Abstract
Reduced exposure to particulate matter with a 50% cut-off aerodynamic diameter of 10 μm (PM10) attenuated age-related lung function decline in our cohort, particularly in the small airways. We hypothesised that polymorphisms in glutathione S-transferase (GST) and haem oxygenase-1 (HMOX1) genes, important for oxidative stress defence, modify these beneficial effects.
A population-based sample of 4,365 adults was followed up after 11 yrs, including questionnaire, spirometry and DNA blood sampling. PM10 exposure was estimated by dispersion modelling and temporal interpolation. The main effects on annual decline in forced expiratory flow at 25–75% of forced vital capacity (FEF25–75%) and interactions with PM10 reduction were investigated for polymorphisms HMOX1 rs2071746 (T/A), rs735266 (T/A) and rs5995098 (G/C), HMOX1 (GT)n promoter repeat, GSTM1 and GSTT1 deletions, and GSTP1 p.Ile105Val, using mixed linear regression models.
HMOX1 rs5995098, HMOX1 haplotype TTG and GSTP1 showed significant genetic main effects. Interactions with PM10 reduction were detected: a 10 μg·m−3 reduction significantly attenuated annual FEF25–75% decline by 15.3 mL·s−1 only in the absence of HMOX1 haplotype ATC. Similarly, carriers of long (GT)n promoter repeat alleles or the GSTP1 Val/Val genotype profited significantly more from a 10 μg·m−3 reduction (26.5 mL·s−1 and 27.3 mL·s−1 respectively) than non-carriers.
Benefits of a reduction in PM10 exposure are not equally distributed across the population but are modified by the individual genetic make-up determining oxidative stress defence.
- Exposure to particles with a 50% cut-off aerodynamic diameter of 10 μm
- forced expiratory flow at 25–75% of forced vital capacity
- general population sample
- glutathione S-transferase
- haem oxygenase-1
- lung function decline
Several studies have shown deleterious effects of air pollution on lung function, including our own 1–3. The underlying inflammatory reactions, triggered by free radicals present in or induced by inhaled dusts, fumes, tobacco smoke and environmental air pollution 4, lead to adverse health effects in the long term 5. Exposure to radicals is termed “oxidative stress”, and its contribution to lung diseases supports the hypothesis that polymorphisms in genes involved in oxidative stress defence importantly determine susceptibility to external noxious exposures. Polymorphisms of the haem oxygenase-1 (HMOX1) gene and glutathione S-transferase (GST) super-gene family are interesting candidates in this respect.
The HMOX1 gene is rapidly upregulated in the alveolar cell layer after exposure to oxidative stress and forms part of the lung's first line of defence 6. The enzyme catalyses haem degradation, producing bilirubin, CO and ferritin, which exert important antioxidative and anti-inflammatory effects 6. A microsatellite polymorphism, i.e. a genetic variant consisting of a variable number of short nucleotide sequence repetitions (in this case “GT”), is known in the promoter region of the HMOX1 gene. In general population samples from France and the Netherlands, long repeat alleles (≥33 repeats) were associated with accelerated decline in several lung function parameters, especially among smokers 7, 8.
The GST super-gene family comprises various isoforms, including GSTT1, GSTM1 and GSTP1 9. GSTs are phase II biotransformation enzymes and important antioxidants. Their substrate, glutathione (GSH), is abundant in the alveolar fluid line and its synthesis is upregulated upon oxidative stress 10. A single nucleotide polymorphism (SNP) in GSTP1, i.e. Ile105Val leads to the substitution of isoleucin (Ile) for valine (Val) in protein synthesis and alters the detoxification of diolepoxides 11. In GSTT1 and GSTM1, homozygous gene deletions with complete loss of protein function are highly prevalent 9. Genetic variation in GSTT1, GSTM1 and GSTP1 has been associated with accelerated lung function decline in the adult general population 12 and reduced lung function growth in children 13.
