Effects of Glutathione-S-Transferase M1, T1, and P1 on Childhood Lung Function Growth

Transcription

1 Effects of Glutathione-S-Transferase M1, T1, and P1 on Childhood Lung Function Growth Frank D. Gilliland, W. James Gauderman, Hita Vora, Edward Rappaport, and Louis Dubeau Departments of Preventive Medicine and Pathology, University of Southern California Keck School of Medicine, Los Angeles, California The effects of glutathione-s-transferase (GST) M1, GSTT1, and all mortality emphasize the importance of attaining maximum GSTP1 genotypes on lung function growth were investigated in lung function (6, 7). 1,940 children enrolled in the Children s Health Study as fourth A broad spectrum of determinants for childhood lung graders (aged 8 11 years) in two cohorts during 1993 and 1996 function growth has been identified (1, 2, 8 11). Experiences and were followed annually over a 4-year period. Genotypes for during the in utero period affect lung function growth, as GSTM1 and GSTT1 and GSTP1 codon 105 variants (ile105 and val105) indicated by associations of adult lung function level with were determined using DNA from buccal cell specimens. We used exposure to maternal smoking, gestational age, and birth two-level regression models to estimate the effects of GSTM1, weight (8, 12 15). Early childhood respiratory infections, GSTT1, and GSTP1 genotypes on the adjusted annual average lung asthma, and airway hyper-responsiveness as well as childhood function growth. GSTM1 null was associated with deficits in annual exposures to air pollution and tobacco smoke are also associgrowth rates for FVC ( 0.21%; 95% confidence interval [CI], 0.40, ated with reduced lung function growth, lower adult lung 0.03) and FEV 1 ( 0.27%; 95% CI, 0.50, 0.04). Children who function, and increased risk for COPD. Familial aggregation were homozygous for the GSTP1 val105 allele had slower lung funcof FEV 1 and associations of COPD with candidate genes tion growth (FVC 0.35%; 95% CI, 0.62, 0.07; and FEV %; 95% CI, 0.68, 0.00) than children with one or more ile105 alleles. such as glutathione-s-transferase (GST) P1, epoxide hy- Children with asthma who were homozygous for the GSTP1 val105 drolase, and 1-antitrypsin gene indicate that a genetic variaallele had substantially larger deficits in FVC, FEV 1, and maximal tion is likely to affect childhood lung growth; however, ge- mid-expiratory flow than children without asthma. The deficits in netic determinants of childhood lung function growth have FVC and FEV 1 growth associated with both GSTM1 null and the GSTP1 not been extensively studied (16 22). Understanding the conval105 allele were largest and were statistically significant in nontype of lung function growth may clarify the pathophysiologic tribution of common genetic variants to the complex pheno- Hispanic white children. We conclude that GSTM1 and GSTP1 genopathways that are important for normal lung development types are associated with lung function growth in school children. and identify susceptible groups for interventions. Keywords: glutathione-s-transferase; FEV; children; polymorphism; We adopted a candidate gene approach to investigate loci growth with common functional allelic variants that may affect lung function growth (23). We identified candidate genes based Normal lung growth and development during childhood are on evidence that antioxidant defenses are important determiessential to reach maximum attainable adult lung function. nants of lung function. The growing lung is exposed to a Reduced growth in lung function leads to a lower attained substantial burden of endogenously produced and inhaled lung volume and air flow, which increase the adulthood risks oxidants and pro-oxidants, including O 2 and a wide variety for acute symptoms from exacerbations of asthma or respira- of toxic aerosols (24 26). If antioxidant defenses are inadetory infections, adverse effects from exposure to respiratory quate, substantial oxidative stress can occur that may intertoxins, and risks for chronic diseases such as chronic obstruc- fere with normal lung growth and may contribute to increased tive pulmonary disease (COPD) (1 5). Associations of low incidence, prevalence, and severity of respiratory diseases FEV 1 with increased risk of cardiovascular disease and over- such as COPD, asthma, and viral infections. Because oxidative stress is involved in the pathogenesis of conditions that affect adult lung function such as asthma and COPD, we hypothesized that polymorphic variants of genes functioning (Received in original form December 20, 2001; accepted in final form May 30, 2002) in antioxidant pathways that modulate oxidative stress are Supported by the National Heart, Lung and Blood Institute (Grant 5 R01 determinants of lung function growth. We focused this study HL61768), the California Air Resources Board (Contract A ), the National on GSTM1, GSTT1, and GSTP1 genotypes because these Institute of Environmental Health Sciences (Grants 1P50 ES P30 ES07048), genes are expressed in the lung, are involved in antioxidant the U.S. Environmental Protection Agency (Grant R ), and The Hastings defense pathways, and have common functional variant al- Foundation. leles (23, 27 29). DISCLAIMER: The statements and conclusions in this report are those of the GSTs (E.C ) are a superfamily of enzymes coninvestigators and not necessarily those of the California Air