Abstract
The Hedgehog signalling pathway plays an important role in lung morphogenesis and cellular responses to lung injury. A genome-wide association study has demonstrated that two single nucleotide polymorphisms (SNPs) near the Hedgehog-interacting protein (Hip) gene, SNP identifiers rs1828591 and rs13118928, are associated with risk of chronic obstructive pulmonary disease (COPD). The aim of the present study was to validate the observed association between genetic variation near the Hip gene and COPD, and to investigate whether risk estimates were modified by smoking behaviour.
The association between the Hip gene SNPs and COPD was investigated in the Rotterdam Study by logistic regression analyses, adjusted for several covariates. In addition, an association meta-analysis was performed that included data from the genome-wide association study on COPD.
Both SNPs were significantly associated with risk of COPD (OR 0.80; 95% CI 0.72–0.91). Homozygosity for the minor G allele resulted in a decreased risk of COPD of ∼40% (95% CI 0.47–0.78). There was a significant interaction with the number of pack-years of smoking (p = 0.004). The meta-analysis yielded an odds ratio for COPD of 0.80 per additional G allele (p = 3.4×10−9).
Genetic variation near the Hip gene was significantly associated with risk of COPD, depending on the number of pack-years of smoking.
- Chronic obstructive pulmonary disease
- epidemiology of chronic obstructive pulmonary disease
- genetics
- Hedgehog-interacting protein
Chronic obstructive pulmonary disease (COPD) is a major worldwide and still increasing health problem 1. The disease is characterised by progressive airflow limitation, driven by an abnormal inflammatory response of the airways to inhaled particles and fumes 2, 3. Smoking is the major aetiological factor for COPD in the Western world 1. However, only 15–25% of cigarette smokers seem to develop the disease, suggesting the presence of susceptibility genes for COPD 4, 5. Previous studies have demonstrated familial aggregation of COPD, and some twin studies on pulmonary function have indicated a genetic contribution to clinically relevant parameters, such as forced expiratory volume in 1 s (FEV1) and forced vital capacity (FVC) 6–8.
The evolutionarily highly conserved Hedgehog (Hh) signalling pathway is crucial to several developmental processes, including lung organogenesis, and is also implicated in the response of the airway epithelium to smoking 9–11. In cells that respond to Hh, two proteins are upregulated, i.e. the Hh receptor Patched protein (Ptc) and the vertebrate-specific Hh- interacting protein (Hip). Both play an important role in the negative feedback regulation of Hh signalling. Loss of Hip function in mouse has already been associated with specific lung defects and results in neonatal lethality 12. As in mice, dysfunction of the Hh pathway during human fetal life is responsible for severe lung malformations 13–15. In adult life, the Hh pathway plays a central role in the repair and regeneration of several tissues, and aberrant activation of this pathway by epithelial stem cells has been linked to various malignancies, including lung cancer 16, 17.
Two recent genome-wide association (GWA) studies unexpectedly demonstrated that two single nucleotide polymorphisms (SNPs) near the Hip locus (SNP identifiers rs13118928 and rs1828591) were associated with COPD and lung function, respectively, although genome-wide significance levels were not reached for the COPD phenotype 18, 19.
The objective of the current study was to study the above-mentioned association between SNPs near the Hip gene and COPD, and to investigate whether smoking or sex modified this association, in a large prospective population-based cohort study with ∼20 yrs of follow-up.
METHODS
Study population
The present study forms part of the Rotterdam Study, a population-based cohort study aimed at assessing the occurrence of and risk factors for chronic diseases in the elderly. The objectives and methods of the Rotterdam Study have been described elsewhere 20, 21. In short, the Rotterdam Study cohort included 7,983 participants aged ≥55 yrs living in Ommoord, a well-defined suburb of the city of Rotterdam (the Netherlands). Almost all (99.8%) of the participants are of Caucasian descent. Baseline data were collected during 1989–1993. Participants were visited at home at the start of the study for a standardised interview regarding their health status. Since the start of the Rotterdam Study, each participant has visited the research centre every 2–3 yrs, during which several measurements were performed. In addition, participants were continuously monitored for the onset of major events, which occurred during follow-up through automated linkage with files from general practitioners. Trained research assistants collected information from the medical records of the general practitioners, hospitals and nursing homes. The medical ethics committee of the Erasmus Medical Center (Rotterdam) approved the study. Participants gave written informed consent and permission to retrieve information from treating physicians.
