17q21 variants modify the association between early respiratory infections and asthma
- L.A.M. Smit*#¶,
- E. Bouzigon+§,f,
- I. Pin**##¶¶,
- V. Siroux**##,
- F. Monier*#,
- H. Aschard+§,f,
- J. Bousquet*#++,
- F. Gormand§§,
- J. Justff,
- N. Le Moual*#,
- R. Nadif*#,
- P. Scheinmann***,
- D. Vervloet###,
- M. Lathropf,¶¶¶,
- F. Demenais+§,f and
- F. Kauffmann*#
- the EGEA Cooperative Group
- *INSERM Unité 780,
- #Université Paris-Sud, IFR69, Villejuif,
- +INSERM Unité 946,
- §Université Paris Diderot, Paris 7, Institut Universitaire d'Hématologie,
- fFondation Jean Dausset-CEPH,
- ffHôpital Trousseau, APHP, Université Paris 6,
- ***Hôpital Necker, Faculté de Médecine Paris Descartes, Paris,
- **INSERM Unité 823,
- ##Université Joseph Fourier,
- ¶¶Centre Hospitalier Universitaire de Grenoble, Grenoble,
- ++Hôpital Arnaud de Villeneuve, Montpellier,
- §§Centre Hospitalier Lyon-Sud, Pierre Bénite,
- ###Hôpital Ste Marguerite, Marseille,
- ¶¶¶Commissariat à l'Energie Atomique, Institut de Génomique, Centre National de Génotypage, Evry, France
- ¶IRAS, Utrecht University, The Netherlands
- +++For a list of the EGEA Cooperative Group members see the Acknowledgements
- L.A.M. Smit, Institute for Risk Assessment Sciences, Division of Environmental Epidemiology, Utrecht Univeristy, P.O. Box 80178, Utrecht, The Netherlands E-mail: L.A.Smit{at}uu.nl
Abstract
Single nucleotide polymorphisms (SNPs) at chromosome 17q21 confer an increased risk of early-onset asthma. The objective was to study whether 17q21 SNPs modify associations between early respiratory infections and asthma.
Association analysis was conducted in 499 children (268 with asthma, median age 11 yrs) from the Epidemiological Study on the Genetics and Environment of Asthma (EGEA). The 12-yr follow-up data were used to assess persistent or remittent asthma in young adulthood. Respiratory infection before 2 yrs of age was assessed retrospectively.
For the 12 17q21 SNPs studied, the odds ratios (OR) for association between infection and early-onset asthma (age at onset ≤4 yrs) were higher in carriers of risk genotypes (OR 3.42–6.36) than in noncarriers (OR 1.84–2.44; p-value for interaction 0.02–0.04 for five SNPs). Risk genotypes also increased the association between infection and childhood asthma that remits in adulthood (OR 4.84–7.16 in carriers and 1.74–2.25 in noncarriers; p-value for interaction 0.008–0.05 for 10 SNPs). In children with 17q21 risk genotypes and early-life environmental tobacco smoke (ETS) exposure, associations between infection and asthma were further enhanced.
17q21 genetic variants and early ETS exposure enhance the association between early respiratory infections and early-onset asthma and childhood asthma that remits in adulthood.
- Environmental tobacco smoke exposure
- epidemiology
- gene–environment interaction
- GSDML
- ORMDL3
- respiratory infection
Infants with early lower respiratory tract infections, which are mainly caused by viruses, such as respiratory syncytial virus (RSV) or rhinovirus, are at increased risk of recurrent wheezing and childhood asthma 1–4. It is not clear whether respiratory infection is a marker identifying infants with a predisposition to develop asthma, or whether infection is causally related to the inception of asthma 5, 6. Early life is a crucial period for lung development, and it has been argued that viral infection itself may actually increase the risk of asthma by influencing the development of immune response or by altering lung function 4, 7, 8. However, a recent Danish twin pair study found no evidence to support a causal role of infection 6.
