Abstract
Studies using genetic isolates with limited genetic variation may be useful in chronic obstructive pulmonary disease (COPD) genetics, but are thus far lacking. The associations between single nucleotide polymorphisms (SNPs) in candidate genes and lung function in COPD were studied in a genetic isolate.
In 91 subjects with Global Initiative for Chronic Obstructive Lung Disease (GOLD) stage ≥1 COPD, who were members of an extended pedigree including 6,175 people from the Genetic Research in Isolated Populations study, 32 SNPs were analysed in 13 candidate genes: a disintegrin and metalloprotease domain 33 gene (ADAM33), transforming growth factor-β1 gene (TGFB1), matrix metalloprotease-1 gene (MMP1), MMP2, MMP9, MMP12, tissue inhibitor of metalloprotease-1 gene (TIMP1), surfactant protein A1 gene (SFTPA1), SFTPA2, SFTPB, SFTPD, glutathione S-transferase P1 gene (GSTP1), and haem oxygenase 1 gene (HMOX1). Their relation to forced expiratory volume in 1 s (FEV1), inspiratory vital capacity (IVC) and FEV1/IVC were studied using restricted maximum likelihood linear mixed modelling, accounting for pedigree structure. Significant associations were replicated in the general Vlagtwedde/Vlaardingen study.
Six SNPs in TGFB1, SFTPA1, SFTPA2 and SFTPD were significantly associated with FEV1/IVC in subjects with GOLD stage ≥1 COPD. Two SNPs in TGFB1 (C to T substitution at nucleotide -509 and substitution of leucine 10 with proline (Leu10Pro)), Leu50Val in SFTPA1 and Ala160Thr in SFTPD showed evidence suggestive of association with FEV1/IVC in subjects with GOLD stage ≥2 COPD. The TGFB1 associations were replicated in GOLD stage ≥2 patients from the Vlagtwedde/Vlaardingen population, with similar effect sizes.
It was shown that a genetic isolate can be used to determine the genetics of lung function, which can be replicated in COPD patients from an independent population.
- Chronic obstructive pulmonary disease
- genetically isolated population
- lung function
- single nucleotide polymorphism
Chronic obstructive pulmonary disease (COPD) is the third leading cause of death worldwide, and is expected to increase in prevalence until 2030 1, 2. The disease has a large personal, societal, and economic impact. COPD is characterised by chronic airway inflammation, airway remodelling and airflow limitation that is not fully reversible. Since not all smokers develop COPD, genetic susceptibility must play a role in the development of this disease, in addition to environmental factors. The genetic determinants for COPD are difficult to study, since COPD is a disease that becomes clinically manifest only at later ages, when parents of COPD patients have already died and their children are probably too young to manifest airway obstruction. This limits the option of performing family-based genetic research. Moreover, published studies frequently use various definitions of disease status, which makes it difficult to compare their results. Therefore, it makes sense to choose a robust phenotype for definition of COPD, such as the level of lung function, which can be more easily compared between studies. Moreover, a low level of lung function is a predictor of mortality due to COPD 3–5.
Another complicating factor in studies on the genetics of COPD is that COPD is considered a complex genetic trait, i.e. multiple, possibly interacting, genetic and environmental factors are involved. Therefore, there are advantages to attempting to identify risk genes in populations that are relatively genetically and environmentally homogeneous, such as genetically isolated populations, in which genetic variation is reduced owing to the small number of founders and drift 6. However, these processes raise the question of whether findings can be extrapolated to the general population. Previous simulation studies suggest that this is the case for common variants with a frequency of >1% 6, but no empirical evidence is available.
A candidate gene study was conducted for level of airflow limitation in patients with COPD who were ascertained as part of the Genetic Research in Isolated Populations (GRIP) study that is being conducted in a young genetically isolated population from the south-western part of the Netherlands. All patients were genotyped using 32 single nucleotide polymorphisms (SNPs) in 13 candidate genes for COPD, chosen based on their previously published association with either COPD, level of lung function or lung function decline, as reported in the general population. Extensive genealogical information was collected, resulting in an extremely large and complex pedigree of 6,175 members. Finally, 1,390 Caucasians from the general Dutch population were studied, including 351 patients with COPD, in order to establish whether or not the present findings could be replicated in the general population. In both studies, it was investigated whether the severity of the disease, as reflected by lung function reduction, is genetically influenced in established COPD.
METHODS
Study populations
The present study forms part of the GRIP programme 7, 8. The GRIP programme is based in a recently genetically isolated population from the south-western part of the Netherlands, which was founded in the middle of the eighteenth century by ∼150 individuals and was genetically isolated until the middle of the twentieth century. The population now includes ∼20,000 inhabitants in eight adjacent communities. GRIP programme participants are generally related via multiple lines of descent and are inbred via multiple consanguineous loops 9, 10.
