The SERPINE2 Gene Is Associated with Chronic Obstructive Pulmonary Disease in Two Large Populations

Abstract

Section:

Rationale: Chronic obstructive pulmonary disease (COPD) is a complex disease influenced by multiple genes and environmental factors. A region on chromosome 2q has been shown to be linked to COPD. A positional candidate gene from the chromosome 2q region SERPINE2 (Serpin peptidase inhibitor, clade E [nexin, plasminogen activator inhibitor type 1], member 2), was previously evaluated as a susceptibility gene for COPD in two association studies, but the results were contradictory.

Objectives: To identify the relationship between SERPINE2 polymorphisms and COPD-related phenotypes using family-based and case-control association studies.

Methods: In the present study, we genotyped 25 single nucleotide polymorphisms (SNPs) from SERPINE2 and analyzed qualitative and quantitative COPD phenotypes in 635 pedigrees with 1,910 individuals and an independent case-control population that included 973 COPD cases and 956 control subjects. The family data were analyzed using family-based association tests. The case-control data were analyzed using logistic regression and linear models.

Measurements and Main Results: Six SNPs demonstrated significant associations with COPD phenotypes in the family-based association analysis (0.0016 ⩽ p ⩽ 0.042). Five of these SNPs demonstrated replicated associations in the case-control analysis (0.021 ⩽ p ⩽ 0.031). In addition, the results of haplotype analyses supported the results from single SNP analyses.

Conclusions: These data provide further support for SERPINE2 as a COPD susceptibility gene.

Keywords:

chronic obstructive pulmonary disease; genetic association, SERPINE2; replication

AT A GLANCE COMMENTARY

Scientific Knowledge on the Subject

A positional candidate gene from the chromosome 2q region SERPINE2 (Serpin peptidase inhibitor, clade E [nexin, plasminogen activator inhibitor type 1], member 2), was previously evaluated as a susceptibility gene for COPD in two association studies, but the results were contradictory.

What This Study Adds to the Field

SERPINE2 demonstrates significant positive association with COPD phenotypes.

Chronic obstructive pulmonary disease (COPD) is one of the leading causes of death and a major source of morbidity. Cigarette smoking is the most important risk factor for COPD, but only a minority of smokers develops the disease (1). Twin studies have demonstrated that pulmonary function is heritable (2, 3). Furthermore, familial aggregation studies suggested a genetic component of COPD susceptibility (4, 5). Genomewide linkage analysis of the Boston Early-Onset COPD Study families using pulmonary function phenotypes demonstrated a significant linkage peak on chromosome 2q (6–8). Linkage of spirometric phenotypes in general-population pedigrees to this region of chromosome 2q has also been reported (9). Using expression-array analysis of murine and human lung tissues, DeMeo and colleagues identified SERPINE2 (Serpin peptidase inhibitor, clade E [nexin, plasminogen activator inhibitor type 1], member 2) as a novel candidate COPD-susceptibility gene on chromosome 2q (10). Association analysis of the Boston Early-Onset COPD Study families showed significant association of multiple SERPINE2 single nucleotide polymorphisms (SNPs) with COPD-related phenotypes. DeMeo and colleagues also replicated these association results in a case-control study (10). However, Chappell and colleagues tested five SERPINE2 SNPs and were unable to replicate the SERPINE2 associations in a large case-control study (11). Because there are contradictory results regarding the association of SERPINE2 with COPD in these previous reports, we conducted genetic association analyses of SERPINE2 SNPs with both qualitative and quantitative COPD-related phenotypes in two large, independent datasets. We genotyped 25 SNPs in the SERPINE2 gene in 635 pedigrees with 1,910 individuals from the International COPD Genetics Network (ICGN), which is the largest family-based COPD collection reported to date. We also analyzed these SNPs in a large case-control collection from Bergen, Norway, which includes 973 cases with COPD and 956 control subjects.

METHODS

Section:

Subject Recruitment and Clinical Evaluation

In the multicenter ICGN study, subjects with known COPD were recruited as probands, and siblings and available parents were ascertained through the probands. Inclusion criteria for probands were airflow limitation (post-bronchodilator FEV₁< 60% predicted and FEV₁/VC < 90% predicted) at a relatively early age (45 to 65 yr), smoking history of 5 or more pack-years, and at least one eligible sibling (with ⩾ 5 pack-year smoking history). COPD in siblings was defined by a post-bronchodilator FEV₁ less than 80% of predicted and FEV₁/VC less than 90% of predicted values. A total of 1,910 white individuals from 635 pedigrees were included in the family-based association analysis.

Subjects for the case-control study were recruited from Bergen, Norway. The entry criteria for COPD cases were FEV₁ of less than 80% predicted and FEV₁/FVC of less than 0.7. The control subjects were selected based on FEV₁ of more than 80% predicted and FEV₁/FVC of more than 0.7. A total of 973 cases with COPD and 956 control subjects were included in the present analysis. Characteristics of the subjects are shown in Table 1. Additional details can be found in the online supplement.