Spirometry
Spirometry was performed by trained paramedical personnel using a SpiroPro® portable spirometer (Erich Jaeger, Hoechberg, Germany) according to American Thoracic Society (ATS)/European Respiratory Society (ERS) guidelines 22. FEV1, FVC and FEV1/FVC ratio were measured; the spirogram (volume–time curve) and maximal expiratory flow–volume curve were also recorded. Spirometry procedures that yielded results that did not meet ATS/ERS criteria for acceptability and reproducibility were classified as not interpretable 23.
Patient identification and validation
For the validation of the COPD cases, access was available to hospital discharge letters, files from general practitioners, spirometry reports and pharmacy dispensing data for patients participating in the Rotterdam Study, as previously described 23. Spirometry was performed in 3,550 participants in the context of the Rotterdam cohort study. In addition, throughout the entire study period, spirometry was also performed on clinical indication by respiratory specialists and specialists in internal medicine with a subspeciality in respiratory medicine. In the absence of spirometry, all medical information on subjects who used respiratory medication for ≥6 months (anatomical therapeutic chemical classification system codes R03AC, R03AK, R03BA, R03BB, R03CC and R03DA) 24 and all hospital discharge letters or mortality reports with a coded diagnosis of COPD (J40–J44 (International Statistical Classification of Diseases and Related Health Problems, 10th revision)) 25 were reviewed. Drug use was only used for potential case finding and not for making the diagnosis of COPD. In the case of confirmed COPD, the day of first drug use was used as an indicator of the date of onset of COPD.
The diagnosis of COPD was classified as definite or probable. Definite COPD was defined by moderate-to-severe obstructive spirometry results (FEV1/FVC of <0.7 and FEV1 of <80% of the predicted value) and/or as COPD diagnosed by a specialist in internal medicine (mainly respiratory physicians or those with a subspecialty in respiratory medicine) based upon the combined results of clinical history, physical examination and spirometry. Probable COPD was defined by mild obstructive spirometry results (FEV1/FVC of <0.7 and FEV1 of ≥80% pred) and/or as COPD diagnosed by a physician in another medical speciality (e.g. a general practitioner).
The index date was defined as the date of diagnosis of COPD found in the medical reports, the date of a first prescription of COPD medication or the date of the obstructive lung function examination, whichever came first.
Genotyping
The Illumina Infinium II HumanHap550K BeadChip (version 3; Illumina, San Diego, CA, USA) was used for genotyping 6,449 participants with available DNA in the Rotterdam Study, as previously described 26. Genotyping was performed at the Dept of Internal Medicine of Erasmus Medical Center using BeadStudio software (version 0.3.10.14; Illumina) for genotyping calling. Participants with a call rate of <97.5%, excess autosomal heterozygosity, sex mismatch or outlying identity-by-state clustering estimates were excluded. A total of 5,974 participants were successfully genotyped and passed quality control. Genotype information on two specific SNPs near the Hip gene located on chromosome 4 (rs13118928 and rs1828591) was extracted from these data.
Statistical analyses
The demographic and clinical characteristics of the study population are expressed as median with interquartile range (IQR) in the text.
First, the association of rs13118928 and rs1828591 with COPD was studied within the Rotterdam Study. All COPD cases, prevalent and incident, were used for analyses in order to obtain maximal power. Consequently, logistic regression analyses were performed because Cox’s proportional hazards analysis was not possible with prevalent cases at baseline. Both SNPs were tested for Hardy–Weinberg equilibrium using Chi-squared statistics. A power calculation using Quanto (University of Southern California, Los Angeles, CA, USA), based on the combined odds ratio (OR) of 0.7 of the Bergen case–control cohort and the National Emphysema Treatment Trial (NETT)/ Normative Aging Study (NAS) population (combined β = -0.332) from the study of Pillai et al. 18, showed that it was possible to demonstrate an OR for COPD of 0.7 with >99% power at a two-sided α of 0.05 27. Additive and genotype-based logistic regression models, adjusted for age at baseline, sex, smoking behaviour (ever- and never-smoker) and the total number of pack-years of smoking, were used to calculate ORs and 95% confidence intervals (CIs). In the additive genetic model, the categorical variable of the number of variant alleles (homozygous wild type (0 variant alleles); heterozygous (1 variant allele); homozygous variant allele (2 variant alleles)) was included as a continuous value in the logistic regression model in order to calculate trends.
