Abstract
Genetic association studies have related the tumour necrosis factor-α gene (TNFA) guanine to adenine substitution of nucleotide -308 (-308G>A) polymorphism to increased risk of asthma, but results are inconsistent. The aim of the present study was to test whether two single-nucleotide polymorphisms, of TNFA and of the lymphotoxin-α gene (LTA), are associated with asthma, bronchial hyperresponsiveness and atopy in adults, by combining the results of two large population-based multicentric studies and conducting a meta-analysis of previously published studies.
The European Community Respiratory Health Survey (ECRHS) and Swiss Cohort Study on Air Pollution and Lung and Heart Diseases in Adults (SAPALDIA) used comparable protocols, including questionnaires for respiratory symptoms and measures of lung function and atopy. DNA samples from 11,136 participants were genotyped at TNFA -308 and LTA 252. Logistic regression employing fixed and random effects models and nonparametric techniques were used.
The prevalence of asthma was 6%. The TNFA -308G>A polymorphism was associated with increased asthma prevalence and with bronchial hyperresponsiveness. No consistent association was found for atopy. The LTA 252A>G polymorphism was not associated with any of the outcomes. A meta-analysis of 17 studies showed an increased asthma risk for the TNFA -308 adenine allele.
The tumour necrosis factor-α gene nucleotide -308 polymorphism is associated with a moderately increased risk of asthma and bronchial hyperresponsiveness, but not with atopy. These results are supported by a meta-analysis of previously published studies.
Asthma is a complex disease with both genetic and environmental components. It is characterised by obstruction of the airways of the lung and is related to atopy and bronchial hyperresponsiveness (BHR). Several chromosome regions and candidate genes have been associated with asthma, although the individual genes identified to date exhibit only modest effects and an unknown pattern of inheritance 1–3.
Tumour necrosis factor (TNF) is a potent pro-inflammatory cytokine involved in the inflammation of asthmatic airways 4. It is located within the class III region of the major histocompatibility complex (MHC) region on chromosome 6p21.3 5, which has previously been linked to asthma in various genome screens 1, 3, 6. The TNF-α gene (TNFA) and lymphotoxin-α (LT-α) gene (LTA, also called TNFB) are members of the TNF superfamily. TNFA plays an important role in generating and maintaining inflammatory responses and airway hyperreactivity 7, 8. TNF-α has been found in increased concentrations in the airways 8 and bronchoalveolar lavage fluid of asthmatic patients 9. Moreover, the TNF-α secretory response to allergens differs between atopic and nonatopic subjects 10. LTA is located closely upstream of TNFA, and both exhibit similar biological activity 11.
Polymorphisms in the two genes may affect the levels of TNF in the airways. The TNFA guanine (G) to adenine (A) substitution of nucleotide -308 (-308G>A) polymorphism, located in the promoter region of TNFA, has been associated with increased secretion and promoter activity 12. The LTA 252A>G polymorphism, located in the first intron of LTA seems to be associated with high LT-α production 13. The TNFA -308A and LTA 252G alleles have been positively associated with asthma in many 14–22 but not all studies 23–27.
The aim of the present study was to evaluate whether or not polymorphisms in TNFA and LTA were associated with asthma, BHR and atopy in adults in two large population-based European cohorts for which comparable methods had been used, the Swiss Cohort Study on Air Pollution and Lung and Heart Diseases in Adults (SAPALDIA) and the European Community Respiratory Health Survey (ECRHS). The role of TNF in asthma susceptibility was also evaluated across smoking categories. Finally, the consistency of the present results was evaluated through a meta-analysis of all published papers on polymorphisms in the two genes.
MATERIALS AND METHODS
Study population
The present study included 11,136 subjects derived from two different population cohorts. The ECRHS was a population-based multicentric cohort study. In the first phase of the study, taking place in most countries in the early 1990s, a random sample of the population aged 20–44 yrs living in the study areas was contacted and asked to complete a short questionnaire concerning respiratory symptoms 28. In a second phase, an ∼20% random subsample of the study population was contacted together with a complementary symptom subsample. The symptom subsample included all subjects reporting asthma-related respiratory symptoms in the short questionnaire who had not been selected in the random sample 29. Subjects in most centres were followed-up with a median duration of follow-up of 8.9 yrs from the first phase (ECRHS) to the second phase (ECRHS-II). For the present analysis, the population studied consisted of 5,065 subjects with interview information, from whom DNA had additionally been extracted during ECRHS-II (19 centres from 10 countries).
The second cohort study population was that of the SAPALDIA 30, 31. SAPALDIA subjects were recruited in 1991 as a random sample of adults aged 18–60 yrs from eight Swiss communities representing different language and climatic regions and varying degrees of urbanisation. The median follow-up time for SAPALDIA was 10.9 yrs. Participants with complete interview data and DNA samples available for genotyping were included (n = 6,071).