Only a minority of infants with respiratory viral infections such as bronchiolitis develop recurrent wheezing and asthma, suggesting that genetic and environmental factors may influence the association between asthma and infection 9. For instance, exposure to environmental tobacco smoke (ETS) may play such a role 10. Single nucleotide polymorphisms (SNPs) in innate immunity and other immune response genes, have been shown to be associated with severity of RSV bronchiolitis 11, 12, however there is a lack of studies investigating genetic factors that influence the association between bronchiolitis and asthma 9.
A genome-wide association study (GWAS) has shown associations between SNPs on chromosome 17q21 and childhood asthma 13, which has been replicated by several studies 14–17. In the Epidemiological Study on the Genetics and Environment of Asthma (EGEA) it has been shown that the increased risk conferred by 17q21 variants was restricted to early-onset asthma, and was further enhanced by early life ETS exposure 16. The latter finding was recently replicated in white subjects 18. It is still unclear how these genetic variants influence the risk of asthma. Interestingly, the expression of one of the 17q21 genes, ORMDL3, was found to be strongly induced by stimulation with polyinosine-polycytidylic acid in lung fibroblasts, suggesting that ORMDL3 might play a role in viral respiratory infections 14. Thus, interactions between 17q21 variants and early life events appear to play an important role in the development of early-onset asthma. It is, however, not known whether these effects persist after childhood.
The objective of the present study was to test the following two hypotheses: 1) the association between early respiratory infections and asthma, particularly early-onset asthma, is influenced by genetic variants in the 17q21 region; and 2) the modifying effect of 17q21 variants on the association between infections and early-onset asthma is further increased by ETS exposure in early life. Furthermore, we tested the same hypotheses for two additional asthma phenotypes: childhood asthma that was either persistent or remittent in young adulthood.
METHODS
Study population and design
EGEA combines a case–control study and a family study. The design and protocol have been reported in detail elsewhere 19. Probands (asthmatic patients aged 7–70 yrs) were recruited from six chest clinics in five French cities between 1991 and 1995 (EGEA1). Family members of asthmatic probands were included, either by including the proband's parents and siblings or by including the proband's spouse and children. In addition, population-based controls were recruited.
In the present study, association analyses were conducted in children (age ≤16 yrs) from the EGEA families and control sample. Of the 604 children enrolled at EGEA1, we used data from 499 children (268 with asthma, 231 without asthma) who had complete information on 17q21 genotypes, age at onset of asthma and respiratory infections (fig. E1 in the supplementary material). A 12-yr follow-up of EGEA1 has been conducted recently (EGEA2; 2003–2007) 20. Activity of asthma at EGEA2 was obtained in 236 (88%) out of 268 children with asthma at EGEA1.
All participants or their parents gave written informed consent.
Genotyping
Genotyping has been described previously 16. In the present study, we examined the 11 SNPs located between 35.23 and 35.34 Mb on chromosome 17q21 that were significantly associated with asthma (p<0.01) in EGEA families 16. We also included an additional SNP, rs7216389, since it was the most studied 17q21 variant in replication studies subsequent to the initial GWAS 13. The 12 SNPs were in strong linkage disequilibrium (LD) with one another, with pairwise LD coefficient D' between 0.87 and 1.0 16.
Health outcomes and exposure variables
Inclusion criteria used at EGEA1 to define asthma in probands, and the definition of asthma in relatives or population-based controls have been described before 19, 21. Early-onset asthma was defined as being ≤4 yrs old at the onset of asthma 16. A self-completed questionnaire was used to assess whether children with asthma at the first survey still had active asthma at EGEA2, which was defined as having had an asthma attack or used asthma medication during the last 12 months. In subjects with asthma at EGEA1, persistent and remittent asthma in young adulthood were defined according to activity of asthma (yes/no) at EGEA2. Subjects without asthma at EGEA1 were the reference group in all analyses.
Respiratory infection in early life was based on parental report at EGEA1, and defined as a positive answer to the question “Did your child have bronchitis or bronchiolitis before the age of 2?”. ETS exposure in early life (≤2 yrs of age) was defined as described previously 16.