The following covariates were considered as potential confounders or effect modifiers: age, sex, smoking behaviour, and the total number of pack-years of smoking. Smoking status was self-reported and categorised as ever- or never-smoker. The category of ever-smoker included both current and former smokers. The number of pack-years was computed from the duration of smoking (in years) multiplied by the number of cigarettes smoked daily divided by 20. The number of pack-years of smoking of each study participant was calculated using prospectively collected data regarding the smoking behaviour of the cohort members, collected during the first, second and third follow-up visits of the Rotterdam Study.
Participants with asthma or a spirometry result suggestive of restrictive lung disease were excluded from the analyses. Interaction terms and subanalyses were introduced to explore potential effect modification by sex and tertiles of pack-years of smoking. A p-value below the conventional level of significance (p<0.05) was considered significant.
Secondly, an inverse-variance-weighted meta-analysis was performed. This analysis combined the results obtained to the recently published data of Pillai et al. 18. The regression coefficients and sem were calculated on the basis of the reported ORs and p-values from the Bergen case–control cohort and NETT/NAS data from the recent GWA study of Pillai et al. 18.
The results of the association meta-analysis were calculated using Excel version 2007 (Microsoft Office). All other statistical analyses were performed using SPSS for Windows version 15 (SPSS, Inc., Chicago, IL, USA).
RESULTS
Baseline characteristics of the study population
Within the Rotterdam Study source population of 7,983 subjects, there were 928 patients with validated prevalent or incident COPD 23. Both SNPs, rs13118928 and rs1828591, were successfully genotyped in 5,974 out of the 7,983 participants for whom DNA was available, including 742 COPD cases. A total of 256 study participants with asthma or a spirometry result suggestive of restrictive pulmonary disease were excluded (fig. 1). Overall, the median age of the study population at the start of the study was 68.4 yrs (IQR 14 yrs), and 59% were female. Additional clinical characteristics are shown in table 1.
Association between variation near the Hip gene and COPD in the Rotterdam Study
Both SNPs studied, rs13118928 and rs1828591, were in Hardy–Weinberg equilibrium (rs13118928: χ2 = 1.413; p = 0.235; rs1828591: χ2 = 1.215; p = 0.270) and in perfect linkage disequilibrium (r2 = 1.00 in HapMap CEU reference samples (Utah residents of Northern and Western European descent)). Owing to the high linkage between the two SNPs, analyses yielded the same results and thus, for reasons of simplicity, only the results for the rs13118928 SNP are shown in the text and tables.
Under an additive genetic model, there was a significant association between both SNPs and the risk of COPD with an OR of 0.80 per additional guanine (G) allele (95% CI 0.72–0.91; p<0.001). Genotypic analysis showed that homozygosity for the minor G allele was associated with a significantly decreased risk of COPD of ∼40% (OR 0.60; 95% CI 0.47–0.78; p<0.001) (table 2). The decrease in risk remained significant when the analyses were restricted to those participants with an available spirometric examination during follow-up in the Rotterdam Study (OR 0.60; 95% CI 0.43–0.89; p = 0.011) (data not shown).
There was no interaction with sex (p = 0.872), but a significant interaction effect was present with the total number of pack-years of smoking (p = 0.004). Additional subanalyses demonstrated that the protective effect of having one or more variant alleles was most pronounced for smoking participants in the highest tertile of pack-years of smoking (>36 pack-yrs), with an OR of 0.80 (95% CI 0.62–0.93; p = 0.007) versus an OR of 0.90 (95% CI 0.67–1.09; p = 0.209) in those who had never smoked (table 3).