Subjects included in the present analysis could be considered to be mainly of European-Caucasian origin. Some subjects from Basle in Switzerland (n = 400) appeared in both datasets and were included only in the SAPALDIA analysis. Ethical approval was obtained for each centre from the appropriate institutional ethics committee and written consent was obtained from each participant.
Asthma, bronchial hyperresponsiveness and atopy
The ECHRS and SAPALDIA used identical questionnaires for the assessment of respiratory symptoms and asthma. Asthma was evaluated at baseline (phase I of both studies) on the basis of reported asthma symptoms and reported physician-diagnosed asthma. The presence of asthma symptoms was based on a positive response to either of two questions concerning: attack of asthma during the 12 months preceding the interview, or current use of asthma medication. Among subjects reporting an asthma attack, 67% also reported use of asthma medication. Alternative definitions of asthma employed in previous studies were also examined. “Wheeze without a cold” was defined as a positive response to the following two consecutive questions (the second asked on the basis of a positive response to the first): “Have you had wheezing or whistling in your chest at any time in the last 12 months?”; “Have you had this wheezing or whistling when you did not have a cold?” Physician-diagnosed asthma was defined as a positive response to the following question: “Have you ever had asthma and was this confirmed by a doctor?”
Data on BHR were available for a total of 8,043 subjects across the two studies. The SAPALDIA and ECRHS used identical spirometric protocols 28, 30, and consenting participants underwent bronchial challenge with methacholine chloride, administered via MEFAR® aerosol dosimeters (Mefar, Bovezzo, Italy) 32. BHR was defined as a 20% fall in forced expiratory volume in one second (FEV1) from the highest post-diluent FEV1 during methacholine challenge with a cumulative dose of 1 mg for both the ECRHS and SAPALDIA 33. BHR associated with the higher cumulative doses delivered in SAPALDIA was not taken into account. A family history of asthma was defined as a report of asthma of either of the parents.
Skin-prick tests (Pharmacia & Upjohn Diagnostics, Uppsala, Sweden) were performed in the ECRHS and SAPALDIA. Subjects atopic at baseline in both studies were defined as yielding positive test results to at least one common inhalant allergen (house dust mites (Dermatophagoides pteronyssinus), timothy grass, cat and Cladosporium herbarum).
Candidate single-nucleotide polymorphism selection and genotyping
Two single-nucleotide polymorphisms (SNPs), TNFA -308G>A (National Center for Biotechnology Information SNP ID rs1800629) and LTA 252A>G (rs909253), were selected on the basis of previous evidence of their correlation with serum levels of TNF-α and LT-α 12, 13 and their association with asthma 14–22. In the SAPALDIA, the TNFA -308G>A and LTA 252A>G polymorphisms were genotyped using a liquid-handling-assisted set-up and a fluorescent 5’-nuclease real-time PCR (TaqMan; Applera Europe, Rotkreuz, Switzerland) assay with an ABI Prism 7900 sequence detection system (Applied Biosystems, Rotkreuz, Switzerland). SNP-specific primers were designed for the PCR by Applied Biosystems (Applera Europe). The SNP-specific minor-groove-binder probes and forward and reverse primers used were as follows: TNFA -308G>A: 5’-CCCGTCC[C/T]CATGCC-3’, 5’-CCAAAAGAAATGGAGGCAATAGGTT-3’, and 5’-GGACCCTGGAGGCTGAAC-3’, respectively; and LTA 252A>G: 5’-CTGCCATG[A/G]TTCCT-3’, 5’-CAGTCTCATTGTCTCTGTCACACAT-3’, and 5’-AGAGAGAGACAGGAAGGGAACAG-3’, respectively. A random sample of 10% of all DNA samples was re-genotyped, and all genotypes were confirmed. The genotype call rate was >99%.
For the ECRHS, genotyping was performed at the Centre for Genomic Regulation of the Spanish National Genotyping Centre (Barcelona, Spain). SNPs were genotyped using the SNPlexTM platform (Applied Biosystems, Foster City, CA, USA) according to the manufacturer’s instructions and analysed on an Applied Biosystems 3730/3730xl DNA Analyzer (Applied Biosystems, Rotkreuz, Switzerland). Allele-calling was performed by cluster analysis using Genemapper (version 4.0) software (Applied Biosystems, Applera Europe). The genotype call rate was >98%. Genotyping quality was controlled in two ways. First, internal positive and negative controls provided by the manufacturer were included in the reaction plates. Secondly, six duplicate samples of two HapMap 34 reference trios were incorporated into the genotyping process. Both genotype concordance and correct Mendelian inheritance were verified. Genotype concordance was tested using SNPator, a web-based tool for genotyping management and SNP analysis developed by the Spanish National Genotyping Centre 35.