Statistical analysis
Odds ratios (OR) and 95% CI for association between infection and asthma were estimated by using generalised estimating equations approach with the logit link function to take into account family dependence among siblings. The sibship was defined as the cluster unit. OR were adjusted for age and sex. To test whether the association of infection with asthma was influenced by 17q21 genetic variants, we introduced the infection effect, the SNP effect, and an SNP x infection term in the model while adjusting for age and sex. For 17q21 SNPs, we assumed a recessive genetic model (homozygous risk allele genotype versus other genotypes) which provided the best fit for association between the 17q21 locus and early-onset asthma as found previously 16. The SNP x infection interaction was tested by using the generalised score test which follows a Chi-squared distribution with one degree of freedom. We used the delta beta statistic to identify influential observations and we assessed the effect of excluding these observations on the results. Empirical p-values for tests of interaction between SNP and infection were also computed by simulations (i.e. marker genotypes were simulated under the null hypothesis of no SNP x infection interaction by using the observed marker allele frequencies and LD pattern among SNPs and keeping the observed phenotypic data in the EGEA children). For each SNP, the p-value was estimated by the proportion of 1,000 simulations leading to a test statistic for interaction as large as or larger than the one observed. Finally, we also tested whether the relationship between infection and asthma was modified by both 17q21 variants and exposure to ETS in early life, using a similar strategy.
RESULTS
Descriptive characteristics
Characteristics of study subjects with and without asthma at recruitment (EGEA1) are presented in table 1. As expected, objectively measured markers of asthma, such as atopy, total immunoglobulin (Ig)E, peripheral blood eosinophils, forced expiratory volume in 1 s (FEV1) and bronchial hyperresponsiveness were significantly associated with diagnosed asthma (p<0.05). The median age of the study subjects at the second survey (EGEA2) was 22.4 yrs (interquartile range 20.0–24.2 yrs). Of 167 children with early-onset asthma, 84 had persisting asthma, 62 had remittent asthma and 21 had missing data at the second survey. Of 101 children with late-onset asthma, 44 had persistent asthma, 46 had remittent asthma and 11 had missing data at the second survey.
Early infections
Early respiratory infection was significantly associated with all asthma phenotypes, but most strongly with early-onset asthma and asthma remittent in young adulthood (table 2). OR changed only marginally after adjusting for age and sex. Further adjustment for ETS or atopy did not change the associations. Associations between respiratory infection and asthma appeared stronger in boys than in girls: the OR for the association between infection and early-onset asthma was 3.70 (2.17–6.32) in boys and 2.29 (1.14–4.59) in girls, but were not significantly different (p = 0.43). There was no association between the 17q21 SNPs and early respiratory infection (p>0.1, data not shown). There were also no associations between 17q21 SNPs and infection when stratified analyses were performed in children with asthma (OR between 0.85 and 1.57; p>0.1) or without asthma (OR between 0.59 and 0.84; p>0.1).
Influence of 17q21 genotype on the association of infection with asthma
The association between respiratory infection and early-onset asthma was stronger in carriers of risk genotypes for each of the 12 17q21 genetic variants (OR ranged between 3.42 and 6.36, according to the genetic variant) than in noncarriers (OR between 1.84 and 2.44) (table 3). Statistically significant interaction between infection and 17q21 genotype was observed for five SNPs (p = 0.02–0.04; table 3). In contrast, associations between infection and asthma with onset at age ≥5 yrs did not differ significantly according to genotype; OR observed in carriers of risk genotypes were 0.7-fold to 1.7-fold OR observed in noncarriers (p-values for interaction ranging between 0.29 and 0.93; data not shown).
We then examined children with asthma in childhood with active or nonactive asthma in young adulthood (persistent or remittent asthma versus no asthma in childhood). OR for the association between infection and childhood asthma that remits in adulthood were 2.1–3.7-times higher in children with risk genotypes than in children with nonrisk genotypes (OR ranged between 4.84 and 7.16 in carriers and between 1.74 and 2.25 in noncarriers; p-value for interaction 0.008–0.05 for 10 SNPs) (table 4). The association between infection and childhood asthma persisting into young adulthood was not significantly modified by 17q21 SNPs (OR differed by a factor of 0.9–1.9, p-value for interaction ranging from 0.18–0.91; data not shown).