Association between variation near the Hip gene and COPD: meta-analysis with published results
The original p-value, including the Bergen case–control cohort, the International COPD Genetics Network (ICGN) population and the NETT/NAS data, was 1.7×10−7. Since it was not possible to obtain β estimates and sem for the ICGN population, the p-value for the combination of the Bergen and NETT/NAS data was also calculated. This resulted in a p-value of 1.0×10−6. The combination of the Rotterdam Study results with the Bergen and NETT/NAS results yielded an overall OR of 0.80 per additional G allele (95% CI 0.70–0.84), with a p-value of 3.4×10−9, which is significant at a genome-wide level (table 4; fig. 2).
DISCUSSION
The results of the present population-based prospective cohort study confirm the recently observed association between genetic variation around the Hip gene and COPD, as described by Pillai et al. 18. It was found that the minor G allele of the rs13118928 and rs1828591 SNPs is associated with a significantly decreased risk of COPD. The risk seemed to be dependent upon the smoking behaviour of the participants, since the decrease in risk was strongest in subjects who had smoked for >30 pack-yrs at the start of the Rotterdam Study.
The Hh signalling pathway is an evolutionarily highly conserved signalling cascade that is crucial to several developmental processes 10, 11. In mammals, there are three Hh proteins: Sonic, Indian, and Desert. Sonic Hh specifically targets the lung and is expressed in the lung epithelium during embryogenesis 28. Signalling is initiated by binding of one of the Hh proteins to the 12-pass transmembrane protein receptor Ptc, which is required to inhibit the activity of Smoothened (Smo), a seven-pass transmembrane protein, which is responsible for regulation of the zinc-finger-containing glioma-associated oncogene homologue (Gli) transcriptional effectors, Gli1, Gli2 and Gli3 13, 29. Upon binding of Hh to Ptc, the Ptc-mediated inhibition of Smo is relieved 30, 31. This triggers a cascade that finally results in activation of the Gli family and transcription of Hh target genes, including Ptc and Gli itself 32. The availability of the Hh ligands is regulated by expression of Hip. The Hip protein was initially discovered by screening murine cDNA expression for proteins that bound to Sonic Hh. The Hip gene, located on chromosome 4 (4q31.21–4q31.3), encodes a transmembrane glycoprotein that binds all three Hh proteins with an affinity equal to that of Ptc and is upregulated upon Hh signalling 28. Hip is expressed in all Hh target tissues and acts by internalising and degrading the Hh protein. Similar to Ptc, Hip thus serves as a natural antagonist of the Hh pathway.
In this way, alterations in the Hip gene can result in altered gene expression of a functional protein, or in a nonfunctional protein and upregulation of the Hh signalling pathway.
Loss of Hip function in mice during fetal life has been associated with specific lung defects resulting in respiratory failure and neonatal lethality 12. In adult animals, downregulation of Hip has been reported in murine lung cancer cell lines 33. As in rodents, dysfunction of the Hh pathway during fetal life in humans is responsible for severe lung malformations 13–15. In adult life, the Hh pathway plays a central role in the repair and regeneration of several human tissues, and aberrant activation of this pathway by epithelial stem cells has been linked to various malignancies, including lung cancer 16, 17. In addition, Hip seemed to be downregulated in human nonsmall cell lung cancer tissues and endothelial cells during angiogenesis 33.
The results of the present study and the two previous GWA reports demonstrate that variants near the Hip gene influence the risk of COPD. Although it is not clear which regulatory effect explains this association, several hypotheses are possible. First, Madison et al. 34 demonstrated that ectopic expression of Hip in mice during fetal life was associated with expanded smooth muscle in the intestinal epithelium 31. If the same effect were present in the small airways, then it could be speculated that changes in Hip influence the thickness of the airway muscle layer and induce more or less airflow limitation. Secondly, a study of Pogach et al. 35 demonstrated that several developmental pathways in the lungs of adult mice, including Hh, reactivate for tissue repair following hyperoxia. In addition, others have shown that embryogenic signalling pathways, such as Hh, are activated in human bronchial epithelial cells exposed to cigarette smoke 36.