TABLE 1. **CLINICAL CHARACTERISTICS FOR TWO STUDY POPULATIONS**
	ICGN Family Data			COPD Case-Control Data
	Probands	Siblings	Cases	Controls
Subjects, n	610	1,300	973	956
Age, yr (± SD)	58.39 (± 5.46)	58.08 (± 9.83)	65.56 (± 10.11)	55.48 (± 9.69)
Female, n (%)	247 (40.49)	648 (49.84)	383 (39.20)	495 (50.61)
Post-FEV₁, L (± SD)	1.11 (± 0.44)	2.36 (± 0.98)	1.58 (± 0.71)	3.24 (± 0.73)
Post-FEV₁, % pred (± SD)	36.26 (± 12.94)	77.48 (± 25.90)	50.34 (± 17.48)	94.05 (± 9.18)
Post-FEV₁/FVC ratio (± SD)	0.37 (± 0.12)	0.61 (± 0.15)	0.51 (± 0.12)	0.79 (± 0.04)
Height, cm (± SD)	167.85 (± 9.45)	167.66 (± 9.54)	170.08 ± 9.02	171.77 ± 8.78
Pack-years of smoking (± SD)	51.48 (± 26.75)	40.61 (± 24.90)	31.98 (± 19.16)	19.58 (± 13.18)
Current smokers, n (%)	207 (33.93%)	661 (50.84%)	465 (47.69%)	409 (41.82%)

Definition of abbreviations: COPD = chronic obstructive pulmonary disease; ICGN = International COPD Genetics Network.

FEV₁/VC was used for the ICGN family data, with VC determined by the higher of FVC and SVC (slow vital capacity).

SNP Selection and Genotyping

Twenty-three SNPs within SERPINE2 and two more SNPs extending into the genes flanking SERPINE2 were selected based on their map positions. These 25 SNPs were genotyped in the ICGN family population and also in the Norwegian case-control population using the Illumina array-based custom SNP genotyping platform (Illumina, Inc., www.illumina.com). Hardy-Weinberg equilibrium for all SNPs was tested in control subjects by using the chi-square goodness-of-fit test to determine if the observed distribution of genotype frequencies was incompatible with the Hardy-Weinberg equilibrium distribution. Population stratification was assessed for the case-control population by genotyping 221 unlinked SNPs and estimating an inflation factor λ for genomic control (12, 13). The PedCheck program was used to detect Mendelian inconsistencies in the genotype data of COPD families (14). Additional details can be found in the online supplement.

Statistical Methods for Association Analysis

FBAT version 1.7.1 (15) was used for the family-based single-SNP association analysis of the mild airflow obstruction phenotype in the ICGN family study. For convenience, “the mild airflow obstruction phenotype” is referred to as “COPD” below. The analyses of quantitative traits (FEV₁ and FEV₁/VC) were performed with covariates including center, age, sex, height, and pack-years of cigarette smoking using PBAT version 3.1 (16). Biallelic tests were conducted for SNPs using an additive genetic model. p Values from the analyses of less than 0.05 were considered as statistically significant. p Values were multiplied by the number of markers analyzed to identify the markers that remain significant after a very conservative Bonferroni correction for multiple statistical testing.

In the case-control population, two models were used in the association analysis. A logistic regression model for the COPD binary phenotype and a linear regression model for the quantitative phenotypes (FEV₁ and FEV₁/FVC), with covariates including age, sex, and pack-years of smoking. For the quantitative trait analysis, only cases with COPD were included. The analyses were done using SAS software 8.2 (SAS Institute, Cary, NC) with an additive genetic model.

Haplotype analyses were conducted using the HBAT function of the FBAT program with the use of Monte Carlo sampling for COPD (17), and using UNPHASED version 3.0 for FEV₁ and FEV₁/VC (18) in the family data. In the case-control data, haplotype analysis was performed using the expectation-maximization algorithm and score tests, implemented in Haplo.stats program, version 1.2.1 (19).

The linkage disequilibrium (LD) structure in the SERPINE2 region was examined with the program Haploview, version 3.3 (20, 21). Additional details can be found in the online supplement.

RESULTS

Section:

Single Marker Association Analyses

We evaluated 25 SNPs in the SERPINE2 region; the locations and characteristics of those SNPs are summarized in Table 2. In ICGN family-based association analysis, six SNPs in SERPINE2 demonstrated significant association with COPD and/or the quantitative spirometric phenotypes, FEV₁ and FEV₁/VC (p < 0.05) (Table 3). Five of these SNPs—rs6734100, rs729631, rs975278, rs7583463, and rs6748795—were significantly associated with COPD (0.0016 ⩽ p ⩽ 0.033); SNP rs6734100 showed the most significant association with COPD (p = 0.0016) and was significant even after a Bonferroni correction (p = 0.0016 × 25 = 0.04) for the number markers analyzed. Another SNP, rs16865390, was significantly associated with FEV₁/VC (p = 0.042) (Table 3).