The genotyping across the two laboratories that analysed samples of the two cohorts were compared using subjects from the centre in Basle (n = 400), who had been included in both the ECRHS and SAPALDIA. The agreement in genotyping was 99.8%. In addition, the Basle ECRHS samples had been previously genotyped for the TNFA -308 marker using both restriction fragment length polymorphism and allele-specific PCR methods 36. Only very small differences in genotype distribution were observed between those results and the results reported in the present study (Chi-squared = 0.02; degrees of freedom (df) = 2; p = 0.99).
Meta-analysis
Previous articles on the association of the TNFA -308G>A or LTA 252A>G polymorphism with asthma were sought on PubMed, and backward searches of articles cited in earlier literature reviews or original papers were conducted. The keywords used in the PubMed search were “asthma AND gene AND (tumour necrosis factor OR TNF)” for TNFA -308G>A and “asthma AND gene AND (lymphotoxin OR LTA OR TNFB)” for LTA 252A>G.
In the meta-analysis, all studies that met the following criteria were included: 1) design: either population-based cohort, case–control or cross-sectional study; 2) outcome: asthma defined as physician-diagnosed 18, 20, 23, 26, 27, 37–40 or self-reported 16, 17, 19, 25, 29, 41, 42, regardless of age of onset; 3) ethnicity: information available, or, if not available in the published report, available through contact with the authors 41; 4) method of genotyping reported; 5) complete genotype information available for subjects; and 6) genotypes in Hardy–Weinberg equilibrium. Results from the SAPALDIA and ECRHS were included in the meta-analysis using asthma symptoms and physician-diagnosed asthma as the main outcomes definition.
Statistical analysis
The statistical analysis was performed using the R genetic package (version 1.2.1) of R statistical software (version 2.4.0) 43. Exact tests were used to test for Hardy-Weinberg equilibrium in control subjects (subjects without asthma symptoms, physician-diagnosed asthma, atopy or BHR) 44. The normalised disequilibrium constant D' and Chi-squared p-values for marker independence were estimated in order to determine linkage disequilibrium between both genetic markers.
Logistic regression analysis was performed in order to determine the adjusted associations between genotypes and disease using co-dominant and additive models. The odds ratio (OR) and p-values corresponding to the 95% confidence interval (CI) were computed using the generalised linear models procedures (glm) from the R statistical package. A p-value of <0.05 was considered significant. Logistic regression models were adjusted for country (ECRHS) or study area (SAPALDIA), sex, age, body mass index (BMI) and smoking status. Haplotype-specific adjusted associations were also evaluated. Haplotypes were reconstructed and analysed using the haplo.glm function of the R library HaploStats.
The impact of population stratification in the present data was assessed by analysing 23 unlinked SNPs (online supplementary material) using a genomic control approach 45. These SNPs were genotyped in the ECRHS study and in a subsample of the SAPALDIA. The significance of the additive model was corrected by the inflation factor (λ) derived from genomic control for each of the three main outcomes in both the ECRHS and the pooled analysis.
Multifactor dimensionality reduction (MDR) was used on genetic and nongenetic potential determinants jointly in order to find genotype combinations within which the dichotomous outcome variability was much lower than between combinations 46–48. It is an extension of the combinatorial partitioning method and can be seen as a data reduction technique in that it reduces the dimensionality of multilocus information to a single dimension. The method is nonparametric, assumes no particular genetic model and generates low false-positive rates 47.
In the meta-analysis, the exact test of Hardy–Weinberg was used to test deviations from Hardy–Weinberg equilibrium only in control groups. ORs were estimated for each study using Fisher’s exact test of independence for 2×2 (using Fisher’s exact test for count data) tables under the prior hypothesis that the rare allele confers susceptibility to asthma. Meta-analysis was performed using the Mantel–Haenszel method using the fixed effects and random effects model with R library rmeta version 2.14. Publication bias was evaluated by measuring the asymmetry of the funnel plot measuring the intercept from regression of standard normal deviates against precision 49.
RESULTS
The characteristics of participants in the ECRHS and SAPALDIA are presented in table 1⇓. Both populations were comparable with regard to sex, BMI and pulmonary function (FEV1 and forced vital capacity), but not with regard to mean age due to differences in the inclusion criteria for age. Smoking status also differed slightly between studies, but the smoking prevalence in the SAPALDIA is within the range observed between different centres in the ECRHS.
Characteristics of participants in the European Community Respiratory Health Survey(ECRHS) and Swiss Cohort Study on Air Pollution and Lung and Heart Diseases in Adults (SAPALDIA) at baseline
The prevalence of atopy and asthma and minor allele frequency (MAF) of TNFA -308G>A and LTA 252A>G are shown by study and country in table 2⇓. Asthmatics (and consequently also atopics) were oversampled in the ECRHS since the subcohort with respiratory symptoms was included. Among the random samples of the ECRHS and SAPALDIA, the prevalence of atopy ranged from 15% in Germany (ECRHS) to 42% in Australia (ECRHS). The prevalence of asthma symptoms ranged from 2% in Germany, Estonia, Spain and Belgium to 7% in the UK and Australia. Even higher prevalences were reported for physician-diagnosed asthma.