Resources Board, the sisting of,,, and families in humans (28, 30). Members U.S. Environmental Protection Agency, or the National Institute of Environmental Health Sciences. The mention of commercial products, their source or their use of the GST families have sequence similarity and shared in connection with material reported herein is not to be construed as either an catalytic properties for reaction of glutathione (GSH) with actual or implied endorsement of such products. reactive substrates. GSTs are well known as phase II xenobi- Correspondence and requests for reprints should be addressed to Frank Gilliland, otic detoxifying enzymes. One of more recently recognized Department of Preventive Medicine, University of Southern California Keck roles of GSTs is in oxidative defenses, where members of School of Medicine, 1540 Alcazar Street, CHP 236, Los Angeles, CA this family function as peroxidases to detoxify products of gillilan@usc.edu oxidative attack (27). Because antioxidants play a role in the This article has an online data supplement, which is accessible from this issue s pathobiology of a variety of diseases and variants in the GST table of contents online at superfamily are common, members of this superfamily may Am J Respir Crit Care Med Vol 166. pp , 2002 DOI: /rccm be determinants of respiratory health. Internet address: Several common variants of GSTs are well characterized

2 Gilliland, Gauderman, Vora, et al.: Effects on Childhood Lung Function Growth 711 METHODS Study Subjects The elements of the Children s Health Study have been described previously (32). Two cohorts of fourth-grade children were recruited from public school classrooms in 12 southern California communities that were selected based on historical measurements of air quality, demographic similarities, and a cooperative school district. The first cohort of 1,806 fourth-grade children was enrolled in 1993, and the second cohort of 2,081 children was enrolled in The parents or guardians of each participating student provided written informed con- sent and completed a written questionnaire, which provided demo- graphic information and characterized history of respiratory illness and its associated risk factors, exposure information, and household charac- teristics. We collected environmental tobacco smoke data using parentcompleted questionnaire items about household smokers, as previously described (33). Personal smoking habits of participants were assessed in a private interview at lung function testing sessions. Students in each community had lung function tests during visits to each school, as previously described (34). Maximum forced expiratory flow volume maneuvers were recorded using rolling-seal spirometers (Spiroflow; P.K. Morgan Ltd., Gillingham, UK). Each subject was asked to perform three satisfactory maneuvers based on the American Thoracic Society s recommendations modified for children (35). At the time of testing, students had height and weight measured and had a private interview about personal smoking habits and recent respiratory illness history and whether they had performed vigorous exercise within 30 minutes of the test. Genotype information was available from 1,940 of the 3,135 children (62%) enrolled in the two fourth-grade cohorts with complete covariate data from the questionnaire and at least two lung function tests. Laboratory Methods Buccal cells were collected from participants as a source of genomic DNA for genotyping assays. Details of buccal cell processing and geno- (28). GSTM1, a member of the family located on chromo- some 1p13.3, has two alleles: a wild-type allele GSTM1*1 and a variant allele that results in no protein expression GSTM1*0. Approximately 50% of Europid populations are homozygous for the GSTM1*0, and individuals with this null genotype express no GSTM1. GSTT1, a member of the family located on chromosome 22q11.2, has a similar null genotype; however, the frequency of the null genotype is approximately 25% in Europid, but it reaches 40% in some Asian populations. GSTP1 variants have been associated with a reduced risk for asthma and related phenotypes. GSTP1 has four alleles formed from two linked single nucleo- tide polymorphisms in codons 105 (exon 5) and 114 (exon 6) (wild-type P1*A [ile105, ala114], P1*B [val105, ala114], P1*C [val105, val114], and P1*D [ile105, val114]) (31). The ile105 wild-type allele has higher catalytic rates than the val105 variant for 1-chloro-2,4-dinitrobenzene but a lower efficiency for polycyclic aromatic hydrocarbon diol epoxides. Because the val105 and val114 are in highly significant linkage disequilibrium and the val114 does not appear to affect substantially the function of the 105 variants, we limited our genotyping the single base G to A change in codon 105 in this study. The Children s Health Study, a cohort study of children s respiratory health, offered an opportunity to investigate the effects of GSTM1, GSTT1, and GSTP1 genotypes on lung function growth in school children. We examined longitudinal lung function data collected over 4 years to determine whether genotypes were associated with annual average FEV 1, FVC, and maximal mid-expiratory flow (MMEF) growth among 1,940 school children who resided in 12 southern California communities. typing assays are provided in an online data supplement. Briefly, DNA was extracted using a PUREGENE DNA isolation kit (cat #D-5000; GENTRA, Minneapolis, MN). Genotypes for GSTM1 were determined using two methods. The first 380 samples were analyzed by enzymatic amplification of the polymorphic GSTM1 locus. Polymerase chain reac- tion products were visualized by electrophoresis on 2.5% agarose gel. The remaining GSTM1 samples and all of the GSTT1 and GSTP1 genotypes were determined using real-time polymerase chain reaction using a TaqMan 7,700 (Applied Biosystems, Foster City, CA). The presence or absence of a fluorescent amplification signal was used as an indication of whether the GSTM1 and GSTT1 alleles were present or absent in a particular genomic DNA sample. Samples showing no signal or late cycle number for start of amplification were repeated and further analyzed with primers and probes for the actin gene to verify the presence of amplifiable DNA. Analysis of the single nucleotide polymorphism at codon 105 in the GSTP1 gene was performed using allele-specific probes. Statistical Analysis We used a two-level regression modeling approach to investigate the relationship between genotypes and lung function growth. The first- level model was a mixed effects linear regression of log-transformed lung function measures on age to estimate the 4-year average growth slope for each subject. This model included adjustment for the following time-varying covariates selected on the basis of lung growth biology reported in the literature, published reports, and previous analyses of the Children s Health Study lung function data (11, 32): height, body mass index (linear and quadratic terms), yearly report of physician- diagnosed asthma (active asthma), personal smoking, respiratory symp- toms on the day of the test, exercise in the 30 minutes before the test, interaction of each of these factors with sex to allow for possible male/ female differences, and spirometer testing variables (barometric pressure, temperature, technician, and spirometer). The second level of the model was an inverse variance-weighted regression of the subjectspecific adjusted growth slopes (from the first model) on genotype. This model included adjustment for community of residence, cohort membership, sex, cohort-specific effects of ethnicity, environmental tobacco smoke exposure, and asthma status at study entry. The quantity of interest was the regression coefficient for genotype, which we report as the percentage of difference in annual lung function growth for subjects with one genotype relative to those with another genotype. For example, for GSTM1, we report the percentage difference in lung function growth for those with the GSTM1 null genotype compared with those with GSTM1 present genotypes. An analogous comparison is made for GSTT1, whereas for GSTP1, we present effect estimates under codominant, dominant, recessive, and additive genetic models for the GSTP1 val105 variant allele. In the additive model, genotype was coded as zero, one, or two for the number of val alleles. We considered a number of a priori potential modifiers of the genetic effects on lung function growth, including ethnicity, sex, environmental tobacco smoke exposure, family income, parental education, personal smoking, family history of asthma or atopy, asthma, and wheezing status. These variables were identified as potential modifiers of the associations of the genotypes with lung function based on consideration of factors that may contribute to genetic and environmental interactions. We tested the statistical significance of effect modifiers by compar- ing models with and without the appropriate interaction terms and report stratified results for variables that show statistical evidence for effect modification of the genetic associations (p 0.10). All analyses were conducted using the MIXED procedure in the SAS software and the MLn program (36, 37), and all reported p values are based on a two-sided alternative hypothesis. RESULTS The group of 1,940 eligible Children s Health Study participants with genotype data consisted of an approximately equal number

3 712 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL TABLE 1. CHARACTERISTICS OF THE ELIGIBLE* CHS move. This loss to follow-up is unlikely to introduce bias into FOURTH-GRADE CHILDREN WITH AND WITHOUT GENOTYPE the study because moving for economic reasons is unlikely to Genotyped Not Genotyped be related to genotype. GSTM1 and GSTT1 null genotypes were present in 43.4% Variables Number Number and 22.8% of children, respectively. Fewer children were homo- (n 1,940) Percentage (n 1,195) Percentage zygous for the GSTP1 val105 variant allele (13.3%). The fre- Sex quency of genotypes and alleles for GSTP1, GSTM1, and GSTT1 Girls varied among the ethnic populations in our study (Table 2). Boys African Americans had the lowest frequency of the GSTM1 null Ethnicity genotype. A higher proportion of Asians had the GSTT1 null White 1, Hispanic genotype than in the other ethnic groups. The allele frequency Black for the val105 variant was highest in Hispanics (0.45) and lowest Asian in Asians (0.32). The alleles of GSTP1 were in Hardy-Weinberg Other equilibrium within each ethnic group. The frequencies in our Asthma study population are within the ranges reported in other study Ever diagnosed populations (38, 39). Never diagnosed 1, , Cohort We found that GSTM1 and GSTP1 genotypes were associ ated with statistically significant deficits in annual