TABLE 2. **TWENTY-FIVE SINGLE NUCLEOTIDE POLYMORPHISMS IN THE *SERPINE2* REGION GENOTYPED IN THIS STUDY**
SNP ID^*	SNP	Position (NCBI 36)^†	Region	Alleles	MAF in Family Data	MAF in Cases	MAF in Controls
1	rs10980	224657851	3′ UTR	G/C	0.443	0.440	0.447
2	rs6754561	224665201	3′ UTR	T/C	0.381	0.360	0.351
3	rs6734100	224667500	Intron	C/G	0.155	0.109	0.127
4	rs729631	224670424	Intron	G/C	0.197	0.151	0.160
5	rs16865390	224671520	Intron	A/C	0.055	0.020	0.020
6	rs975278	224673212	Intron	G/A	0.199	0.150	0.160
7	rs7583463	224674614	Intron	C/A	0.233	0.168	0.179
8	rs6748795	224676228	Intron	C/G	0.232	0.168	0.178
9	rs3795880	224676812	Intron	G/C	0.326	0.317	0.308
10	rs6712954	224682155	Exon	A/G	0.068	0.030	0.039
11	rs2076924	224682549	Intron	C/T	0.348	0.314	0.302
12	rs16865421	224683598	Intron	A/G	0.092	0.093	0.077
13	rs6721140	224684148	Intron	G/A	0.351	0.314	0.301
14	rs11695803	224686666	Intron	A/G	0.100	0.101	0.087
15	rs3795879	224688326	Intron	G/A	0.231	0.176	0.189
16	rs6747096	224688347	Exon	A/G	0.222	0.174	0.185
17	rs10196778	224689101	Intron	A/G	0.226	0.176	0.189
18	rs13392412	224689945	Intron	T/C	0.223	0.174	0.185
19	rs3795877	224691682	Intron	C/T	0.222	0.175	0.185
20	rs3795875	224691970	ILE > MET	C/G	0.001	0.001	0.001
21	rs1530020	224696875	Intron	G/T	0.298	0.274	0.264
22	rs7590948	224699195	Intron	G/A	0.487	0.458	0.457
23	rs4674839	224704712	Intron	C/T	0.117	0.103	0.092
24	rs920251	224718450	Intron	T/C	0.360	0.338	0.345
25	rs1438831	224731820	Exon	T/C	0.357	0.331	0.338

Definition of abbreviations: MAF = minor allele frequency; SNP = single nucleotide polymorphism; UTR = untranslated region.

^*SNP 1 (rs10980) and SNP 25 (rs1438831) are located in genes MRPL44 and LOC440935, respectively; the other SNPs are in SERPINE2 gene.

^†Base pair position in NCBI build 36.

TABLE 3. **RESULTS OF ASSOCIATION OF INDIVIDUAL *SERPINE2* SINGLE NUCLEOTIDE POLYMORPHISMS WITH CHRONIC OBSTRUCTIVE PULMONARY DISEASE PHENOTYPES IN INTERNATIONAL COPD GENETICS NETWORK FAMILY DATA**
			No. of Informative Families	p Value
SNP ID	SNP	Risk Allele	No. of Informative Families		COPD	FEV₁	FEV₁/VC
1	rs10980	C	194	0.622	0.601	0.640
2	rs6754561	T	182	0.949	0.719	0.887
3	rs6734100	C	118	0.0016	0.348	0.260
4	rs729631	G	145	0.024	0.451	0.583
5	rs16865390	A	43	0.368	0.068	0.042
6	rs975278	G	146	0.025	0.475	0.628
7	rs7583463	C	161	0.030	0.079	0.198
8	rs6748795	C	161	0.033	0.097	0.206
9	rs3795880	C	181	0.650	0.096	0.297
10	rs6712954	A	56	0.699	0.848	0.996
11	rs2076924	T	184	0.816	0.586	0.311
12	rs16865421	G	64	0.424	0.945	0.928
13	rs6721140	A	184	0.765	0.607	0.306
14	rs11695803	G	77	0.564	0.757	0.606
15	rs3795879	G	138	0.982	0.325	0.174
16	rs6747096	A	138	0.975	0.200	0.061
17	rs10196778	A	139	0.919	0.202	0.079
18	rs13392412	T	138	0.876	0.204	0.094
19	rs3795877	C	137	0.948	0.200	0.061
20	rs3795875	G	1	NA	NA	NA
21	rs1530020	G	164	0.665	0.958	0.338
22	rs7590948	G	183	0.478	0.579	0.745
23	rs4674839	T	82	0.353	0.241	0.512
24	rs920251	T	184	0.844	0.196	0.336
25	rs1438831	T	186	0.595	0.464	0.769

Definition of abbreviations: COPD = chronic obstructive pulmonary disease; NA = not available due to the low allele frequency; SNP = single nucleotide polymorphism.

We attempted to replicate the results of the family-based association analysis in the case-control population. The results of the analyses to identify population stratification in the case-control study showed that the mean test statistic of the genomic control SNPs, inflation factor (λ), was 1.027; thus, no significant evidence of population stratification was found between the cases and control subjects. All SERPINE2 SNPs were in Hardy-Weinberg equilibrium in the control subjects (p > 0.05). In the COPD case-control data, six SNPs in SERPINE2 demonstrated significant association with COPD, FEV₁, and/or FEV₁/FVC (Table 4). Among them, rs16865421 was significantly associated with increased odds for COPD (p = 0.040; odds ratio, 1.33; 95% confidence interval, 1.01–1.74); rs6734100 was significantly associated with a reduction in FEV₁ in cases with COPD (p = 0.008); and five SNPs—rs6734100, rs729631, rs975278, rs7583463, and rs6748795—were significantly associated with reductions in FEV₁/FVC in cases with COPD (0.021 < p < 0.031) (Table 4). Thus, associations of five SNPs with COPD-related phenotypes were replicated in the ICGN family data and the case-control data.