The delta beta statistic did not identify individuals who had a strong influence on the effect estimates shown in tables 3 and 4. We estimated the parameters of the model again without the 10 most influential observations. The p-values for interaction that were <0.05 in table 3 still reached that significance level, except for rs8076131 (p = 0.07); and the p-values for interaction that were <0.05 in table 4 still reached that significance level, except for rs903281 (p = 0.08), rs4795405 (p = 0.07) and rs4794820 (p = 0.11).
ETS exposure
We further studied the influence of both 17q21 genotypes and ETS exposure in early life on the associations between respiratory infection and asthma. Figure 1 shows association between infection and either early-onset asthma or remittent asthma according to ETS exposure and risk genotype at rs8069176; the SNP showing the most significant gene–environment interactions. The association between infection and early-onset asthma and remittent asthma was very strong among ETS exposed subjects with the rs8069176 GG genotype (early-onset asthma OR 9.98 (3.56–27.97), asthma remittent in adulthood OR 10.68 (3.58–31.82)). OR for the association between infection and early-onset asthma were significantly different between children exposed to ETS and children not exposed to ETS, but only in carriers of the risk genotype (interaction infection x ETS; p = 0.03 in children carrying rs8069176 GG). The other 11 SNPs showed similar patterns, with a very strong association between infection and early-onset asthma among children exposed to ETS and carrying the risk genotype (ORs ranged between 4.65 (1.62–13.34) for rs9303277 CC and 10.04 (3.63–27.69) for rs2305480 GG), and weaker associations in the other children (data not shown).
Estimation of the parameters of the model without the 10 most influential observations led to a slight decrease in evidence for associations between infection and early-onset asthma: among ETS-exposed subjects with the rs8069176 GG genotype, the OR was 9.18 (3.06–27.6).
DISCUSSION
This study shows that 17q21 risk genotypes increased the positive association between early respiratory infection and asthma, and this was restricted to early-onset asthma and asthma that remits in young adulthood. The association between infection and early-onset asthma (or remittent asthma) was further enhanced when children with 17q21 risk variants were exposed to ETS in early life. The present study supports earlier findings showing the importance of early life events in the association between 17q21 variants and asthma 16.
Infections and asthma
Lower respiratory tract infections may identify children with pre-existing physiological or immunological abnormalities, or may provide a first trigger to start wheezing in infants who are prone to develop asthma. Otherwise, it has been suggested that viral respiratory infections and asthma may be causally related. There is some evidence of a causal link, including recent findings from an experimental mouse model showing that innate immune activation by viral respiratory infection can lead to the development of chronic inflammatory airway disease 7, 8, 22. Infants with 17q21 risk variants may respond differently to lung injury caused by respiratory viral infections, which may increase their susceptibility to develop recurrent wheezing or asthma. However, we cannot draw conclusions regarding the causality of association between early respiratory infection and asthma, and if we assume a noncausal association between infection and asthma, an explanation for our findings might be that 17q21 variants increase the risk to develop asthma by interacting with risk factors underlying early virus-induced wheezing. Gene–gene or gene–environment interaction between 17q21 variants and common risk factors for respiratory infection and asthma (such as innate immune gene variants) could thus influence the risk of asthma.
A large genomic region of interest
The function of the studied 17q21 genes is largely unknown, especially with respect to asthma. A relationship between ORMDL3, viral infection, and asthma was suggested by Hirota et al. 14 who found that ORMDL3 expression was strongly induced in an experimental model of viral stimulation. A recent study reported that ORMDL3 facilitated endoplasmic reticulum mediated inflammatory responses 23. Among the 12 SNPs studied, one belongs to ORMDL3 while the other ones are intergenic (three SNPs) or belong to other genes, IKZF3 (one SNP), involved in the regulation of lymphocyte development, ZPBP2, or zona pellucida-binding protein 2 (one SNP), and GSDML (six SNPs), encoding one of the gasdermin proteins implicated in epithelial barrier function and skin differentiation 24. These 17q21 SNPs are strongly associated with transcript levels of ORMDL3 13, and, as found more recently, with transcript levels of GSDML, indicating that both ORMDL3 and GSDML are coregulated by cis-acting genetic variants 25. Therefore, our epidemiologic observations are consistent with biological findings which suggest that genetic variants act over a large genomic region, playing a role in transcriptional activity of at least three genes (ZPBP2, GSDML and ORMDL3) 25.