The Swiss Study on Air Pollution and Lung Diseases in Adults (SAPALDIA) recently showed that a reduction in long-term exposure to particulate matter with a 50% cut-off aerodynamic diameter of 10 μm (PM10) was associated with a significant attenuation of the age-related decline in lung function over 11 yrs of follow-up 2. Attenuations were strongest for forced expiratory flow at 25–75% of forced vital capacity (FEF25–75%), and weakened with increasing baseline PM10 exposure. In the present study we thus focused on the decline in FEF25–75% and investigated whether the effect of PM10 reduction differed according to the participants' genetic make-up regarding HMOX1, GSTM1, GSTT1 and GSTP1 genes, which all play a major role in the oxidative stress defence of the body.
METHODS
Study design and population
The methods of the SAPALDIA study have been described in detail elsewhere 14. In short, the study population consists of a random population sample of adults aged 18–60 yrs from eight areas of Switzerland. A total of 9,651 people participated in the first assessment in 1991, and at the follow-up examination in 2002, 8,047 were reassessed. Written consent was obtained from all study participants. Approval of the study was given by the Swiss Academy of Medical Sciences (Basel, Switzerland) and the regional ethics committees.
The present study sample consists of 4,365 participants with complete data regarding spirometry, residential history, smoking history, PM10 exposure and available genotyping information (fig. 1).
Selection of study participants. PM10: particulate matter with a 50% cut-off aerodynamic diameter of 10 μm.
Assessment procedures
Health questionnaire
At both surveys, participants underwent a computer-assisted interview comprising questions about smoking behaviour, exposure to environmental tobacco smoke at home and at the workplace, workplace exposure to dust and fumes, presence of respiratory symptoms or chronic illness, medication use and socio-economic factors. Never-smokers were defined as people who, at the time of follow-up, had smoked fewer than 20 packs of cigarettes or 360 g of tobacco during their lifetime, and smokers and ex-smokers reported current smoking or quitting at least 1 month before follow-up, respectively. For smokers and ex-smokers, smoking history in pack-years before the first assessment and between surveys were included in the analyses.
Pulmonary function testing
Spirometry testing was performed according to the European Community Respiratory Health Survey protocol 15 and complied with American Thoracic Society criteria 16. Device checks and calibrations were performed on a daily basis during assessment periods. The same spirometric devices were used at both examinations (Sensor Medics 2200 SP; Sensor Medics, Yorba Linda, CA, USA). Forced vital capacity (FVC), forced expiratory volume in 1 s (FEV1) and FEF25–75% were recorded. Annual decline in FEF25–75% was calculated by subtracting the baseline measurement from the follow-up value and dividing the difference by the individual time of follow-up in years.
Atopy testing
In 1991, skin-prick testing was performed for cat fur and dog epithelia, timothy grass, Parietaria, birch, house-dust mite, Alternaria tenuis and Cladosporium herbarum. A person was considered atopic if the mean diameter of at least one wheal was 3 mm greater than that of a control sting free of antigen 17, 18. The mean diameter was calculated by averaging the sum of the widest diameter and the longest line perpendicular to it within the wheal perimeter.
Genotyping procedures
A detailed description of the genotyping procedures and specific genetic terms is given in the online supplementary document.
Briefly, genomic DNA was extracted manually from EDTA-buffered whole blood.
Genotyping of GSTT1, GSTM1 gene deletions and GSTP1 p.Ile105Val SNP in the SAPALDIA cohort have been previously reported 12.
Three SNPs of the HMOX1 gene, rs2071746 (T/A), rs5995098 (G/C) and rs735266 (T/A) were genotyped using real-time PCR (TaqMan®; Applied Biosystems, Rotkreuz, Switzerland). HMOX1 haplotypes were inferred using PHASE software v2.1 19, 20. Haplotypes with prevalences below 5% were not analysed.
HMOX1 (GT)n promoter repeats were determined using PCR and fragment size analysis with GeneMapper® software v3.5 (Applied Biosystems). For comparability with previous studies showing associations of ≥33 repeats with lung function 7, 8, repeat genotypes were classified as having at least one long allele (≥33 repeats) or not. The allele distribution in the study population is presented in figure 2.
Distribution of (GT)n repeats of the HMOX1 promoter polymorphism.