Symptom prevalence at baseline and minor allele frequency(MAF) by study site
Geographical differences were also observed for the MAFs of both polymorphisms. The lowest MAFs were observed in France (12% for TNFA -308G>A and 27% for LTA 252A>G) and the highest in the UK (21% for TNFA -308G>A and 41% for LTA 252A>G). The two polymorphisms were in strong linkage disequilibrium (Chi-squared = 7516.29; D' = 0.98; r2 = 0.60; p<2.22×10−16). The genotype distribution for both alleles was consistent with the Hardy–Weinberg equilibrium in the control group (p>0.05), except for the French ECRHS centre in Grenoble (p<0.001). The analyses presented in the current article include data from Grenoble. The results were only minimally modified when all of the analyses were repeated excluding this centre (data not shown). No strong effects of population stratification were detected in the present sample, obtaining λ of ∼1 (1.05 in atopy, 1.06 in asthma and 1.30 in BHR).
The associations of TNFA -308G>A with atopy, asthma symptoms and BHR adjusted for country (ECRHS) or centre (SAPALDIA), sex, age, BMI and smoking status are summarised in table 3⇓. A significant association was found for asthma symptoms and TNFA -308GA heterozygotes (OR = 1.38; p = 0.001) and for the TNFA -308A allele (OR = 1.30; p = 0.002). An analysis of TNFA -308G>A by study showed an increased risk of asthma symptoms in the ECRHS but not in the SAPALDIA (tables 3⇓ and 4⇓; fig. 1⇓). The result of the test for heterogeneity between the two studies was significant (Q-statistic 5.92; df = 1; p = 0.015). In the ECRHS, a significant risk increase for asthma symptoms was observed for the GA and AA genotypes and the A allele (OR = 1.49; p = 6.3×10−5). Stratification by country in the ECRHS showed no differences in risk (Q = 2.66; df = 8; p = 0.95; table 4⇓; fig. 1⇓) and an increased risk (OR>1) was observed for all countries. ORs for the random subsample of the ECRHS tended to be lower than those of the asthma-enriched subsample (data not shown). In the SAPALDIA, no difference in effect of TNFA -308G>A was observed between Latin- and German-speaking regions (Q = 0.44; df = 1; p = 0.51). Exclusion of SAPALDIA subjects who were aged >45 yrs at baseline (so as to compare with a similar population structure as in the ECRHS) did not affect the risk estimates for asthma. The OR for the A allele in all SAPALDIA subjects was 0.94 (p = 0.71), whereas the OR for SAPALDIA subjects aged <45 yrs was 0.89 (p = 0.57). The observed associations between TNFA -308G>A and asthma symptoms were not modified by either sex or atopy. In both the ECRHS and SAPALDIA, TNFA -308G>A was associated with a slight increase in BHR prevalence (A allele OR = 1.15; p = 0.03), with similar risks found in the two studies (A allele OR = 1.13 and 1.18, respectively). No significant association was found for TNFA -308G>A with atopy (table 3⇓). The ORs for the random subsample of the ECRHS tended to be lower than those of the asthma-enriched subsample (data not shown).
Meta-analysis of tumour necrosis factor-α gene guanine to adenine (A) substitution of nucleotide -308 polymorphism and asthma symptoms, including the European Community Respiratory Health Survey (ECRHS) and Swiss Cohort Study on Air Pollution and Lung and Heart Diseases in Adults (SAPALDIA). Data are presented as A allele odds ratio (OR; ▪ (size reflects weighting)) for asthma symptoms and 95% confidence interval (CI; horizontal bars). The centres of the diamonds indicate the combined mean effect of the studies and their extremities the 95% CI; ……: line of no effect. The arrowheads indicate the positions of ORs of low weight. #: British and Irish; ¶: South Asian; +: North India; §: West India.
Adjusted association of tumour necrosis factor-α gene nucleotide -308 genotype with atopy, asthma symptoms and bronchial hyperresponsiveness (BHR)
Tumour necrosis factor-α gene guanine to adenine (A) substitution of nucleotide -308 polymorphism and asthma symptoms in the studies used for the meta-analysis and study weighting
The strength of the association of TNFA -308G>A genotypes was different for distinct asthma-related phenotypes. ORs for the A allele and for distinct phenotypes are shown in table 5⇓. A positive association was found for most, but not all, phenotypes examined, although the differences between ECRHS and SAPALDIA remained with regard to phenotypes based on reported asthma symptoms.