Lung function and atopy
Association between 17q21 SNPs and FEV1 was not found in EGEA and other populations, but in these populations lung function was not measured in early childhood 15, 16. There is ample evidence that diminished airway size is a risk factor for wheezing during viral infections in early life 4. It could, therefore, also be speculated that 17q21 variants modify association between a reduced pulmonary function in infancy and early-onset asthma. A reduced maximal flow at functional residual capacity (V′max,FRC) during the first year of life was associated with transient, but not persistent, early wheezing, a phenotype which is strongly related to viral infections, in the Tucson Children's Respiratory Study 26. However, it has also been shown that diminished values of V′max,FRC at 1 month are associated with persistent wheeze at 11 yrs of age in children from the Perth study 27. EGEA probands were recruited at age ≥7 yrs and, therefore, our study cannot be directly compared with birth cohort studies such as the Tucson and Perth studies 26–28, in as much as children with asthma at school age were included more frequently in EGEA.
Both atopy and early life respiratory viral infections are independently associated with subsequent asthma, and the combination of atopy and respiratory infections has been associated with a strongly increased risk to develop asthma 29. Most studies show that 17q21 SNPs are not associated with atopy or total IgE as such, but possibly with atopic asthma 15–17. However, because of the strong association between atopy and asthma in EGEA children (almost 90% of the children with asthma were atopic), we could not perform separate analyses for atopic and nonatopic asthma.
Persistent and remittent asthma in young adulthood
The longitudinal design of EGEA provides a unique opportunity to study the persistence of childhood asthma into young adulthood. The study did not have sufficient statistical power to perform separate analyses according to the age at onset among subjects with persistent or remittent asthma; however, the majority of children with asthma (62%) had a first asthma attack at age ≤4 yrs. The modifying effect of 17q21 variants on the association between infection and asthma was most pronounced for asthma that remitted in young adulthood, suggesting that the functional role of 17q21 variants may be restricted to early life events, which may be outgrown at a later age.
ETS, early events and epigenetics
We studied whether ETS exposure in early life further modified associations between respiratory infection and asthma, although this resulted in relatively small subgroups. The additional modifying effect of early life ETS exposure is, however, plausible, since ETS exposure has been shown to influence the child's lung function and immune system. Children exposed to ETS in utero or in early life exhibit small but clear deficits in lung function, and there has been clear evidence of a dose-dependent relationship between the amount smoked by mothers and lung function in children 30. It has also been shown that exposure to ETS in early life can influence signalling through innate immune receptors, which may be implicated in the increased predisposition to infection in ETS exposed infants 31. Moreover, in mouse models, cigarette smoke has been shown to affect antiviral pro-inflammatory processes 10. Verlaan et al. 25 suggested that environmental triggers and common 17q21 variants may influence epigenetic states by shifting the equilibrium in favour of one of the chromatin conformations resulting in changes in gene expression. The very early window of expression and the interaction with ETS both support the hypothesis of a role of epigenetic changes.