lung function 2 1, growth (Table 3). The GSTM1 null allele was associated with a ETS 0.21% annual deficit in FVC growth and a 0.27% annual deficit Present in FEV 1 growth. For GSTP1, the recessive model for the val105 Absent 1, Personal smoking allele indicates a 0.35% deficit per year in FVC growth and a Smokers % reduction in growth for FEV 1. The codominant model Nonsmokers 1, , was consistent with decreased growth in FVC and FEV 1 for the Mean (SD) homozygous val105 genotype. These findings are consistent with Height, m 1.4 (0.1) 1.4 (0.1) a decrease in lung volume growth. Weight, kg 35.9 (8.8) 36.9 (9.5) Children with asthma had larger deficits in lung function BMI, kg/m (3.4) 18.6 (3.6) Age, years 10.1 (0.5) 10.1 (0.6) growth than children without asthma (Table 4). The annual deficits in FEV 1 growth associated with the homozygous val105 Definition of abbreviations: BMI body mass index; CHS Children s Health genotype were statistically significantly larger among children Study; ETS environmental tobacco smoke. with asthma compared with those without asthma. We found no * At least two-lung function tests available during the 4 years of follow-up for statistically significant effects of genotypes at the three loci on 2 fourth-grade cohorts recruited in 1993 and 1996; distributions at study entry. Physician-diagnosed asthma at study entry. MMEF in children with or without asthma; however, children Child smoker at the end of 4 years of follow-up. with asthma and the homozygous val105 genotype had a deficit Mean and SD at study entry. of 1.4% compared with children with asthma and the ile105 genotypes that was larger than for children without asthma (interaction p value of 0.06). The annual deficits in lung function growth associated with of boys and girls (Table 1). The majority of participants were GSTM1 and GSTP1 genotypes were larger and statistically signon-hispanic white (60.5%) and Hispanic (27.1%). Physician- nificant when the analyses were restricted to non-hispanic white diagnosed asthma was reported for 15.1% of children. Children children but not when restricted to Hispanic children (Table 5). ranged in age from 8 to 11 years at study entry, with a mean Among non-hispanic whites, deficits in FEV 1 growth were age of 10.1 years. An average of 4.4 lung function tests was 0.39% for the GSTM1 null genotype and 0.56% for the GSTP1 recorded on each participant. Over the 4-year period of study, val105/val105 genotype. Deficits in FVC among non-hispanic participants FEV 1 and FVC both grew at an average rate of whites were also larger and statistically significant for the GSTM1 11.9% per year. Children in the Children s Health Study with null genotype and the GSTP1 val105/val105 genotype. Among and without genotype data did not differ substantially by age, Hispanic children, the deficits in FVC and FEV 1 associated with sex, height, and weight. Compared with children with genotyping GSTM1 null genotype and GSTP1 val105/val105 genotypes were data, children without genotyping data participated more fre- smaller and were not statistically significantly different from quently in the earlier cohort (52% versus 43%), had more envi- zero. We found little evidence for modification of the genotype ronmental and personal tobacco smoke exposure (24% versus associations sex, family income, environmental tobacco smoke 16%), had slightly less asthma (13% versus 15%), and were exposure, family income, parental education, personal smoking, less likely to be non-hispanic white (51% versus 60%). The family history of asthma, or atopy. differences in these characteristics were related to differences Consideration of the joint effects of GSTM1 and GSTP1 in the socioeconomic status of the groups, as indicated by differ- indicates that the genes may be operating through independent ences in parental education and family income. In the group biologic pathways. In models among non-hispanic white chilwith genotype date, 26% of parents completed college and 27% dren, deficits in FVC and FEV 1 growth associated with GSTM1 had a high school education or less compared with 16% and null genotype and the GSTP1 val105/val105 genotype were ap- 40%, respectively, among the groups without genotyping. The proximately additive (Table 6). proportion of families with annual income that was less that $22,500 was higher in the group without genotyping (39%) than in the group with genotyping (26%). The loss to follow-up for DISCUSSION buccal cell sample collection was largely caused by families movthose We found that children who had the GSTM1 null genotype and ing out of the study area for economic reasons, as ascertained who were homozygous for the GSTP1 val105 variant allele by a questionnaire that asked about the reason for moving. The had lower lung function growth than children with the more families with lower education and income were more likely to common alleles at the two loci. The deficits in FVC and FEV 1