TABLE 4. **RESULTS OF ASSOCIATION OF INDIVIDUAL *SERPINE2* SINGLE NUCLEOTIDE POLYMORPHISMS WITH CHRONIC OBSTRUCTIVE PULMONARY DISEASE PHENOTYPES IN CHRONIC OBSTRUCTIVE PULMONARY DISEASE CASE-CONTROL DATA**
			COPD			FEV₁
SNP ID	SNP	Risk Allele	OR (95% CI)^*	p Value	β (SE)^*	p Value	β (SE)	p Value
1	rs10980	C	0.92 (0.78–1.07)	0.289	0.012 (0.026)	0.631	0.003 (0.006)	0.533
2	rs6754561	T	1.07 (0.92–1.26)	0.343	−0.007 (0.027)	0.787	−0.004 (0.005)	0.485
3	rs6734100	C	0.82 (0.65–1.03)	0.098	−0.113 (0.042)	0.008	−0.021 (0.009)	0.021
4	rs729631	G	0.88 (0.71–1.08)	0.236	−0.050 (0.036)	0.111	−0.017 (0.008)	0.031
5	rs16865390	A	1.11 (0.65–1.89)	0.686	0.004 (0.040)	0.914	−0.005 (0.008)	0.566
6	rs975278	G	0.87 (0.71–1.08)	0.222	−0.06 (0.034)	0.097	−0.017 (0.008)	0.027
7	rs7583463	C	0.88 (0.72–1.07)	0.218	0.061 (0.034)	0.081	−0.017 (0.007)	0.025
8	rs6748795	C	0.88 (0.73–1.08)	0.225	−0.066 (0.035)	0.057	−0.018 (0.001)	0.021
9	rs3795880	C	1.09 (0.93–1.29)	0.295	−0.006 (0.028)	0.826	−0.002 (0.005)	0.787
10	rs6712954	A	0.74 (0.48–1.13)	0.163	−0.097 (0.079)	0.223	−0.028 (0.017)	0.107
11	rs2076924	T	1.05 (0.89–1.24)	0.545	−0.013 (0.028)	0.635	−0.005 (0.006)	0.348
12	rs16865421	G	1.33 (1.01–1.74)	0.040	−0.004 (0.040)	0.915	−0.005 (0.009)	0.566
13	rs6721140	A	1.06 (0.90–1.25)	0.505	−0.014 (0.028)	0.598	−0.006 (0.006)	0.314
14	rs11695803	G	1.20 (0.93–1.57)	0.155	0.003 (0.044)	0.934	0.003 (0.009)	0.774
15	rs3795879	G	0.86 (0.71–1.05)	0.150	−0.015 (0.034)	0.649	−0.008 (0.007)	0.302
16	rs6747096	A	0.88 (0.72–1.08)	0.219	−0.010 (0.034)	0.765	−0.006 (0.007)	0.403
17	rs10196778	A	0.87 (0.71–1.06)	0.167	−0.009 (0.034)	0.782	−0.006 (0.007)	0.393
18	rs13392412	T	0.88 (0.72–1.07)	0.207	−0.014 (0.035)	0.672	−0.007 (0.006)	0.395
19	rs3795877	C	0.89 (0.73–1.08)	0.233	−0.010 (0.033)	0.766	−0.006 (0.007)	0.401
20	rs3795875	G	NA	NA	NA	NA	NA	NA
21	rs1530020	G	0.99 (0.84–1.18)	0.981	0.016 (0.030)	0.589	0.003 (0.006)	0.568
22	rs7590948	G	0.94 (0.81–1.09)	0.419	−0.005 (0.027)	0.853	−0.002 (0.005)	0.718
23	rs4674839	T	1.09 (0.84–1.42)	0.536	0.007 (0.044)	0.878	−0.001 (0.009)	0.929
24	rs920251	T	0.97 (0.83–1.14)	0.743	−0.003 (0.028)	0.893	0.002 (0.006)	0.757
25	rs1438831	T	0.97 (0.82–1.14)	0.727	0.001 (0.027)	0.976	0.005 (0.006)	0.410

Definition of abbreviations: CI = confidence interval; COPD = chronic obstructive pulmonary disease; NA = not available due to the low allele frequency; OR = odds ratio; SNP = single nucleotide polymorphism.

^*Effect estimates are ORs and β (regression coefficients) for qualitative and quantitative phenotypes, respectively.

Haplotype Block Structure and LD

Figure 1 shows pairwise LD (r²) values for the 25 SNPs in the 74-kb sequence encompassing SERPINE2 and the flanking regions. Five haplotype blocks were revealed in ICGN family data and COPD case-control data, respectively. In ICGN family data, the significant SNPs—rs6734100, rs729631, rs16865390, rs975278, and rs7583463—were located in block 1, and rs6748795 was located in block 2. In the COPD case-control population, the significant SNPs—rs6734100, rs729631, rs975278, rs7583463, and rs6748795—were located in block 1, and rs16865421 in block 2. All of the replicated SNPs are in one single LD block in the case-control population. Because all significant SNPs are within the LD blocks in the SERPINE2 gene and not in the flanking regions, it is reasonable to speculate that the significant associations are from the SERPINE2 gene itself.