Limitations
Sub-group analyses led to relatively small numbers of subjects, especially when taking ETS exposure into account. We cannot rule out false negative results due to inadequate power. Although results were not modified by removal of influential subjects from the study sample, interpretation should be cautious and replication is needed to confirm our observations. Another main limitation of our study was that viral respiratory infection was assessed retrospectively by questionnaire. We did not have objective data confirming that the self-reported early-life respiratory infections were viral, although it is well-established that the majority of respiratory illnesses in infancy are caused by viral infections 32. Parents of children with asthma could have been more likely to recall a respiratory infection episode during the first two years of their child's life than parents of children without asthma. Although a recall bias may have resulted in an overestimation of the association between infection and asthma, it cannot explain the modifying effect of 17q21 variants on this association. We could not assess the role of severity, nor of specific viral infections, such as rhinovirus or RSV bronchiolitis. Recently, outpatient wheezing illness with rhinovirus infection was shown to be a much stronger predictor of subsequent asthma than wheezing with RSV 1. It would be of interest not only to replicate our findings in a prospective study, but also to explore whether 17q21 variants interact differently with rhinovirus or RSV bronchiolitis. To prevent false-positive results due to potential confounders or data stratification, p-values for tests of interaction of 17q21 SNPs with infection were computed by simulations and led to the same results as those shown here. We did not apply the Bonferroni correction to correct for multiple testing since it would be much too conservative given the high level of LD among the 12 SNPs analysed. Moreover, further haplotype analysis led to the same results as those shown here.
In conclusion, this study shows an important role of 17q21 variants, early respiratory infection and early exposure to ETS in early-onset asthma. Our longitudinal data also suggest that 17q21 variants modified the association of early respiratory infection and an asthma phenotype that does not persist into adulthood. Therefore, our results add to the evidence that the relevant window of expression for 17q21 variants is early in life.
Acknowledgments
The authors would like to thank the EGEA Cooperative Group: Co-ordination: F Kauffmann; F Demenais (genetics); I Pin (clinical aspects). Respiratory epidemiology: M Korobaeff (Egea1), F Neukirch (Egea1) (both Inserm U 700, Paris, France); I Annesi-Maesano (Inserm U 707, Paris, France); F Kauffmann, N Le Moual, R Nadif, MP Oryszczyn (Inserm U 780, Villejuif, France); V Siroux (Inserm U 823, Grenoble, France). Genetics: J Feingold (Inserm U 393, Paris, France); E Bouzigon, F Demenais, MH Dizier (Inserm U 946, Paris, France); I Gut, M Lathrop (CNG, Evry, France). Clinical centres: I Pin, C Pison (Grenoble, France); D Ecochard (Egea1), F Gormand, Y Pacheco (Lyon, France); D Charpin (Egea1), D Vervloet (Marseille, France); J Bousquet (Montpellier, France); A Lockhart (Egea1)(Cochin, Paris, France), R Matran (Marseille, France); E Paty, P Scheinmann (Necker, Paris, France); A Grimfeld, J Just (Trousseau, Paris). Data and quality management: J Hochez (Egea1) (Inserm ex-U155, Paris); N Le Moual, C Ravault (Inserm U 780, Villejuif); N Chateigner (Inserm U 794, Evry); J Ferran (Grenoble, France).
The authors thank all those who participated to the setting of the study and on the various aspects of the examinations involved. The authors are grateful to the three CIC-Inserm of Necker, Grenoble and Marseille who supported the study and in which subjects were examined at the second survey. They are indebted to all the individuals who participated, and without whom the study would not have been possible.
Footnotes
This article has supplementary material available from www.erj.ersjournals.com
Support Statement
This study was supported by the French Ministry of Higher Education and Research, University Paris Diderot-Paris 7, grants from the French Agency for Environmental and Occupational Health Safety (grant AFSSET-APR-SE-2004), the French National Agency for Research (grants ANR 05-SEST-020-02/05-9-97 and ANR 06-CEBS), Merck Sharp & Dohme (MSD), Hospital programme of clinical research (PHRC)-Paris, and GABRIEL, a multidisciplinary study to identify the genetic and environmental causes of asthma in the European Community (contract n° 01896 under the Integrated Programme LSH-2004-1.2.5-1 Post genomic approaches to understand the molecular bias of asthma aiming at a preventive or therapeutic control). L.A.M. Smit is supported by a European Academy of Allergology and Clinical Immunology-Global Allergy and Asthma European Network (EAACI-GA2LEN) exchange fellowship award.
Statement of Interest
A statement of interest for this study can be found at www.erj.ersjournals.com/misc/statements.dtl
- Received September 30, 2009.
- Accepted December 8, 2009.
- ©ERS 2010