Subjects with general-practitioner-diagnosed COPD were invited to the research centre to undergo spirometry and complete a questionnaire 11. Spirometry was performed by trained pulmonary research technicians using a pneumotachograph (Viasys, Houten, the Netherlands; formerly Jaeger spirometry system). Predicted values for forced expiratory volume in 1 s (FEV1) were calculated using adjusted Quanjer et al. 12 equations for Caucasian subjects. DNA was isolated from blood using Puregene® DNA Purification Kits (Gentra, Inc., Minneapolis, MN, USA). All participants gave written informed consent.
In order to verify the findings from the GRIP study in the general population, cross-sectional data from the general-population-based Vlagtwedde/Vlaardingen cohort were used. Questionnaires, spirometric results and DNA were collected 13, 14. For this study, 351 subjects were selected, according to Global Initiative for Chronic Obstructive Lung Disease (GOLD) criteria, with GOLD stage ≥1 COPD at the last 1989/1990 survey, of whom 167 had GOLD stage ≥2 COPD 15.
Genotyping
SNPs in candidate genes for lung function and COPD, based on their previously published significant associations, were genotyped (table 1⇓). The selected SNPs were either the most significant SNPs in previous studies, tagging SNPs for the gene, or SNPs with a known functional effect on gene expression or function. Genotyping was performed using Applied Biosystems TaqMan® SNP Genotyping Assays (Applied Biosystems, Nieuwerkerk aan de IJssel, the Netherlands). Sequences of primers and probes are available on request.
Candidate genes and single nucleotide polymorphisms(SNPs) genotyped in the study population
Statistical analysis
In order to analyse pedigree data, use was made of the measured genotype (MG) approach 33, which models quantitative traits as
where yi is the phenotype of the ith individual, g the vector of genotypes at the marker under study, k the marker genotype effect, cij the value of the jth covariate or fixed effect for the individual i, βj an estimate of the jth fixed effect or covariate and Gi and ei random additive polygenic and residual effects, respectively. The random effects are assumed to follow multivariate normal distribution with a mean of zero. The variance for the polygenic effects is defined as ΦσG2, where Φ is the relationship matrix and σG2 the additive genetic variance due to polygenes. For the residual random effects, the variance is defined as Iσe2, where I is the identity matrix and σe2 the residual variance.
Since the pedigree under analysis was very large, fast genome-wide rapid association using mixed model and regression (GRAMMAR) approximation to the full MG approach was used 34. The GRAMMAR consists of a fast though conservative test at the screening stage, followed up with full MG analysis of polymorphisms that pass the relaxed screening significance threshold (p<0.1). All analyses involving pedigree were performed using ASReml v2.0 35, a package for linear mixed model analysis using restricted maximum likelihood. This is a joint venture between the biometrics programme of the New South Wales Department of Primary Industries (Orange, Australia) and the Biomathematics University of Rothamsted Research (Harpenden, UK). Statisticians in the UK and Australia have collaborated in its development.
Significant associations were tested using linear regression analyses in the Vlagtwedde/Vlaardingen population. All analyses were adjusted for age, height and sex.
RESULTS
GRIP study population
A total of 157 individuals who were diagnosed with COPD by their general practitioners were ascertained. Spirometric measures confirmed COPD in 91 subjects, i.e. subjects with GOLD stage ≥1 COPD (defined by an FEV1/inspiratory vital capacity (IVC) of <70%) 15. The rest of the subjects could not be defined as having COPD according to their spirometric results and were, therefore, excluded from the analyses. The familial relationship of these 91 subjects was determined in the larger GRIP study database. This resulted in a large extended pedigree structure of 6,175 members. The characteristics of the GRIP COPD population and the Vlagtwedde/Vlaardingen replication cohort are shown in table 2⇓.
Characteristics of the Genetic Research in Isolated Populations(GRIP) and Vlagtwedde/Vlaardingen (Vla/Vla) study populations
Association of genes with lung function parameters in GRIP, and replication in Vlagtwedde/Vlaardingen
The effects of SNPs in the studied genes on percentage predicted FEV1, IVC and FEV1/IVC were first analysed in the 91 subjects with GOLD stage ≥1 COPD. None of the SNPs were associated with percentage predicted FEV1 or IVC. Six SNPs in the transforming growth factor-β1 gene (TGFB1), surfactant protein A1 gene (SFTPA1), SFTPA2 and SFTPD were significantly associated with FEV1/IVC (table 3⇓). None of these associations were replicated in subjects from the Vlagtwedde/Vlaardingen cohort with GOLD stage ≥1 COPD (data not shown).
