittsburgh, PA; A. Sharafkhaneh, Houston, TX; T. Siler, St Charles, MO; E. Silverman, Boston, MA; A. Wanner, Miami, FL; R. Wise, Baltimore, MD; R. ZuWallack, Hartford, CT.

Steering Committee: H. Coxson (Canada), C. Crim (GlaxoSmithKline, USA), L. Edwards (GlaxoSmithKline, USA), D. Lomas (UK), W. MacNee (UK), E. Silverman (USA), R. Tal Singer (Co-chair, GlaxoSmithKline, USA), J. Vestbo (Co-chair, Denmark), J. Yates (GlaxoSmithKline, USA).

Scientific Committee: A. Agusti (Spain), P. Calverley (UK), B. Celli (USA), C. Crim (GlaxoSmithKline, USA), B. Miller (GlaxoSmithKline, USA), W. MacNee (Chair, UK), S. Rennard (USA), R. Tal-Singer (GlaxoSmithKline, USA), E. Wouters (The Netherlands), J. Yates (GlaxoSmithKline, USA).

Co-investigators in the NETT Genetics Ancillary Study also include J. Benditt, G. Criner, M. DeCamp, P. Diaz, M. Ginsburg, L. Kaiser, M. Katz, M. Krasna, N. MacIntyre, R. McKenna, F. Martinez, Z. Mosenifar, J. Reilly, A. Ries, P. Scanlon, F. Sciurba and J. Utz.

International COPD Genetics Network (ICGN) investigators: Edwin K. Silverman, Brigham & Women's Hospital, Boston, MA, USA; David A. Lomas, Cambridge Institute for Medical Research, University of Cambridge, Cambridge, UK; Barry J. Make, National Jewish Medical and Research Center, Denver, CO, USA; Alvar Agusti and Jaume Sauleda, Hospital Universitari Son Dureta, Fundación Caubet-Cimera and Ciber Enfermedades Respiratorias, Spain; Peter M.A. Calverley, University of Liverpool, UK; Claudio F. Donner, Division of Pulmonary Disease, S. Maugeri Foundation, Veruno (NO), Italy; Robert D. Levy, University of British Columbia, Vancouver, Canada; Peter D. Paré, University of British Columbia, Vancouver, Canada; Stephen Rennard, Section of Pulmonary & Critical Care, University of Nebraska Medical Center, Omaha, NE, USA; Jørgen Vestbo, Department of Cardiology and Respiratory Medicine, Hvidovre Hospital, Copenhagen, Denmark; Emiel F.M. Wouters, University Hospital Maastricht, The Netherlands.