Adjusted association of tumour necrosis factor-α gene -308 adenine allele with various asthma-related phenotypes and their combinations in the two cohorts
Results for LTA 252A>G are summarised in table 6⇓. Overall, ORs tended to be lower than for TNFA -308G>A. A significantly increased risk of asthma symptoms was observed for the heterozygous GA genotype (OR = 1.21; p = 0.05), which was primarily driven by an association of the G allele with asthma in the ECRHS sample (OR = 1.22; p = 0.01).
Adjusted association of lymphotoxin-α gene nucleotide 252 genotype with atopy, asthma symptoms and bronchial hyperresponsiveness (BHR)
Haplotype analysis
The 252G/-308A haplotype was positively associated with asthma symptoms (OR = 1.29; p = 0.003) compared to the most common haplotype, 252A/-308G. The prevalence of the 252A/-308A haplotype was <0.01 and could, therefore, not be tested for an association with asthma. No significant overall results were found for the 252G/-308G haplotype (online supplementary material).
Smoking effect
An association of TNFA -308G>A (GA and AA genotypes compared to GG genotype) with asthma was significant among never-smokers (OR = 1.33; p = 0.03) and ex-smokers (OR = 1.64; p = 0.01), whereas a nonsignificant excess risk was found for current smokers (OR = 1.20). This differential effect was only observed in the ECRHS and not in the SAPALDIA. In the ECRHS, the effect of TNFA -308G>A was significant among never-smokers (OR = 1.50; p = 0.01) and ex-smokers (OR = 2.36; p = 0.0009). However, the p-value for interaction was nonsignificant when examining never- versus ever-smokers (p = 0.52) or smoking status in three categories (p = 0.23).
Multifactor dimensionality reduction
In the nonparametric MDR analysis, multiple genetic loci and environmental exposures associated with asthma were detected simultaneously in the absence of a main effect. An increasing number of interactions was examined, starting from including singular effects up to three-way interactions, and the best fit of each combination of variables was tested through permutation tests using 5,000 replications. Among the seven variables examined, significant results were found for a model including LTA 252A>G, smoking and region (with 5,000 permutations, permutation-based p-value of <0.001). The best fit was for a model with two variables included, smoking and region (permutation-based p-value of <0.001).
Meta-analysis
The meta-analysis of TNFA -308G>A was performed including the country-specific results from ECRHS countries and the two region-specific results from the SAPALDIA as individual observations along with the 16 published studies listed in table 4⇑ and figure 1⇑. The funnel plot for publication bias was not asymmetric (p>0.10), suggesting a lack of publication bias in the present meta-analysis. The ECRHS and SAPALDIA results were based on asthma symptoms. In total, 4,341 cases of asthma and 13,459 controls were examined (table 4⇑; fig. 1⇑). The combined OR for asthma using a fixed-effects model was 1.32 (p<0.001) and using a random-effects model 1.35 (p = 0.001). Similarly to a previous meta-analysis 37, significant heterogeneity was observed between the Caucasian populations (Q = 56.57; df = 17; p<0.001). Excluding the study of Albuquerque et al. 26, which had an MAF of 31% that was different from all other studies, and the study of Shin et al. 27, with a different MAF to other Asian population studies (17%), the heterogeneity diminished considerably (Q = 9.64; df = 15; p = 0.84). The combined OR in Caucasian populations after the exclusion of the study of Albuquerque et al. 26 was 1.24 (p<0.0001) for the fixed-effects model and 1.29 (p = 0.001) for the random-effects model. Heterogeneity was nonsignificant in Asian populations, although the ORs for a study in Korea 27 were markedly different from those of all other studies. The same risk was observed when physician-diagnosed was included instead of asthma symptoms in the meta-analysis, with an OR of 1.27 (p<0.001). A meta-analysis of 10 previous studies on LTA 252A>G and asthma 14, 17, 19, 20, 25–27, 38, 40, 41, 42 was also performed, including the data from the present study. The meta-analysis examined a total of 3,120 cases and 12,026 controls, with the combined OR being 1.01 (95% CI 0.99–1.16; p = 0.06) for the fixed-effects model and 1.08 (95% CI 0.95–1.23; p = 0.22) for the random-effects model.
DISCUSSION
A genetic association study on the TNFA -308G>A and LTA 252A>G polymorphisms with asthma and related phenotypes was performed in two large European prospective cohorts. This is the largest association study of TNFA -308G>A/LTA 252A>G in subjects well phenotyped for atopy and respiratory symptoms using validated questionnaires and measures of lung function, BHR and atopy. A moderate but significant association of the TNFA -308GA genotype and the TNFA -308A allele with increased asthma risk was found. The risks for BHR were also increased but to a lesser extent, whereas no consistent associations were found for atopy. The present results were verified using alternative nonparametric analyses and a meta-analysis of all published studies. Weaker associations were found for LTA. The effects of TNFA can be attributed to linkage disequilibrium with other asthma genes, to a main effect of TNFA or to an interaction and modification of the effect by environmental exposures.