Associations of single nucleotide polymorphisms(SNPs) with forced expiratory volume in 1 s (FEV1)/inspiratory vital capacity (IVC) in the Genetic Research in Isolated Populations (GRIP) and Vlagtwedde/Vlaardingen (Vla/Vla) study populations
In addition, the effects of SNPs in the studied genes were analysed using a more stringent definition of COPD, namely GOLD stage ≥ = 2 (defined as FEV1/IVC of <70% and FEV1 of <80% pred). This resulted in 67 cases in the GRIP population. In these subjects, two SNPs in TGFB1 (cytosine to thymidine substitution at nucleotide -509 (-509C>T) and substitution of leucine 10 with proline (Leu10Pro)), Leu50Val in SFTPA1 and Ala160Thr in SFTPD showed evidence suggestive of association with FEV1/IVC (p<0.10) (table 3⇑). The TGFB1 -509C>T and Leu10Pro associations were replicated in GOLD stage ≥2 subjects from the Vlagtwedde/Vlaardingen population (n = 167), with similar effect sizes (see table 3⇑).
DISCUSSION
The present study is the first to use a genetically isolated population to analyse genetic effects on level of lung function in COPD. Interestingly, significant effects of SNPs in COPD candidate genes were found on severity of COPD, assessed by lung function in subjects with COPD, even though the present study population was small. The present results show that levels of FEV1/IVC, measures of airway obstruction, are genetically influenced in established COPD. This means that, even within patients with phenotypic COPD, genotypes can be identified that are associated with severity of disease. This is of clinical importance since low lung function has been shown to predict mortality in COPD, not only in the general population but also within COPD patients 3–5.
The TGFB1 SNPs that were associated with FEV1/IVC in the present populations have previously been associated with development of COPD or with lower FEV1 and FEV1/VC in several 17–19, but not all previous studies 14, 36, 37. The present results (in both the genetically isolated and general population) thus confirm the former studies that implicate a role of TGFB1 in the severity of airflow limitation. The SFTPA1 and SFTPD SNPs have been associated with COPD previously 20, 38. It is now shown for the first time that these SNPs may also play a role in severity of COPD. This is plausible since surfactant proteins decrease surface tension at the air–liquid interface and, therefore, reduce the tendency of alveoli to collapse during expiration. The latter contributes to the severity of airway obstruction, as measured by FEV1/IVC.
No significant associations of a disintegrin and metalloprotease domain 33 gene (ADAM33), matrix metalloprotease-1 gene (MMP1), MMP2, MMP9, MMP12, tissue inhibitor of metalloprotease-1 gene (TIMP1), SFTPB, glutathione S-transferase P1 gene (GSTP1) and haem oxygenase 1 gene (HMOX1) with level of lung function were found in COPD patients. This does not, however, imply that these genes do not play any role whatsoever in COPD. To date, no studies have analysed genetic effects on the severity of airway obstruction within patients with established COPD. The present study shows that SNPs in TGFB1, SFTPA1 and SFTPD may be important in progression of COPD, whereas the SNPs in the other genes, i.e. ADAM33, MMP1, MMP2, MMP9, MMP12, TIMP1, GSTP1 and HMOX1, may simply constitute SNPs that are important in the development of COPD.
One important advantage of testing genes in a genetically isolated population is that it provides an opportunity of finding genes associated with disease in a relatively small sample size due to increased homogeneity of the population, as recently demonstrated for multiple sclerosis 39. Thus, for a lower cost and effort, many genes can be tested regarding their significance in contributing to disease severity, which can subsequently be replicated in a larger sample of the general population. The most important requirement for such studies is that the genetic isolate is representative of the general population or disease-specific study populations. This is indeed the case since it was shown that, in selected subjects with COPD from the general population, the associations found in the young genetic isolate can be replicated in a substantial part. Thus it is possible to translate findings in a genetic isolate to the general population, but correct and comparable phenotyping of the study populations remains crucial to replicate associations between populations.
It was not possible to replicate the results of any of the SNPs in subjects with GOLD stage ≥1 COPD from the Vlagtwedde/Vlaardingen population. On closer investigation, it appeared that the GRIP patients with GOLD stage ≥1 COPD had more severe COPD, i.e. lower lung function and more symptoms, than COPD patients of similar disease stage in the Vlagtwedde/Vlaardingen population. A more strict definition of COPD (GOLD stage ≥ = 2) in the Vlagtwedde/Vlaardingen and GRIP populations gave a phenotypically better comparison. Indeed, when analysing subjects with GOLD stage ≥2 COPD from the Vlagtwedde/Vlaardingen population, the TGFB1 SNPs -509C>T and Leu10Pro were significantly associated with FEV1/IVC, as they were in the GRIP GOLD stage ≥2 COPD patients.