Figure 1. Linkage disequilibrium (LD) map across the SERPINE2 region. LD block structure of 25 single nucleotide polymorphisms (SNPs) within the SERPINE2 region in the International COPD Genetics Network family-based population (A) and COPD case-control population (B). Values of r² (× 100) are shown. Black squares, r² = 1; white squares, r² = 0; squares in shades of gray, 0 < r² < 1 (the intensity of the gray shading is proportional to r²). Haplotype block structure was estimated with the Haploview program, version 3.3.

Download Figure (Part 1) | Download Figure (Part 2) | Download Powerpoint

[More] [Minimize]

Haplotype Analyses

Using a sliding-window approach, we performed adjacent 2- and 3-SNP haplotype analyses in both ICGN family data and COPD case-control data to identify the regions that most likely contribute directly to the observed associations. We listed significant results of haplotype-based association analysis in the two study populations in Table 5; both global (p-global) and individual (p-specific) haplotype results are reported. In the family data, 11 adjacent SNP combinations with a significant score test were detected (0.0078 ⩽ p-specific ⩽ 0.0226 and 0.0062 ⩽ p-global ⩽ 0.0423) with COPD; for 2-SNP haplotypes, the haplotype including the SNPs rs6734100 and rs729631 showed the strongest association (p-global = 0.0062 and p-specific = 0.0078); for 3-SNP haplotypes, haplotypes including rs6734100, rs729631, and rs16865390 showed the most significant association (p-global = 0.0165, p-specific = 0.0082). We did not detect significant haplotype associations with FEV₁ and FEV₁/VC in the family data. In the case-control data, two haplotypes demonstrated significant association (0.0149 ⩽ p-specific ⩽ 0.0257 and 0.0296 ⩽ p-global ⩽ 0.0398) with FEV₁ among COPD cases; five haplotypes demonstrated significant association (0.0291 ⩽ p-specific ⩽ 0.0396 and 0.0383 ⩽ p-global ⩽ 0.0492) with FEV₁/FVC among cases with COPD for 2-SNP haplotypes. Five haplotypes demonstrated significant association (0.0297 ⩽ p-specific ⩽ 0.0479 and 0.0409 ⩽ p-global ⩽ 0.0491) with FEV₁/FVC for 3-SNP haplotypes. We did not detect significant haplotype associations with COPD in the case-control data. Significant results of haplotype analyses for FEV₁/FVC in the case-control data were similar to those of associated haplotypes for COPD in the family data.

TABLE 5. **SIGNIFICANT RESULTS FROM THE HAPLOTYPES ANALYSES**
Haplotype	P-Specific	P-Global	Haplo_Freq
COPD in ICGN Family Data
2–3	0.0143	0.0394	0.124
3–4	0.0078	0.0062	0.837
4–5	0.0095	0.0166	0.801
5–6	0.0182	0.0110	0.799
6–7	0.0129	0.0322	0.161
7–8	0.0226	0.0250	0.805
3–4–5	0.0082	0.0165	0.815
4–5–6	0.0126	0.0234	0.823
5–6–7	0.0221	0.0166	0.135
6–7–8	0.0128	0.0323	0.136
8–9–10	0.0089	0.0423	0.148
FEV₁ in COPD Case-Control Data
3–4	0.0149	0.0296	0.849
3–4–5	0.0257	0.0398	0.830
FEV₁/FVC in COPD Case-Control Data
3–4	0.0396	0.0408	0.849
4–5	0.0348	0.0383	0.847
5–6	0.0313	0.0432	0.848
7–8	0.0315	0.0451	0.823
8–9	0.0291	0.0492	0.165
3–4–5	0.0297	0.0429	0.846
4–5–6	0.0312	0.0422	0.848
5–6–7	0.0479	0.0491	0.151
6–7–8	0.0459	0.0485	0.150
7–8–9	0.0297	0.0409	0.166

Definition of abbreviations: COPD = chronic obstructive pulmonary disease; Haplo_Freq = frequency of the best specific haplotype; ICGN = International COPD Genetics Network.

Haplotypes with a global p < 0.05 are shown. P-specific represents the best specific haplotype p value. The relative position of the SNPs in the haplotypes combinations corresponds to the single nucleotide polymorphism IDs given in Table 2.

DISCUSSION

Section:

COPD is likely influenced by multiple genetic determinants, but severe α₁-antitrypsin deficiency is the only proven genetic risk factor. In the present study, we investigated whether SNPs in the SERPINE2 gene were associated with COPD-related phenotypes in two independent, large datasets. Our study confirmed the previously reported association between SERPINE2 SNPs and COPD-related phenotypes (10). By testing 25 SNPs, including 13 novel ones that were not included by DeMeo and coworkers (10), we found that six SNPs were significantly associated with COPD-related phenotypes in ICGN family data, and six SNPs were associated in COPD case-control data. Five of these six significant SNPs were replicated in both populations, and three of these five SNPs (rs6734100, rs729631, and rs975278) were shown to be associated with COPD-related phenotypes by DeMeo and colleagues (10). SNP rs6734100 showed the most significant evidence of association with COPD. Our haplotype analyses supported the results of single SNP analysis.