The prevalence, morbidity and mortality of COPD in females are increasing 37. In addition, some reports suggest that females may be at greater risk of developing severe early-onset COPD due to sex differences in the metabolism of cigarette smoke 38. However, the results of these studies remain controversial since differences in lung development, smoking behaviour, diagnosis and treatment of COPD are possible confounders. Since sex can be of interest in the pathogenesis of COPD, a potential interaction effect of sex on the observed association between variation near the Hip gene and COPD was tested for. However, no significant influence was found (p = 0.872).
The results of these two studies raise the possibility that the Hh pathway is involved in the response of the airway epithelium to tissue injury, and that alterations in Hh signalling can lead to susceptibility to COPD, depending upon the severity of the smoking behaviour. Thirdly, since a recent GWA study investigating genetic variants that influence adult height identified several relevant genes in the Hh signalling pathway (Indian Hh, Hip and Ptc), height could be considered a potential confounder for the observed association between Hip and COPD 39. However, additional analyses including height as a covariate did not influence the present estimates (data not shown).
The strengths of the present study are the high-quality information available regarding exposures prior to outcome, with prospective data collection, the general population-based setting, the large number of subjects that participated in the Rotterdam Study and the long duration of follow-up. The high response rate and virtually complete follow-up for each participant makes information and selection bias unlikely for these data. A power analysis showed that the present study was of sufficient size to distinguish between the proposed effect and no effect. The study was conducted in a data set independent from the original GWA study.
A similar significant protective risk was demonstrated with the same SNPs, and the combined analysis led to a smaller p-value than in the original report, below the stringent threshold frequently used to define genome-wide significance. Hence the present study satisfies all criteria suggested by the National Cancer Institute/National Human Genome Research Institute Working Group on Replication in Association Studies for establishing positive replication 40.
As for all other association studies, no conclusions regarding causality or the mechanism through which the associated SNPs influence COPD risk can be formed. It is possible that the two SNPs studied are not the causal variants, but are in linkage disequilibrium with other polymorphisms that are responsible for the increased COPD risk. Since the SNPs are not situated within the Hip gene and do not seem to exhibit a strong linkage with other SNPs in the Hip gene itself, a regulatory function is possible. Additional functional and translational research examining the role of Hip in COPD is certainly warranted.
In conclusion, the present study confirms the previously reported association between two SNPs near the Hip gene and COPD in a Caucasian population-based cohort study. In addition, a potential gene-by-environment interaction was identified, since the association was most apparent in heavily smoking participants.
Acknowledgments
The authors thank J. Verkroost (Erasmus University Medical Center, Rotterdam, the Netherlands) for outstanding work in preparing the data set for the validation of the COPD cases, and G. Van Pottelberge (Ghent University, Ghent, Belgium) for helping to validate the COPD cases. We also thank J. Heeringa (Erasmus University Medical Center), the research assistants and all our other colleagues in the Ommoord research centre (Erasmus University Medical Center) for their efforts in the data collection.
Footnotes
↵This article has supplementary material accessible from www.erj.ersjournals.com
Support Statement
The Rotterdam Study is supported by the Erasmus University Medical Center (Rotterdam), Erasmus University (Rotterdam), Netherlands Organization for Scientific Research (NWO; The Hague), Netherlands Organization for Health Research and Development (The Hague), Research Institute for Diseases in the Elderly (The Hague), Ministry of Education, Culture and Science (The Hague) and Ministry of Health, Welfare and Sport (The Hague) (all the Netherlands). This study was supported by NWO grants 904-61-093 and 918-46-615, and Netherlands Genomics Initiative/NWO project number 050-060-810. Y.M.T.A van Durme received a travel grant from the Belgian Thoracic Society and is a doctoral research fellow of the Fund for Scientific Research – Flanders (Brussels, Belgium). M. Eijgelsheim is funded by Netherlands Heart Foundation (The Hague, the Netherlands) project number 2007B221.
Statement of Interest
A statement of interest for G.G. Brusselle can be found at www.erj.ersjournals.com/misc/statements.dtl
- Received August 14, 2009.
- Accepted November 17, 2009.
- ©ERS 2010