Since the percentage of subjects with, amongst others chronic cough, was different in both cohorts, the analyses were repeated using straightforward linear regression models with chronic cough in the model to check for stability of the effect estimates. Analyses on FEV1/IVC in the GRIP GOLD stage ≥2 population, taking, for example, chronic cough into account, resulted in similar regression estimates for the SNPs in TGFB1 and SFTPA1, but with smaller p-values and slightly higher explained variances, whereas the suggestive associations of the other SNPs disappeared. Additional adjustment for chronic cough in the Vlagtwedde/Vlaardingen GOLD stage ≥2 population resulted in similar significant regression estimates for the SNPs in TGFB1 with FEV1/IVC. Therefore, the effect estimates appear to be stable within both GOLD stage ≥2 groups, irrespective of differences in characteristics between the GRIP and Vlagtwedde/Vlaardingen GOLD stage ≥2 populations.
Several explanations may exist for the lack of replication for SFTPA1 and SFTPD (Met11Thr) SNP FEV1/IVC results in the Vlagtwedde/Vlaardingen GOLD stage ≥2 population. First, the original GRIP findings on these genes could be falsely positive. Indeed, multiple (though correlated) outcomes and SNPs were studied in GRIP. Another, more biological, explanation for the lack of replication may be that the prevalence of certain alleles in genetically isolated populations differs from that in a general population as a result of genetic drift and founder effects. Indeed, the genotype frequencies for the SFTPA1 Leu50Val SNP were significantly different between the two populations, but not for the other SNPs (table 4⇓). A third explanation may be that differences in characteristics exist between the study populations. The GRIP population had more severe COPD and was slightly older than the Vlagtwedde/Vlaardingen COPD population.
Genotype frequencies of significant single nucleotide polymorphisms in the Genetic Research in Isolated Populations(GRIP) compared to the Vlagtwedde/Vlaardingen (Vla/Vla) Global Initiative for Chronic Obstructive Lung Disease ≥2 population
In addition, differences in environment may affect the lack of replication of the surfactant protein gene data. The genetically isolated population shares the same environment, similar socioeconomic status and the same general practitioners. The possibility cannot be ruled out that the COPD patients in the GRIP population exhibited a higher prevalence of chronic bronchitis and airway disease, whereas the airway obstruction in the Vlagtwedde/Vlaardingen population may have been caused by emphysema 40–42. Further research is needed in order to separately assess these phenomena, since computed tomographic scans are necessary, which were not available for any of the present patients.
In conclusion, the present study provides two important messages. First, significant effects of SNPs were found on the severity of COPD, i.e. level of lung function in patients with established COPD, in a relatively small genetically isolated population with a large pedigree structure. Secondly, two of these associations were replicated in COPD patients selected from the general population on the condition that they were phenotypically similar. These findings are important since more severe airway obstruction is associated with progression and mortality of COPD. Future studies using this genetic isolate should focus on progression of COPD, since this population seems to be highly suitable for determining genetic risk factors for severity of airway obstruction in established COPD that can be translated to the general population.
Support statement
The Dutch Asthma Foundation (Leusden, the Netherlands) funded the collection of lung function data (NAF 3.4.04.041). C.C. van Diemen is assigned through the Dutch Asthma Foundation (NAF3.2.02.51). The Genetic Research in Isolated Populations programme is supported by grants from the Netherlands Organisation for Scientific Research (The Hague, the Netherlands; Pioneer grant to C.M. van Duijn) and the Center for Medical Systems Biology (Leiden, the Netherlands).
Statement of interest
None declared.
Acknowledgments
We thank L. Testers (Erasmus Medical Center Rotterdam, Rotterdam, the Netherlands) for help in fieldwork logistics and DNA collection, and M. Farenhorst, T. van Hoogdalem, J. Post, A. Verbokkem and K. Vink-Klooster (all University Medical Center Groningen, University of Groningen, Groningen, the Netherlands) for collecting the lung function data. We would like to thank P. Veraart, H. Kornman and E. Boeren (all Erasmus Medical Center Rotterdam) for their contribution to the genealogical research. We would like to thank all of the participants in the Genetic Research in Isolated Populations (GRIP) study for their cooperation, as well as the general practitioners that made this work possible. All of the research assistants of the GRIP study are acknowledged for their help in data collection.
- Received April 9, 2008.
- Accepted August 28, 2009.
- © ERS Journals Ltd