Consisting of nine exons, the SERPINE2 gene encodes a 44-kD thrombin and urokinase inhibitor and is the most important physiological regulator of α-thrombin in tissues (22). It is highly expressed and developmentally regulated in the nervous system where it is concentrated at neuromuscular junctions and central synapses in the hippocampus and striatum. Mansuy and colleagues found SERPINE2 expression in the embryonic mouse lung within the conducting-airway epithelium (23). DeMeo and colleagues showed that SERPINE2 was expressed highly in the developing mouse lung during alveogenesis, and that it was also expressed in airway epithelial cells and vascular adventitia of adult human lungs (10). In addition, several reports have shown that the expression of SERPINE2 is increased by tumor necrosis factor (TNF)–α, IL-1β, and transforming growth factor-β (24, 25). SERPINE2 belongs to the serpin family of proteins, as does α₁-antitrypsin. The deficiency of α₁-antitrypsin is the only known genetic cause of COPD. The results from the study reported by DeMeo and associates suggest that the overexpression of SERPINE2, rather than deficiency, is associated with COPD (10). Cross-talk mechanisms between serine proteases and metalloproteinases have been reported, particularly within the lung and in relation to the pathogenesis of emphysema (26). However, the physiological function of SERPINE2 and its role in the development of COPD remain to be characterized.

Association studies offer a potentially powerful approach to identify genetic variants that influence susceptibility to common complex diseases, but they are plagued by inconsistent results (27). The inconsistency may be due to false-positive studies, false-negative studies, population stratification, or true variability in genetic determinants among different populations (28). To address these issues, we have undertaken several strategies in the present study:

We recruited two large populations, a family-based population (635 pedigrees and 1,910 individuals) and a case-control population (973 cases and 956 controls), which had enough power to detect association to modest genetic effects; many previously reported studies on COPD candidate genes were hampered by relatively small population sizes (29).
We carefully considered the impact of population stratification. The family-based association analyses are immune to population stratification, and genomic control methods were used to assess for population stratification in our case-control association analysis.
The two study populations were phenotyped using similar methods. Although the ICGN family-based samples were ascertained through probands with relatively early onset COPD, the Norwegian case-control study includes subjects with COPD of milder degree as well as older cases. Thus, we anticipate that the results should be applicable to the general COPD population of European ancestry.

The lack of replication by Chappell and colleagues (11) of SERPINE2 associations could relate to several factors. They tested five SNPs from SERPINE2 in a large case-control population with 1,018 cases and 911 control subjects. Only limited information about their analysis is provided in their Letter to the Editor, but they did not find significant association results. Regarding the five SNPs that Chappell and colleagues studied, four of them—rs1438831, rs920251, rs6747096, and rs3795879—were also tested in the present analyses, but were not significantly associated with COPD-related phenotypes in the present study (Tables 3 and 4). Genetic heterogeneity between study populations and false-negative results (potentially related to population stratification in their samples) could have contributed to their lack of replication of SERPINE2 associations.

The present study also has several weaknesses:

The case-control population is from a single center in Norway, and the age and pack-years of smoking were different between the cases and control subjects. Therefore, we used a logistic regression model for the association analysis, correcting for age and pack-years of smoking. The lower average smoking intensity among control subjects could reduce power to detect significant associations. The fact that the case-control population is from a single center can be considered as an advantage because the population is more homogeneous and the possibility of false-positive results by population stratification is minimal. The analysis of population stratification using a set of 221 unlinked SNPs showed no evidence for population stratification.
Only SNP rs6734100 was still significantly associated with COPD after a conservative correction for multiple comparisons. However, the replication in two independent populations and validation by haplotype analysis suggest that the association results are valid. Because five SNPs showed significant associations in both populations, our results are unlikely to have occurred by random chance. Although various procedures have been studied for correction of multiple testing, including Bonferroni correction and permutation testing, there has not been an ideal statistical framework to deal with raw p values for SNP analyses (30), especially for replication of a previously reported association result.
We did not find associations with the same SNPs for the same phenotypes in both populations. However, five significant SNPs for COPD in the family data were replicated for the most commonly used measure of airflow obstruction (FEV₁/FVC) among cases in the case-control population.
It has been shown that chromosome 2q is linked to FEV₁/FVC, and SERPINE2 is only one of the possible candidates in this region. So, the possibility of genetic association with another gene or genes in this locus is not excluded in the present study.
The SERPINE2 SNPs associated with COPD in this study have no known function, and it is likely that these SNPs are in LD with the functional SNPs that determine disease susceptibility. SERPINE2 gene expression is regulated by adaptor related protein complex 1 (AP-1)–like elements, a cAMP-responsive element (CRE)–like element, and specificity protein 1 (Sp1)-binding sites in the proximal promoter region (31). It has been reported that there is a nuclear factor-κB–like motif located at −400 bp, which regulates the basal, but not TNF-α-induced, SERPINE2 promoter activity (32). It is possible that SNPs in the regulatory sequences of SERPINE2 are functional mutations. Only one putative functional SNP was genotyped in this study, rs3795875 (amino acid position 51 ILE > MET), but the minor allele frequency was only 0.1% and hence did not provide any significant association. More work has to be done to identify the functional mutations and assess the effect of specific variants on gene function.