The two polymorphisms examined are located in the MHC class III region, near human leukocyte antigen (HLA)-B 5. The TNFA -308A allele is in strong linkage disequilibrium with the HLA-A1, -B8 and -DR3 alleles 12, which are also associated with higher levels of TNFA -308G>A 50. Although some studies suggest that the LTA 252A>G/TNFA -308G>A association is independent of MHC class II alleles 21, 42, Moffat et al. 14 found that two haplotypes containing the TNFA -308A allele (LTA 252G/TNFA -308A/HLA-DRB1*3 and LTA 252G/TNFA -308A/HLA-DRB1*2) were more strongly associated with asthma and BHR than other haplotypes containing only LTA 252A>G/TNFA -308G>A polymorphisms. Identification of the individual effects of LTA 252A>G and TNFA -308G>A is difficult due to the linkage disequilibrium. The present results indicate that TNFA -308G>A is associated with several asthma-related phenotypes, whereas less consistent results were observed for the main effects of LTA 252A>G. Haplotype analysis showed that only the model with the TNFA -308A allele was associated with asthma, and that this association was equivalent to the model with TNFA -308G>A alone. This evidence suggests that the associations observed for LTA 252A>G in asthma and BHR may be due to its linkage with TNFA -308G>A.
TNF-α is a potent pro-inflammatory cytokine found in high concentrations in bronchoalveolar fluid from asthmatics 8, 9, 51. TNFA nucleotide -308 is located in the promoter region of TNFA, and in vitro studies have reported increased TNFA transcription associated with the TNFA -308G>A variant 12, 52. The A allele has been associated with increased expression 12 and secretion of TNF-α 52, 53, although this association is not uniformly supported 54, 55. No association of the TNFA A allele with atopy was observed, whereas it was observed for asthma symptoms and BHR. Furthermore no modification of the association of TNFA -308G>A and asthma by atopic status was observed. One study 20 reported that TNFA -308G>A was a risk factor for atopic asthma but not for nonatopic asthma, although a number of other studies did not find this interaction 23, 27, 42.
In the present study, the pattern of the interaction between TNF polymorphisms and tobacco smoke was not clear. TNF-α is central to acute cigarette-smoke-induced inflammation and the resulting connective tissue breakdown. Oxidative stress involved in inflammation is partly regulated by cytokines such as TNF-α. TNFA -308G>A effects on the inflammatory response to oxidative stress have been suggested in other studies in relation to exposure to ozone, occupational endotoxin and environmental tobacco smoke, but the results have been inconsistent. Interaction of the LTA 252A>G and TNFA -308G>A polymorphisms with environmental factors has been shown in several studies in relation to smoking, ozone and other environmental or occupational exposures, although the results have not been consistent 56–59.
The present study had low power for the evaluation of differences between countries and regions due to the low MAF of the TNFA -308A allele. No effect of population stratification was observed in the present sample, although the number of markers tested might be insufficient for the detection of lower stratification between European populations 60. The association analyses of both markers revealed differences in risk between the ECRHS and SAPALDIA cohorts for asthma symptoms, whereas more consistent results were obtained for BHR and atopy. The study protocols of both studies were similar, and phenotypes were defined using the same questions. Nevertheless, differences across the two cohorts tended to relate primarily to asthma phenotypes, defined by questionnaire, rather than to phenotypes derived from functional or biological tests. However, a study assessing the internal consistency of respiratory symptoms suggests that international comparisons are not affected by errors due to cross-cultural variations in the reporting of symptoms 61. The major difference between the two studies was the inclusion in the ECRHS of a subsample with subjects that reported respiratory symptoms in the initial screening questionnaire. This subsample represented >60% of the asthma cases in the ECRHS and led to differences in symptom prevalences compared to the SAPALDIA. The magnitude of asthma risk in relation to TNFA genotypes in the ECRHS after exclusion of the asthma-enriched sample tended to be to weaker, although still positive, indicating that inclusion of the asthma-enriched sample could not explain the differences in risk between the two studies. The slightly stronger association observed in the oversampled asthmatics from the ECRHS might, possibly, reflect an underlying stronger effect of the TNFA -308G>A genotype in more severe forms of asthma and in asthma persistence as well as progression. There is, however, only very limited evidence supporting such an association 15, 62. Other potential differences, such as errors in genotyping, could be disregarded. Given the above, the most likely explanation for the absence of an association with self-reported asthma phenotypes in the SAPALDIA is the lack of statistical precision due to a much lower number of cases compared to the ECHRS. The meta-analysis of all published studies reinforces the present results of a positive association between the TNFA -308G>A genotype and asthma prevalence, although some geographical variability could be observed. An inverse association was observed in only two studies 26, 27. However, the combined estimate confirms the association of TNFA -308G>A with an increased risk of asthma in European and Asian populations. Population stratification is a concern in large genetic association studies with heterogeneous population 60. The subjects in the present analysis were almost entirely of European ancestry. However, even within the present study population, differences in allelic frequency were observed between countries, with higher MAFs obtained in the UK. The two previous studies that found an inverse association with asthma risk 26, 27 were outliers with regard to TNFA -308A allele frequency. In the present meta-analysis, large variation in the allelic frequency of the TNFA -308A allele was observed between populations, with lower frequencies in those from Korea, Japan, China and Taiwan. One limitation of the present meta-analysis is that the selection criteria were broad, leading to the inclusion of different age groups and definitions of asthma (e.g. paediatric asthma, adult asthma and atopic asthma), and the biological mechanisms involved in each of these asthma-related phenotypes might be different. This is particularly important given the differences in risk observed for different asthma phenotypes in the present study. Since there are limited published data on BHR, it was not possible to perform a meaningful meta-analysis for this end-point.
Conclusions
In the present large international population-based prospective study, the tumour necrosis factor-α gene guanine to adenine substitution of nucleotide -308 polymorphism was associated with a moderately increased risk of asthma and bronchial hyperresponsiveness, whereas no association was found for atopy. These results were supported by a meta-analysis of the published evidence.
Support statement
This sudy was supported by grants from the MaratoTV3 Foundation (Sant Joan Despí, Spain), Swiss National Science Foundation (Berne, Switzerland), Zurich Lung League (Zurich, Switzerland) and Genome Spain Foundation (Madrid, Spain).
Statement of interest
A statement of interest for P. Burney can be found at www.erj.ersjournals.com/misc/statements.shtml
Acknowledgments
Author affiliations are as follows. F. Castro-Giner and J.R. Gonzalez: Centre for Research in Environmental Epidemiology (CREAL), Barcelona; Municipal Institute of Medical Research (IMIM-Hospital del Mar), Barcelona; and CIBER Epidemiologia y Salud Pública (CIBERESP); all Spain. M. Kogevinas: Centre for Research in Environmental Epidemiology (CREAL), Barcelona; Municipal Institute of Medical Research (IMIM-Hospital del Mar), Barcelona; and CIBER Epidemiologia y Salud Pública (CIBERESP); all Spain; and Medical School, University of Crete, Heraklion, Greece. M. Mächler and N.M. Probst-Hensch: Institute of Social and Preventive Medicine, and Institute of Surgical Pathology, Molecular Epidemiology/Cancer Registry, University of Zurich/University Hospital Zurich, Zurich, Switzerland. R. de Cid: Genes and Disease Program, Center for Genomic Regulation, Barcelona National Genotyping Center, Barcelona; and CIBER Epidemiologia y Salud Pública (CIBERESP); both Spain. K. Van Steen: Dept of Applied Mathematics and Computer Science, Ghent University, Ghent; Ghent University Hospital, Ghent; and StepGen CVBA, Merelbeke; all Belgium. M. Imboden: Institute of Social and Preventive Medicine, and Institute of Surgical Pathology, Molecular Epidemiology/Cancer Registry, University of Zurich/University Hospital Zurich; and Institute of Medical Genetics, Division of Medical Molecular Genetics and Gene Diagnostics, University of Zurich; both Zurich, Switzerland. C. Schindler: Institute of Social and Preventive Medicine, University of Basle, Basle, Switzerland. W. Berger: Institute of Medical Genetics, Division of Medical Molecular Genetics and Gene Diagnostics, University of Zurich, Zurich, Switzerland. K.A. Franklin: Dept of Respiratory Medicine, University Hospital, Umeå, Sweden. C. Janson: Dept of Medical Sciences, Respiratory Medicine and Allergology, Uppsala University, Uppsala, Sweden. D. Jarvis and P. Burney: Respiratory Epidemiology and Public Health Group, National Heart and Lung Institute, Imperial College, London, UK. E. Omenaas: Haukeland University Hospital, Bergen, Norway. T. Rochat: Division of Pulmonary Medicine, University Hospitals of Geneva, Geneva, Switzerland. X. Estivill: Genes and Disease Program, Center for Genomic Regulation, Barcelona National Genotyping Center, Barcelona; Universitat Pompeu Fabra, Barcelona; and CIBER Epidemiologia y Salud Pública (CIBERESP); all Spain. J.M. Antó: Centre for Research in Environmental Epidemiology (CREAL), Barcelona; Municipal Institute of Medical Research (IMIM-Hospital del Mar), Barcelona; Universitat Pompeu Fabra, Barcelona; and CIBER Epidemiologia y Salud Pública (CIBERESP); all Spain. M. Wjst: German Research Centre for Environmental Health, Helmholtz Centre, Munich, Germany.