4 Gilliland, Gauderman, Vora, et al.: Effects on Childhood Lung Function Growth 713 TABLE 2. ALLELE AND GENOTYPE FREQUENCIES FOR GSTP1, GSTM1, AND GSTT1 AMONG PARTICIPANTS IN THE CHILDREN S HEALTH STUDY Ethnicity NHW Hispanics Blacks Asians Others Genotypes n % n % n % n % n % GSTP1 Ile/Ile Ile/Val Val/Val Observed allele frequency Ile Val GSTM1 Null GSTT1 Null Definition of abbreviations: GST glutathione-s-transferase; NHW non-hispanic whites. GSTP1 alleles are consistent with Hardy Weinberg Equilibrium in each group. growth were largest and statistically significant for children with asthma and non-hispanic white children. Although the magnitude of the cumulative effects on lung function over the period of study was relatively small, approximately 1%, a larger absolute deficit in maximum attained lung function may occur, and individuals with the variant genotypes may be susceptible to respiratory symptoms and conditions associated with low adult lung function, such as COPD. Furthermore, children with asthma had larger cumulative deficits, approximately 5%, which may compound deficits in lung function associated with asthma. We are not aware of other studies that have examined the associations between genetic variants and lung function growth in children. We observed effects of GSTM1 and GSTP1 genotypes on FVC and FEV 1 growth, but not MMEF. The similar magnitude of effects on FVC and FEV 1 growth suggests that GSTM1 and GSTP1 genotypes primarily affect growth in lung volume with little effect on measures of air flow. The explanation for the predominant effects on growth of lung volume is unclear. In adults, GSTM1 and GSTP1 variant alleles have been associated with COPD and asthma, suggesting a role for these loci in the pathogenesis of airflow limitation at older ages (40 43). It may be that the rapidly increasing lung volumes of children are more susceptible to reduced growth rates than are small airway flows such as MMEF. In addition, the measurement of MMEF has larger uncertainties than measurements of FVC and FEV 1, and the larger measurement error may result in artifactually reduced estimates for MMEF. No formal efforts were made to adjust tests of significance for multiple comparisons in these analyses. The level of significance and the consistency of the results for lung function measurements reduce the likelihood that the results were caused by chance alone. We found some indication of ethnic differences in the effects of GSTM1 and GSTP1 genotypes on FVC and FEV 1 growth. Lung function growth is a complex trait that probably involves multiple genes and pathways. The effects of variation at one TABLE 3. EFFECTS OF GSTM1, GSTT1, AND GSTP1 GENOTYPES ON ANNUAL LUNG FUNCTION GROWTH AMONG CHS PARTICIPANTS FVC FEV 1 MMEF Gene Genetic Model Percentage Deficit (95% CI) Percentage Deficit (95% CI) Percentage Deficit (95% CI) GSTM1 Present Null 0.21 ( 0.40, 0.03) 0.27 ( 0.50, 0.04) 0.30 ( 0.78, 0.19) Null 0.13 ( 0.36, 0.10) 0.09 ( 0.37, 0.20) 0.27 ( 0.32, 0.86) val/val 0.35 ( 0.62, 0.07) 0.34 ( 0.68, 0.00) 0.03 ( 0.74, 0.69) Dominant ile/ile val/val or ile/val 0.04 ( 0.16, 0.24) 0.05 ( 0.20, 0.29) 0.09 ( 0.41, 0.59) Codominant ile/ile ile/val 0.13 ( 0.08, 0.33) 0.13 ( 0.12, 0.39) 0.10 ( 0.42, 0.63) val/val 0.27 ( 0.58, 0.03) 0.27 ( 0.64, 0.11) 0.03 ( 0.74, 0.81) Additive 0.07 ( 0.21, 0.07) 0.06 ( 0.24, 0.11) 0.04 ( 0.32, 0.40) Definition of abbreviations: CHS Children s Health Study; CI confidence interval; GST glutathione-s-transferase; MMEF maximal mid-expiratory flow rate. p

5 714 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL TABLE 4. EFFECTS OF GSTM1, GSTT1, AND GSTP1 GENOTYPES ON ANNUAL LUNG FUNCTION GROWTH AMONG CHILDREN WITH AND WITHOUT PHYSICIAN-DIAGNOSED ASTHMA AT STUDY ENTRY Control Subjects Patients with Asthma (n 1,620) (n 287) p Value for Parameter Gene Genetic Model Percentage Deficit (95% CI) Percentage Deficit (95% CI) Interaction FVC GSTM1 Present Null 0.19 ( 0.40, 0.01) 0.52 ( 1.02, 0.01) 0.70 Null 0.12 ( 0.37, 0.12) 0.15 ( 0.83, 0.53) 0.70 val/val 0.27 ( 0.57, 0.04) 0.88 ( 1.60, 0.15) 0.20 FEV 1 GSTM1 Present Null 0.27 ( 0.52, 0.02) 0.49 ( 1.12, 0.14) 0.98 Null 0.07 ( 0.37, 0.24) 0.14 ( 0.97, 0.70) 0.51 val/val 0.18 ( 0.55, 0.20) 1.21 ( 2.10, 0.32) 0.04 MMEF GSTM1 Present Null 0.34 ( 0.84, 0.17) 0.16 ( 1.60, 1.30) 0.61 Null 0.33 ( 0.29, 0.95) 0.01 ( 1.91, 1.94) 0.45 val/val 0.31 ( 0.44, 1.08) 1.38 ( 3.43, 0.72) 0.06 Definition of abbreviations: CI confidence interval; GST glutathione-s-transferase; MMEF maximal mid-expiratory flow rate. Two-way p value for interaction between asthma at baseline and gene. p locus may depend on the genetic background of each ethnic group. Non-Hispanic white children may have a genetic background that allows a larger effect of GSTM1 and GSTP1 genotypes than Hispanic children. In addition, the associations may have been confounded by ethnic admixture. Lung function varies by ethnicity and genes such as GSTP1 have different frequencies in our Hispanics and non-hispanic white populations. Hispanics are likely to show stronger population stratification caused by TABLE 5. EFFECTS OF GSTM1, GSTT1, AND GSTP1 GENOTYPES ON ANNUAL LUNG FUNCTION GROWTH FOR HISPANIC AND NONHISPANIC WHITE CHILDREN Hispanics Non-Hispanic Whites (n 526) (n 1,173) p Value for Parameter Gene Genetic Model Percentage Deficit (95% CI) Percentage Deficit (95% CI) Interaction FVC GSTM1 Present Null 0.06 ( 0.45, 0.33) 0.28 ( 0.51, 0.05) 0.32 Null 0.09 ( 0.56, 0.38) 0.14 ( 0.42, 0.15) 0.81 val/val 0.23 ( 0.72, 0.26) 0.46 ( 0.83, 0.09) 0.35 FEV 1 GSTM1 Present Null 0.10 ( 0.60, 0.40) 0.39 ( 0.67, 0.11) 0.28 Null 0.02 ( 0.58, 0.63) 0.12 ( 0.47, 0.23) 0.67 val/val 0.01 ( 0.62, 0.64) 0.56 ( 1.01, 0.11) 0.09 Definition of abbreviations: CI confidence interval; GST glutathione-s-transferase. Two-way p value for interaction between race and gene. p 0.01.