Air pollution exposure
The attribution of PM10 exposure has been described in detail previously 2, 21. Each participant was assigned annual average PM10 concentrations for 1991 and 2002 based on their home address using a Gaussian dispersion model (PolluMap version 2.0 22) with predictions for the years 1990 and 2000, and an algorithm interpolating historical trends in central site measurements. Individual annual change in PM10 exposure, our exposure of interest, was calculated by subtracting 1991 values from those in 2002 and dividing the result by the number of years of follow-up. PM10 exposure in Switzerland declined throughout the study period, therefore the median change in PM10 exposure was negative.
Statistical analysis
A detailed characterisation of the study population can be found in online table O1. To address the influence of self-selected participation, our study sample was compared with all participants with complete data at the baseline examination, but not at follow-up.
The main effects of alleles on annual decline in FEF25–75% were assessed using multivariable linear regression with random effects for study areas under specification of additive, dominant and recessive genetic models (“additive” meaning a trend with the number of mutant alleles, “dominant” effects in heterozygous and homozygous and “recessive” only in homozygous mutant alleles). In accordance with Downs et al. 2, all models included the average annual PM10 exposure at baseline and its annual change, and adjusted for sex, age, age squared, height and atopy as baseline variables, smoking status (current smoker, former smoker, never-smoker) at follow-up, smoking intensity (number of cigarettes per day) at both surveys, smoking history in pack-years up to the baseline and between surveys as cumulative smoking exposure, parental smoking, workplace exposure to dust and fumes at each survey, body mass index (BMI) and its change (baseline BMI, change in BMI and an interaction term between the two), level of education and its change, nationality (Swiss and non-Swiss) and seasonal terms (sine and cosine functions of the day of examination).
Interaction between PM10 decline and genotypes was tested by recoding all polymorphisms into categorical variables representing dominant, co-dominant and recessive effects and including genetic variables and their products with change in PM10 into the regression models. Reparametrisation of annual PM10 decline into genotype-specific variables yielded genotype-specific PM10 effects and 95% confidence intervals. Reported effects refer to a 10 μg·m−3 decline in PM10 over the average follow-up period of 10.92 yrs. To control for the influence of FVC on the measurement of FEF25–75% 23, the analysis was repeated after adjusting FEF25–75% values at baseline and follow-up for FVC via linear regression and recalculating change in FEF25–75% using the residuals.
Separate analyses of main effects and interactions were done for males and females.
In a sensitivity analysis, the study sample was stratified into never- and ever-smokers (being a current smoker or ex-smoker at baseline or follow-up) to check whether the findings were stable across smoking categories. PM10 effects within genotype strata were tested for effect modification by ever-smoking.
To assess a potential multiple testing problem, we subjected all significant gene−air pollution interactions to permutation testing using STATA's “permute” procedure with 10,000 runs. This test permutes the outcome variable and rematches to the genotype data, creating new data sets where null associations are expected. The p-values for gene−air pollution interactions are computed as the fraction of permuted tests that were more significant than the original one.
Statistical analyses were performed using SAS software version 9.1 (SAS Institute, Cary, NC, USA) and STATA version 9.2 (StataCorp, College Station, TX, USA). Significance levels for two-sided tests of main effects were chosen at α = 0.05, and at α = 0.1 for tests of effect modification.
RESULTS
Characteristics of study participants
Characteristics of participants included in the present study are presented in table 1. At baseline, 52.9% of our study sample were female, the mean age was 41.4 yrs, and 29.1% of participants were smokers with a median of 18.4 pack-yrs smoked. Mean values for FVC at the baseline examination were 4.5 L, for FEV1 3.6 L and for FEF25–75% 3.4 L·s−1. A more detailed characterisation can be found in online table O1.
When compared with individuals who completed baseline assessment but were not included in this study for reasons described in figure 1 (see Methods section), our study sample was slightly older, leaner and showed a higher percentage of never-smokers, had better education and slightly higher lung function values.
Main effects of gene polymorphisms
Estimates of gene main effects on change in FEF25–75% during follow-up in the whole study sample are presented in table 2, sex-specific results are shown in online table O2.
HMOX1 polymorphisms
The homozygous mutant SNP alleles of all three HMOX1 SNPs and related haplotypes showed mostly favourable effects, attenuating the natural decline in FEF25–75% in the whole study sample by up to 5.9 mL·s−1·yr−1 (95% CI −0.5–12.3). Associations were statistically significant for HMOX1 rs5995098 (p = 0.047) and HMOX1 haplotype TTG (p = 0.031) under a recessive genetic model.
No significant association with FEF25–75% decline was found for having at least one long (GT)n promoter repeat allele.
GST polymorphisms
GSTP1 was associated with annual FEF25–75% decline (p = 0.044) under a dominant genetic model, and the effect was marginally significant under an additive model (p = 0.054): valine alleles accelerated decline by up to 4.4 mL·s−1 (table 2).
Interaction of genetic polymorphisms with reduction in PM10 exposure on lung function decline
Interactions of gene polymorphisms and decline in PM10 exposure in the whole study population on FEF25–75% decline are presented in table 3. Sex-specific results are presented in online tables O3 and O4. Analyses equivalent to the ones for FEF25–75% were done for change in FEV1 and FVC, but only weak interaction signals were found (online table O5).
HMOX1 polymorphisms
In the whole study sample, no significant interaction with decline in PM10 exposure was found for single HMOX1 SNPs, but rather for haplotypes: in participants not carrying haplotype ATC a PM10 decline of 10 μg·m−3 during follow-up was significantly associated with an attenuation of the annual decline in FEF25–75% by 15.3 mL·s−1 (95% CI 7.8–22.7), opposed to no significant effects in haplotype ATC carriers (table 3). The p-value for interaction was 0.009 when testing a dominant genetic model. There was also an interaction between decline in PM10 and having at least one long HMOX1 (GT)n promoter repeat allele: participants with long repeat alleles presented an attenuation of the annual FEF25–75% decline by 26.5 mL·s−1 (95% CI 11.7–41.2), as opposed to only 11.7 mL·s−1 (95% CI 4.3–19.2) in those without long repeat alleles. This interaction was statistically significant (p = 0.044).
GST polymorphisms
No significant interactions between GSTM1 or GSTT1 deletions and PM10 exposure reduction were found.
Regarding the GSTP1 p.Ile105Val polymorphism, participants with genotype Val/Val showed the most beneficial reaction to PM10 reduction with an attenuation of the decline in FEF25–75% by 27.3 mL·s−1 (95% CI 11.4–43.2). Those with one or no Val allele presented attenuations of 9.8 mL·s−1 (95% CI 0.7–19.0) and 13.5 mL·s−1 (95% CI 5.2–21.8), respectively. Under a recessive genetic model, this interaction was statistically significant (p = 0.052).
Reanalysis after adjustment for FVC values
Repeating the analysis using FEF25–75% adjusted for FVC to calculate change in FEF25–75% did not yield different results. Apart from small changes in effect size, the patterns of interaction persisted (table 3).
Sex-specific interactions between genetic polymorphisms and PM10 decline
The sex-specific results suggested that the interaction of PM10-decline with HMOX1 haplotype ATC was accentuated in males (fig. 3), while that with long HMOX1 (GT)n promoter repeat alleles was only seen in females. Genotype-specific PM10-effects are presented in online tables O3 and O4.
Estimated attenuation of the annual rate of decline in forced expiratory flow at 25–75% of forced vital capacity (FEF25–75%) associated with a 10 μg·m−3 decline in exposure to particulate matter with a 50% cut-off aerodynamic diameter of 10 μm (PM10) over 11 yrs of follow-up, in a) males and b) females, stratified by genotype. Positive values indicate attenuation of decline in FEF25–75%. #: pinteraction = 0.008 assuming dominant genetic effects; ¶: pinteraction = 0.007 assuming dominant genetic effects; +: pinteraction = 0.028; §: pinteraction = 0.052 assuming recessive genetic effects; ƒ: long allele is ≥33 GT promoter repeats.
Sensitivity analysis
Stratification and statistical interaction testing showed that the interaction effects between PM10 decline and HMOX1 and GST polymorphisms did not differ between never- and ever-smokers (data not shown).
Assessment of multiple testing
The adjusted p-values for the interaction between PM10 decline and HMOX1 haplotype ATC were p = 0.009 (dominant genetic model), p = 0.041 for a long HMOX1 (GT)n promoter repeat and p = 0.049 for GSTP1 Val/Val genotype (recessive genetic model).
DISCUSSION
The SAPALDIA study previously showed that an improvement in long-term PM10 exposure was associated with an important attenuation in age-related lung function decline 2.
In the current work we demonstrate that polymorphisms in GST and HMOX1 affect the age-related decline in FEF25–75% per se, but their impact is especially pronounced in modifying the effect of reduced PM10 exposure during the time of follow-up. Participants with mutant alleles in the HMOX1 SNPs or haplotypes, with long HMOX1 (GT)n promoter repeat alleles or with GSTP1 valine homozygosity, profited most from the improvement in air quality. For FEV1 decline, a similar but weaker signal was detected in GSTP1 valine homozygous participants. For HMOX1 haplotype ATC, effects on FEV1 and FVC decline appeared unstable, despite marginally significant interactions. The emphasised effect on FEF25–75% decline may be due to preferential deposition of the fine, more health-relevant portion of PM10 (PM2.5 and smaller) in smaller airways 24, and is not unexpected, as PM10 reduction affected FEF25–75% decline most strongly in our previous study 2.
The main effects and interactions with PM10 reduction differed by polymorphism in HMOX1. The GSTP1 SNP exhibited main and PM10 modifying effects. The GSTP1 main effect was most significant under a dominant (p = 0.044) and marginally significant under an additive genetic model, while the interaction was strongest under a recessive model (p = 0.052). This discrepancy is in line with findings from the few studies to date that have assessed genetic susceptibility to air pollutants, and all included GSTP1. The effect of the GSTP1 alleles was dependent on the chosen outcome and exposure: the homozygous GSTP1 isoleucine allele was associated with asthma in two childhood studies only in the presence of high urban background air pollution or ozone exposure 25, 26. In contrast, GSTP1 valine alleles were associated with respiratory symptoms in children highly exposed to ozone and with allergic sensitisation in adults exposed to traffic air pollution 27, 28. Experimental studies in ragweed-sensitised adults showed stronger inflammatory reactions upon allergen exposure for GSTM1-null or GSTP1 105Ile genotypes only in the co-presence of diesel exhaust or environmental tobacco smoke 29. These studies further confirm the importance of gene−environment interactions in the development of respiratory disease.
Our study benefits from a large population-based sample with detailed information on health parameters and validated exposure to air pollution (PM10) on an individual scale, as well as high-quality data on genetic variants. Limitations of our study include a possible selection of healthier individuals, as indicated by the figures in table 1. However, the mean baseline value of FEF25–75% did not differ between our study sample and the cohort subjects not included. Moreover, participation is unlikely to be influenced by the genetic profile of the individual, and any residual factors clustering at the area level are captured by random effects. With respect to the lower proportion of smokers in the study sample, our sensitivity analysis showed that our main findings were also present in participants with past or current smoking exposure. Our study determined change in lung function values using only two spirometry measurements, which is a limitation. However, we expect the measurement error to be randomly distributed across the investigated genetic variants. The observed associations are therefore rather underestimated.
Another concern is population stratification. No deviance from Hardy−Weinberg equilibrium was detected for bi-allelic polymorphisms in our study sample, and the proportion of HMOX1 long promoter repeat alleles did not significantly differ across study and language areas. Lack of suitable genetic markers prevented the application of genomic control methods 30, but since genetic homogeneity of Caucasian Western-Central European populations has been described 31, we do not expect population stratification to invalidate our findings.
We have not comprehensively assessed the genetic variation in HMOX1, and the functional impact of the HMOX1 SNPs and corresponding haplotypes is still unknown. Thus, our results must be confirmed upon availability of functionally well-characterised HMOX1 genetic variants.
Finally, it was beyond the scope of this current study to investigate how modifiable lifestyle factors like diet, physical activity or comorbidities determine susceptibility to air pollution. This will be an important aspect for future studies including our own.