Swiss Cohort Study on Air Pollution and Lung and Heart Diseases in Adults
The present study could not have been performed without the help of the study participants, technical and administrative support, and the medical teams and fieldworkers at the local study sites, as well as the entire Swiss Cohort Study on Air Pollution and Lung and Heart Diseases in Adults (SAPALDIA) team. The authors also thank E. Glaus for DNA extraction and O. Senn for genotyping (both University of Zurich, Zurich, Switzerland).
Local fieldworkers: Aarau: M. Broglie, M. Bünter, and D. Gashi; Basle: R. Armbruster, T. Damm, U. Egermann, M. Gut, L. Maier, A. Vögelin, and L. Walter; Davos: D. Jud, and N. Lutz; Geneva: M. Ares, M. Bennour, B. Galobardes, and E. Namer; Lugano: B. Baumberger, S. Boccia Soldati, E. Gehrig-Van Essen, and S. Ronchetto; Montana: C. Bonvin, and C. Burrus; Payerne: S. Blanc, A.V. Ebinger, M.L. Fragnière, and J. Jordan; and Wald: R. Gimmi, N. Kourkoulos, and U. Schafroth. Administrative staff: N. Bauer, D. Baehler, C. Gabriel, and R. Nilly.
SAPALDIA team
Study directorate: U. Ackermann-Liebrich (epidemiology), J.M. Gaspoz (cardiology), P. Leuenberger (pneumology), L.J.S. Liu (exposure), N.M. Probst Hensch (epidemiology/genetic and molecular biology), C. Schindler (statistics), and T. Rochat (pneumology).
Scientific team: J.C. Barthélémy (cardiology), W. Berger (genetic and molecular biology), R. Bettschart (pneumology), A. Bircher (allergology), G. Bolognini (pneumology), O. Brändli (pneumology), M. Brutsche (pneumology), L. Burdet (pneumology), M. Frey (pneumology), M.W. Gerbase (pneumology), D. Gold (epidemiology/cardiology/pneumology), W. Karrer (pneumology), R. Keller (pneumology), B. Knöpfli (pneumology), N. Künzli (epidemiology/exposure), U. Neu (exposure), L. Nicod (pneumology), M. Pons (pneumology), E. Russi (pneumology), P. Schmid-Grendelmeyer (allergology), J. Schwartz (epidemiology), P. Straehl (exposure), J.M. Tschopp (pneumology), A. von Eckardstein (clinical chemistry), J.P. Zellweger (pneumology), and E. Zemp Stutz (epidemiology).
Scientific team at coordinating centres: P.O. Bridevaux (pneumology), I. Curjuric (epidemiology), S.H. Downs (epidemiology/statistics), D. Felber Dietrich (cardiology), A. Gemperli (statistics), D. Keidel (statistics), M. Imboden (genetic and molecular biology), P. Staedele-Kessler (statistics), and GA Thun (genetic and molecular biology).
European Community Respiratory Health Survey
Steering Committee: F. Neukirch and B. Leynaert (Paris, France); J. Heinrich and M. Wjst (Erfurt, Germany); C. Svanes (Bergen, Norway); J.M. Antó and J. Sunyer (Barcelona, Spain); C. Janson (Uppsala, Sweden); N. Künzli (Basle, Switzerland); and P. Burney, S. Chinn and D Jarvis (London, UK).
Principal Investigators and Senior Scientific Team: Australia: M. Abramson, R. Woods, E.H. Walters, F. Thien and G. Benke (Melbourne); Belgium: P. Vermeire, J. Weyler, M. Van Sprundel and V. Nelen (South Antwerp/Antwerp City); Estonia: R. Jogi and A. Soon (Tartu); France: R. Liard and M. Zureik (Paris), and I. Pin and J. Ferran-Quentin (Grenoble); Germany: C. Frye and I. Meyer (Erfurt); Norway: A. Gulsvik, E. Omenaas and B. Laerum (Bergen); Spain: M. Kogevinas, J.P. Zock, X. Basagana, A. Jaen, F. Burgos and F. Castro (Barcelona), J. Maldonado, A. Pereira and J.L. Sanchez (Huelva), J. Martinez-Moratalla Rovira and E. Almar (Albacete), N. Muniozguren and I. Urritia (Galdakao), and F. Payo (Oviedo); Sweden: G. Boman, D. Norback and M. Gunnbjornsdottir (Uppsala), K. Toren, L. Lillienberg, A. Dahlman-Höglund and R. Sundberg (Gothenburg), and E. Norrman, M. Soderberg, K. Franklin, B. Lundback, B. Forsberg and L. Nystrom (Umeå); Switzerland: B. Dibbert, M. Hazenkamp, M. Brutsche and U. Ackermann-Liebrich (Basle); and UK: B. Harrison (Norwich), and R. Hall and D. Seaton (Ipswich).
Footnotes
-
This article has supplementary material accessible from www.erj.ersjournals.com
- Received November 19, 2007.
- Accepted March 11, 2008.
- © ERS Journals Ltd
References