6 Gilliland, Gauderman, Vora, et al.: Effects on Childhood Lung Function Growth 715 TABLE 6. ASSOCIATIONS OF GSTP1 AND GSTM1 GENOTYPES WITH LUNG FUNCTION GROWTH RATES IN NONHISPANIC WHITES GSTM1 FVC FEV 1 GSTP1 Present Null Present Null ile/ile or ile/val 0.24 ( 0.48, 0.01) 0.30 ( 0.60, 0.01) val/val 0.37 ( 0.89, 0.15) 0.76 ( 1.27, 0.24) 0.44 ( 1.07, 0.19) 0.95 ( 1.57, 0.33) Definition of abbreviations: GST glutathione-s-transferase. Two-way interaction between GSTP1 and GSTM1, p Two-way interaction between GSTP1 and GSTM1, p p of studies focusing on genes involved in oxidative stress pathways. The relatively modest effects of GSTM1 and GSTP1 vari- ants on lung function growth suggest that other genes may influence lung function growth (23). Investigations of the effects of additional loci on lung function growth are warranted. The larger effects of GSTP1 genotype on lung function growth among children with asthma are consistent with the proposed oxidative stress mechanism. A number of reports indicate that children with asthma have lower lung function growth than children without asthma (49). The reasons for the slower growth have yet to be clarified, but treatment with inhaled steroids does not appear to contribute to or reduce the deficits (50). An alternative explanation is that the excess oxidative stress in the airways of children with asthma may interfere with normal lung growth and development. Children with the val105 genotype may be less able to defend their airways from the adverse effects of excess oxidative stress associated with asthma. We conclude that school children with a GSTM1 null allele or a GSTP1 val105/val105 genotype show decreased growth in FVC and FEV 1. We consider lung function to be a quantitative trait that is determined by multiple genes and exposures. Each gene is likely to contribute modestly to variation in lung function growth. In this study, we show that the GSTM1 and GSTP1 may contribute to variation in growth and, based on lung physiology and the functional biology of the genes, implicate oxidative stress as a potentially important pathway for further investigation. Although the effects of GSTM1 and GSTP1 val105 allele are modest in magnitude for individuals, this genetic variation may have public health importance, especially for children with asthma. The variants are common, and small changes in the population distribution of lung function mean level can substantially increase the number of individuals with symptomatically low lung function. Children with these genotypes, especially those with asthma, may have lower attained lung function at maturity and be more susceptible to a spectrum of adverse respi- ratory outcomes associated with chronic excess oxidative stress. Further follow-up of the Children s Health Study cohort will be necessary to determine whether the reduction in lung function growth results in lower attained lung function. Investigation of the relationships between genes involved in oxidative stress path- ways and lung function growth is warranted. recent admixture that could result in biased estimates from uncontrolled confounding. We did not collect information to allow control for confounding from admixture within the ethnic groups. Because ethnic differences in lung function growth are small, we believe that the potential magnitude of a bias from confounding of the lung function growth results by admixture is small. The genes are on different chromosomes and thus are not in linkage disequilibrium. However, each locus could be in linkage disequilibrium with an unknown causal gene(s). The potential for linkage disequilibrium is a fundamental limitation of the candidate gene association approach and depends on the linkage disequilibrium surrounding each locus in the study populations. Different patterns of linkage disequilibrium could explain the ethnic differences. The sample of the study population with genotyping had a greater proportion of non-hispanic white children than the group without genotyping data (60% versus 50%). We adjusted for ethnicity and found no other substantial differences in characteristics of those with genotypes compared with those without genotypes, indicating that any bias from selection by availability of DNA is likely to be small. We had an insufficient number of children from African American, Asian, and other ethnic groups to allow assessment of the genetic effects within these groups. We identified three GSTs with common functional polymor- phisms as candidate genes based on a mechanistic hypothesis about pathways involved in lung function growth (23, 27 30, 44). A broad spectrum of determinants for reduced lung function in children and lung function decline and COPD in adults has been identified; however, a unifying mechanistic framework for the inter-related determinants has yet to be established (22, 25, 43, 45). An emerging body of evidence indicates that COPD, asthma, and respiratory infections increase pulmonary and systemic lev- els of oxidative stress (22, 25, 46 48). We hypothesized that excess oxidative stress provides a mechanistic framework that unifies inter-relationships between childhood lung function growth, asthma and respiratory infections, environmental exposures such as air pollution and tobacco smoke, and factors such as diet and genetics (23). Excess oxidative stress is the proximal event leading to inflammation, cell death, and subsequent airway remodeling among individuals with inadequate defenses. Acute events may lead to long-term effects in the setting of chronic excess oxidative stress. The measurement of chronic levels of oxidative stress is not currently feasible for large epidemiologic studies. Genetic variation may provide a useful method to classify chronic levels of oxidative stress. GSTM1 and GSTP1 are important antioxidant enzymes in the lung that function as anti- oxidants in xenobiotic, peroxide, and hydroperoxides metabolism pathways to reduce oxidative stress (29, 30). Our findings support the proposed mechanistic framework and the usefulness References 1. Burrows B, Taussig LM. As the twig is bent, the tree inclines (perhaps). Am Rev Respir Dis 1980;122: Dezateux C, Stocks J. Lung development and early origins of childhood respiratory illness. Br Med Bull 1997;53: Sherrill DL, Lebowitz MD, Knudson RJ, Burrows B. Smoking and symp- tom effects on the curves of lung function growth and decline. Am Rev Respir Dis 1991;144:17 22.