In conclusion, our results show that a reduction in long-term exposure to PM10 does not exert uniform beneficial effects for all members of a population. They rather indicate that individuals who differ in coping with oxidative stress due to genetic variation in HMOX1 and GST enzymes also differ in their response to air quality improvements. Genetic susceptibility must be considered in setting limits for air pollution levels, which should protect the most susceptible members of society 32.
Acknowledgments
SAPALDIA team members are as follows.
Study directorate: T. Rochat (pneumology; Geneva, Switzerland), U. Ackermann-Liebrich (epidemiology; Basel, Switzerland), J.M. Gaspoz (cardiology; Geneva), P. Leuenberger (pneumology; Lausanne, Switzerland), L.J.S. Liu (exposure; Basel), N.M. Probst Hensch (epidemiology, and genetic and molecular biology; Zurich, Switzerland) and C. Schindler (statistics; Basel).
Scientific team: J.C. Barthélémy (cardiology; St Etienne, France), W. Berger (genetic and molecular biology; Schwerzenbach, Zurich), R. Bettschart (pneumology; Aarau, Switzerland), A. Bircher (allergology; Basel), G. Bolognini (pneumology; Lugano, Switzerland), O. Br�ndli (pneumology; Wald, Switzerland), M. Brutsche (pneumology; St Gallen, Switzerland), L. Burdet (pneumology; Payerne, Switzerland), M. Frey (pneumology; Aarau), M.W. Gerbase (pneumology; Geneva), D. Gold (pneumology, epidemiology and cardiology; Basel), W. Karrer (pneumology; Montana, Switzerland), R. Keller (pneumology; Aarau), B. Knöpfli (pneumology; Berne, Switzerland), N. Künzli (epidemiology and exposure; Basel), U. Neu (exposure; Berne), L.P. Nicod (pneumology; Lausanne), M. Pons (pneumology; Lugano), E.W. Russi (pneumology; Zurich, Switzerland), P. Schmid-Grendelmeyer (allergology; Zurich), J. Schwartz (epidemiology; Boston, MA, USA), P. Straehl (exposure; Berne), J.M. Tschopp (pneumology; Montana), A. von Eckardstein (clinical chemistry; Zurich), J.P. Zellweger (pneumology; Lausanne) and E. Zemp Stutz (epidemiology; Basel).
Scientific team at coordinating centres: P.O. Bridevaux (pneumology; Geneva), I. Curjuric (epidemiology; Zurich), J. Dratva (epidemiology; Basel), D. Felber Dietrich (cardiology; Basel), A. Gemperli (statistics; Basel), D. Keidel (statistics; Basel), M. Imboden (genetic and molecular biology; Zurich), H. Phuleria (exposure; Basel) and G.A. Thun (genetic and molecular biology; Zurich).
The study could not have been done without the help of the study participants, technical and administrative support and the medical teams and field workers at the local study sites.
Local fieldworkers: Aarau: M. Broglie, M. Bünter and D. Gashi; Basel: 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 (Basel), D. Baehler (Geneva), C. Gabriel (Geneva) and R. Nilly (Basel).
Footnotes
↵This article has supplementary material available fromwww.erj.ersjournals.com
Support Statement
Support was provided by the Swiss National Science Foundation (Berne, Switzerland) (grants 33CSCO-108796, 3247BO-104283, 3247BO-104288, 3247BO-104284, 32-65896.01, 32-59302.99, 32-52720.97, 32-4253.94, 4026-28099), the Federal Office for Forest, Environment and Landscape (Berne), the Federal Office of Public Health (Berne), the Federal Office of Roads and Transport (Berne), the canton governments of Aargau, Basel Stadt, Basel-Land, Geneva, Luzern, Ticino and Zurich, the Swiss Lung League (Berne), the canton Lung League of Basel Stadt/Basel Landschaft, Geneva, Ticino and Zurich, and UBS Wealth Foundation (Zurich, Switzerland) grant BA29s8Q7-DZZ.
Statement of Interest
A statement of interest for D.S. Postma can be found at www.erj.ersjournals.com/misc/statements.dtl
- Received March 18, 2009.
- Accepted August 24, 2009.
- ©ERS Journals Ltd
REFERENCES