In conclusion, we conducted a robust genetic association study and found that variants in the SERPINE2 gene are likely to contribute to the development of COPD. Functional tests need to be performed to find the molecular mechanism that drives the genetic association between COPD phenotypes and SERPINE2.

The authors gratefully acknowledge the assistance of the site coordinators for the family and case-control collections. Mark Hall, Sandra Hammond, Rachel Taylor, Sara Alalouf, and Santhi Subramanian of GlaxoSmithKline for data management support.

International COPD Genetics Network (ICGN) Investigators:Alvar Agusti, Son Dureta Hospital and Fundación Caubet-Cimera, Palma de Mallorca, Spain; Peter M. A. Calverley, University of Liverpool, Liverpool, UK; Claudio F. Donner, Division of Pulmonary Disease, S. Maugeri Foundation, Veruno (NO), Italy; Robert D. Levy, University of British Columbia, Vancouver, Canada; Barry J. Make, National Jewish Medical and Research Centre, Denver, Colorado; Peter D. Paré, University of British Columbia, Vancouver, Canada; Stephen I. Rennard, University of Nebraska, Omaha, Nebraska; Jørgen Vestbo, Department of Cardiology and Respiratory Medicine, Hvidovre Hospital, Copenhagen, Denmark; Emiel F. M. Wouters, University Hospital Maastricht, The Netherlands.

References

Section:

1.	Hanrahan JP, Sherman CB, Bresnitz EA, Emmons KM, Mannino DM. Cigarette smoking and health. Am J Respir Crit Care Med 1996;153:861–865. Abstract, Medline, Google Scholar
2.	Redline S, Tishler PV, Rosner B, Lewitter FI, Vandenburgh M, Weiss ST. Speizer FEl. Genotypic and phenotypic similarities in pulmonary function among family members of adult monozygotic and dizygotic twins. Am J Epidemiol 1989;129:827–836. Crossref, Medline, Google Scholar
3.	Webster PM, Lorimer EG. man SF, Woolf CR, Zamel N. Pulmonary function in identical twins: comparison of nonsmokers and smokers. Am Rev Respir Dis 1979;119:223–228. Abstract, Medline, Google Scholar
4.	Kauffmann F, Tager IB, Munoz A, Speizer FE. Familial factors related to lung function in children aged 6–10 years: results from the PAARC epidemiologic study. Am J Epidemiol 1989;129:1289–1299. Crossref, Medline, Google Scholar
5.	Lebowitz MD, Knudson RJ, Burrows B. Family aggregation of pulmonary function measurements. Am Rev Respir Dis 1984;129:8–11. Abstract, Medline, Google Scholar
6.	Silverman EK, Mosley JD, Palmer LJ, Barth M, Senter JM, Brown A, Drazen JM, Kwiatkowski DJ, Chapman HA, Campbell EJ, et al. Genome-wide linkage analysis of severe, early-onset chronic obstructive pulmonary disease: airflow obstruction and chronic bronchitis phenotypes. Hum Mol Genet 2002;11:623–632. Crossref, Medline, Google Scholar
7.	Silverman EK, Palmer LJ, Mosley JD, Barth M, Senter JM, Brown A, Drazen JM, Kwiatkowski DJ, Chapman HA, Campbell EJ, et al. Genomewide linkage analysis of quantitative spirometric phenotypes in severe early-onset chronic obstructive pulmonary disease. Am J Hum Genet 2002;70:1229–1239. Crossref, Medline, Google Scholar
8.	Palmer LJ, Celedón JC, Chapman HA, Speizer FE, Weiss ST, Silverman EK. Genome-wide linkage analysis of bronchodilator responsiveness and post-bronchodilator spirometric phenotypes in chronic obstructive pulmonary disease. Hum Mol Genet 2003;12:1199–1210. Crossref, Medline, Google Scholar
9.	Malhotra A, Peiffer AP, Ryujin DT, Elsner T, Kanner RE, Leppert MF, Hasstedt SJ. Further evidence for the role of genes on chromosome 2 and chromosome 5 in the inheritance of pulmonary function. Am J Respir Crit Care Med 2003;168:556–561. Abstract, Medline, Google Scholar
10.	DeMeo DL, Mariani TJ, Lange C, Srisuma S, Litonjua AA, Celedon JC, Lake SL, Reilly JJ, Chapman HA, Mecham BH, et al. The SERPINE2 gene is associated with chronic obstructive pulmonary disease. Am J Hum Genet 2006;78:253–264. Crossref, Medline, Google Scholar
11.	Chappell S, Daly L, Morgan K, Baranes TG, Roca J, Rabinovich R, Millar A, Donnelly SC, Keatings V, MacNee W, et al. The SERPINE2 gene and chronic obstructive pulmonary disease [letter]. Am J Hum Genet 2006;79:184–186. Crossref, Medline, Google Scholar
12.	Devlin B, Bacanu S, Roeder K. Genomic control to the extreme. Nat Genet 2004;36:1129–1130. Crossref, Medline, Google Scholar
13.	Devlin B, Roeder K. Genomic control for association studies. Biometrics 1999;55:997–1004. Crossref, Medline, Google Scholar
14.	O'Connell JR, Weeks DE. PedCheck: a program for identification of genotype incompatibilities in linkage analysis. Am J Hum Genet 1998;63:259–266. Crossref, Medline, Google Scholar
15.	Horvath S, Xu X, Laird NM. The family based association test method: strategies for studying general genotype-phenotype associations. Eur J Hum Genet 2001;9:301–306. Crossref, Medline, Google Scholar
16.	Lange C, DeMeo D, Silverman EK, Weiss ST, Laird NM. PBAT: tools for family-based association studies. Am J Hum Genet 2004;74:367–369. Crossref, Medline, Google Scholar
17.	Horvath S, Xu X, Lake SL, Silverman EK, Weiss ST, Laird NM. Family-based tests for associating haplotypes with general phenotype data: application to asthma genetics. Genet Epidemiol 2004;26:61–69. Crossref, Medline, Google Scholar
18.	Dudbridge F. Pedigree disequilibrium tests for multilocus haplotypes. Genet Epidemiol 2003;25:115–121. Crossref, Medline, Google Scholar
19.	Schaid DJ, Rowland CM, Tines DE, Jacobson RM, Poland GA. Score tests for association between traits and haplotypes when linkage phase is ambiguous. Am J Hum Genet 2002;70:425–434. Crossref, Medline, Google Scholar
20.	Barrett JC, Fry B, Maller J, Daly MJ. Haploview: analysis and visualization of LD and haplotype maps. Bioinformatics 2005;21:263–265. Crossref, Medline, Google Scholar
21.	Gabriel SB, Schaffner SF, Nguyen H, Moore JM, Roy J, Blumenstiel B, Higgins J, DeFelice M, Lochner A, Faggart M, et al. The structure of haplotype blocks in the human genome. Science 2002;296:2225–2229. Crossref, Medline, Google Scholar
22.	Carter RE, Cerosaletti KM, Burkin DJ, Fournier REK, Jones C, Greenberg BD, Citron BA, Festoff BW. The gene for the serpin thrombin inhibitor (P17), protease nexin 1, is located on human chromosome 2q33-q35 and on syntenic regions in the mouse and sheep genomes. Genomics 1995;27:196–199. Crossref, Medline, Google Scholar
23.	Mansuy IM, van der Putten H, Schmid P, Meins M, Botteri FM, Monard D. Variable and multiple expression of protease nexin-1 during mouse organogenesis and nervous system development. Development 1993;119:1119–1134. Crossref, Medline, Google Scholar
24.	Mbebi C, Hantaï D, Jandrot-Perrus M, Doyennette MA, Verdière-Sahuqué M. Protease nexin I expression is up-regulated in human skeletal muscle by injury-related factors. J Cell Physiol 1999;179:305–314. Crossref, Medline, Google Scholar
25.	Vaughan PJ, Cunningham DD. Regulation of protease nexin-1 synthesis and secretion in cultured brain cells by injury-related factors. J Biol Chem 1993;268:3720–3727. Crossref, Medline, Google Scholar
26.	Shapiro SD, Goldstein NM, Houghton AM, Kobayashi DK, Kelley D, Belaaouaj A. Neutrophil elastase contributes to cigarette smoke-induced emphysema in mice. Am J Pathol 2003;163:2329–2335. Crossref, Medline, Google Scholar
27.	Lohmueller KE, Pearce CL, Pike M, Lander ES, Hirschhorn JN. Meta-analysis of genetic association studies supports a contribution of common variants to susceptibility to common disease. Nat Genet 2003;33:177–182. Crossref, Medline, Google Scholar
28.	Silverman EK, Palmer LJ. Case-control association studies for the genetics of complex respiratory diseases. Am J Respir Cell Mol Biol 2000;22:645–648. Abstract, Medline, Google Scholar
29.	Joos L, Pare PD, Sandford AJ. Genetic risk factors for chronic obstructive pulmonary disease. Swiss Med Wkly 2002;132:27–37. Medline, Google Scholar
30.	Roeder K, Bacanu SA, Sonpar V, Zhang X, Devlin B. Analysis of single-locus tests to detect gene/disease associations. Genet Epidemiol 2005;28:207–219. Crossref, Medline, Google Scholar
31.	Cousin E, Medcalf RL, Bergonzelli GE, Kruithof EK. Regulatory elements involved in constitutive and phorbol ester-inducible expression of the plasminogen activator inhibitor type 2 gene promoter. Nucleic Acids Res 1991;19:3881–3886. Crossref, Medline, Google Scholar
32.	Mahony D, Kalionis B, Antalis TM. Plasminogen activator inhibitor type-2 (PAI-2) gene transcription requires a novel NF-κB-like transcriptional regulatory motif. Eur J Biochem 1999;263:765–772. Crossref, Medline, Google Scholar

Correspondence and requests for reprints should be addressed to Sreekumar G. Pillai, Ph.D., GlaxoSmithKline R&D, 5 Moore Drive, Research Triangle Park, NC 27709. E-mail: [email protected]

American Journal of Respiratory and Critical Care Medicine

Abstract

Abstract

Related