7 716 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL Tager IB, Weiss ST, Munoz A, Rosner B, Speizer FE. Longitudinal study ferase supergene family to resistance to oxidative stress. Free Radic of the effects of maternal smoking on pulmonary function in children. Res 1995;22: NEnglJMed1983;309: Hayes JD, Pulford DJ. The glutathione S-transferase supergene family: 5. Tager IB, Segal MR, Speizer FE, Weiss ST. The natural history of forced regulation of GST and the contribution of the isoenzymes to cancer expiratory volumes: effect of cigarette smoking and respiratory symp- chemoprotection and drug resistance. Crit Rev Biochem Mol Biol 1995; toms. Am Rev Respir Dis 1988;138: : Ebi-Kryston KL. Respiratory symptoms and pulmonary function as pretransferase 29. Strange RC, Spiteri MA, Ramachandran S, Fryer AA. Glutathione-Sdictors of 10-year mortality from respiratory disease, cardiovascular family of enzymes. Mutat Res 2001;482: disease, and all causes in the Whitehall study. J Clin Epidemiol 1988; 30. Hayes JD, Strange RC. Glutathione S-transferase polymorphisms and 41: their biological consequences. Pharmacology 2000;61: Hole DJ, Watt GC, Davey-Smith G, Hart CL, Gillis CR, Hawthorne 31. Hemmingsen A, Fryer AA, Hepple M, Strange RC, Spiteri MA. Simulta- VM. Impaired lung function and mortality risk in men and women: neous identification of GSTP1 Ile105 Val105 and Ala114 Val114 findings from the Renfrew and Paisley prospective population study. substitutions using an amplification refractory mutation system poly- Br Med J 1996;313: ; discussion merase chain reaction assay: studies in patients with asthma. Respir 8. Gilliland FD, Berhane K, McConnell R, Gauderman WJ, Vora H, Rappa- Res 2001;2: Peters JM, Avol E, Navidi W, London SJ, Gauderman WJ, Lurmann F, port EB, Avol E, Peters JM. Maternal smoking during pregnancy, Linn WS, Margolis H, Rappaport E, Gong H, et al. A study of twelve environmental tobacco smoke exposure and childhood lung function. southern California communities with differing levels and types of air Thorax 2000;55: pollution. I. Prevalence of respiratory morbidity. Am J Respir Crit 9. Gold DR, Tager IB, Weiss ST, Tosteson TD, Speizer FE. Acute lower Care Med 1999;159: respiratory illness in childhood as a predictor of lung function and 33. Gilliland FD, Berhane K, McConnell R, Gauderman WJ, Vora H, Rappachronic respiratory symptoms. Am Rev Respir Dis 1989;140: port E, Avol E, Peters JM. Maternal smoking during pregnancy, envi- 10. Cook DG, Strachan DP, Carey IM. Parental smoking and spirometric ronmental tobacco smoke exposure and children lung function. Thorax indices in children. Thorax 1998;53: ;55: Gauderman WJ, McConnell R, Gilliland F, London S, Thomas D, Avol 34. Peters JM, Avol E, Gauderman WJ, Linn WS, Navidi W, London SJ, E, Vora H, Berhane K, Rappaport EB, Lurmann F, et al. Association Margolis H, Rappaport E, Vora H, Gong H Jr, Thomas DC. A study between air pollution and lung function growth in southern California of twelve southern California communities with differing levels and children. Am J Respir Crit Care Med 2000;162: types of air pollution. II. Effects on pulmonary function. Am J Respir 12. Shaheen SO, Barker DJ, Holgate ST. Do lower respiratory tract infec- Crit Care Med 1999;159: tions in early childhood cause chronic obstructive pulmonary disease? 35. American Thoracic Society. Standardization of spirometry 1987 update: Am J Respir Crit Care Med 1995;151: ; discussion statement of the American Thoracic Society. Am Rev Respir Dis 13. Shaheen SO, Barker DJ, Shiell AW, Crocker FJ, Wield GA, Holgate 1987;136: ST. The relationship between pneumonia in early childhood and imtute; 36. SAS Institute. SAS/STAT user s guide, version 8. Cary, NC: SAS Instipaired lung function in late adult life. Am J Respir Crit Care Med ;149: Woodhouse G, Rasbash J, Goldstein H, Yang J, Heworth J, Plenis I. 14. Barker DJ, Godfrey KM, Fall C, Osmond C, Winter PD, Shaheen SO. MLn users guide. London: University of London; Relation of birth weight and childhood respiratory infection to adult 38. Garte S, Gaspari L, Alexandrie AK, Ambrosone C, Autrup H, Autrup lung function and death from chronic obstructive airways disease. BMJ JL, Baranova H, Bathum L, Benhamou S, Boffetta P, et al. Metabolic gene polymorphism frequencies in control populations. Cancer Epide- 1991;303: miol Biomarkers Prev 2001;10: Wjst M, Popescu M, Trepka MJ, Heinrich J, Wichmann HE. Pulmonary 39. Geisler SA, Olshan AF. GSTM1, GSTT1, and the risk of squamous cell function in children with initial low birth weight. Pediatr Allergy Immucarcinoma of the head and neck: a mini-huge review. Am J Epidemiol nol 1998;9: ;154: Chen Y, Dosman JA, Rennie DC, Lockinger LA. Major genetic effects 40. Baranova H, Perriot J, Albuisson E, Ivaschenko T, Baranov VS, Hemery on airway-parenchymal dysanapsis of the lung: the Humboldt family B, Mouraire P, Riol N, Malet P. Peculiarities of the GSTM1 0/0 genostudy. Genet Epidemiol 1999;16: type in French heavy smokers with various types of chronic bronchitis. 17. Silverman EK. Genetics of chronic obstructive pulmonary disease. No- Hum Genet 1997;99: vartis Found Symp 2001;234: Harrison DJ, Cantlay AM, Rae F, Lamb D, Smith CA. Frequency of 18. Chen Y. Genetics and pulmonary medicine: 10: Genetic epidemiology glutathione S-transferase M1 deletion in smokers with emphysema of pulmonary function. Thorax 1999;54: and lung cancer. Hum Exp Toxicol 1997;16: Kauffmann F, Tager IB, Munoz A, Speizer FE. Familial factors related 42. Ishii T, Matsuse T, Teramoto S, Matsui H, Miyao M, Hosoi T, Takahashi to lung function in children aged 6 10 years: results from the PAARC H, Fukuchi Y, Ouchi Y. Glutathione S-transferase P1 (GSTP1) polyepidemiologic study. Am J Epidemiol 1989;129: morphism in patients with chronic obstructive pulmonary disease. Tho- 20. Sandford AJ, Chagani T, Weir TD, Connett JE, Anthonisen NR, Pare rax 1999;54: PD. Susceptibility genes for rapid decline of lung function in the lung 43. Sandford AJ, Pare PD. Genetic risk factors for chronic obstructive pulmohealth study. Am J Respir Crit Care Med 2001;163: nary disease. Clin Chest Med 2000;21: Ishii T, Matsuse T, Teramoto S, Matsui H, Miyao M, Hosoi T, Takahashi 44. Fryer AA, Bianco A, Hepple M, Jones PW, Strange RC, Spiteri MA. H, Fukuchi Y, Ouchi Y. Glutathione S-transferase P1 (GSTP1) polymarker for bronchial hyperresponsiveness and asthma. Am J Respir Polymorphism at the glutathione S-transferase GSTP1 locus: a new morphism in patients with chronic obstructive pulmonary disease. Thorax 1999;54: Crit Care Med 2000;161: Barnes PJ. Chronic obstructive pulmonary disease. N Engl J Med 2000; 22. Koyama H, Geddes DM. Genes, oxidative stress, and the risk of chronic 343: obstructive pulmonary disease. Thorax 1998;53:S10 S Peterhans E. Oxidants and antioxidants in viral diseases: disease mecha- 23. Gilliland FD, McConnell R, Peters JM, Gong H Jr. A theoretical basis for nisms and metabolic regulation. J Nutr 1997;127:962S 965S. investigating ambient air pollution and children s respiratory health. 47. Schwarz KB. Oxidative stress during viral infection: a review. Free Radic Environ Health Perspect 1999;107: Biol Med 1996;21: Forsberg L, de Faire U, Morgenstern R. Oxidative stress, human genetic 48. Greene LS. Asthma, oxidant stress, and diet. Nutrition 1999;15: variation, and disease. Arch Biochem Biophys 2001;389: Berhane K, McConnell R, Gilliland F, Islam T, James Gauderman W, 25. MacNee W. Oxidants/antioxidants and COPD. Chest 2000;117:303S Avol E, London SJ. Rappaport E, Margolis HG, Peters JM. Sex- 317S. specific effects of asthma on pulmonary function in children. Am J 26. Schunemann HJ, Muti P, Freudenheim JL, Armstrong D, Browne R, Respir Crit Care Med 2000;162: Klocke RA, Trevisan M. Oxidative stress and lung function. Am J 50. Long-term effects of budesonide or nedocromil in children with asthma: Epidemiol 1997;146: the Childhood Asthma Management Program Research Group. N 27. Hayes JD, Strange RC. Potential contribution of the glutathione S-trans- Engl J Med 2000;343: