The genes that contribute to the genetic susceptibility to chronic obstructive pulmonary disease (COPD) remain largely unknown. We hypothesized that widely divergent rates of decline in lung function in smokers would be a robust phenotype for detection of genes that contribute to COPD severity. We selected 283 rapid decliners ( Δ FEV1 = − 154 ± 3 ml/yr) and 308 nondecliners ( Δ FEV1 = + 15 ± 2 ml/yr) from among smokers followed for 5 yr in the NHLBI Lung Health Study. Rapid decline of FEV1 was associated with the MZ genotype of the α1-antitrypsin gene (odds ratio [OR] = 2.8, p = 0.03). This association was stronger for a combination of a family history of COPD with MZ (OR = 9.7, p = 0.03). These data suggest that the MZ genotype results in an increased rate of decline in lung function and interacts with other familial factors. Haplotype frequencies of the microsomal epoxide hydrolase (mEH) gene were significantly different between rapid decliners and nondecliners (p = 0.03). A combination of a family history of COPD with homozygosity for the His113/His139 mEH haplotype was also associated with rapid decline of lung function (OR = 4.9, p = 0.04). The α1-antitrypsin S and 3 ′ polymorphisms, vitamin D-binding protein isoforms, and tumor necrosis factor (TNF- α G-308A and TNF- β A252G) polymorphisms were not associated with rate of decline of lung function.
Chronic obstructive pulmonary disease (COPD) is a major cause of morbidity and mortality worldwide. In a recent review of the global burden of human illness COPD ranked twelfth as a cause of lost quantity and quality of life and was projected to rank fifth by the year 2020 (1). Cigarette smoking is the major risk factor, but only 10–20% of smokers develop symptomatic COPD (2) and differences in cigarette smoke exposure account for only ∼ 15% of the variation in lung function (3). Family (4) and twin studies (5) have shown that COPD is a complex genetic disease.
COPD is characterized by decreased expiratory flow rates, increased pulmonary resistance, hyperinflation, and abnormal gas exchange. The disease is caused by proteolytic destruction of the lung parenchyma resulting in emphysema and loss of lung elastic recoil, as well as by inflammatory narrowing of peripheral airways. Typically, both pathophysiological processes are found in individual patients with COPD although one may predominate.
Lung function, as measured by forced expiratory volume in 1 s (FEV1), normally increases to a maximal value at adulthood, remains stable for 10–15 yr, and then declines (6). COPD can develop because of a reduced maximal lung function, an earlier age of onset of decline, or an accelerated rate of decline. Although the latter is thought to be related to genetic susceptibility to cigarette smoke few genetic studies have examined this phenotype.
The Lung Health Study (LHS) was conducted by the National Heart, Lung, and Blood Institute and was designed to describe the natural history of cigarette-induced COPD (7). A total of 5,887 male and female smokers were recruited, aged 35 to 60 yr, who had spirometric signs of early COPD. During 5 yr of follow-up 3,216 subjects continued to smoke and from this group we selected the 283 with the fastest rate of decline of FEV1 and the 308 who had no decline. These subjects were genotyped for polymorphisms in the α1-antitrypsin, microsomal epoxide hydrolase, vitamin D-binding protein, and tumor necrosis factor genes.
Homozygosity for the α1-antitrypsin (α1-AT) Z allele is clearly a cause of accelerated decline in FEV1 due to reduced antiproteolysis in the lung parenchyma. However, it remains to be determined whether heterozygosity for the Z allele, the S deficiency allele (8), and the 3′ allele (9) is also a risk factor for rapid decline in lung function.
Microsomal epoxide hydrolase (mEH) is a xenobiotic- metabolizing enzyme expressed in bronchial epithelial cells that metabolizes highly reactive epoxides in cigarette smoke. Polymorphisms in exon 3 (Tyr113→ His) and exon 4 (His139→ Arg) have been correlated with levels of mEH activity (10). The slowest activity haplotype (His113/His139) was associated with COPD in a U.K. population (11) and with more severe COPD in a Japanese population (12). However, these results were not confirmed in a Korean population (13).
The vitamin D-binding protein (VDBP) has three common isoforms known as 1F, 1S, and 2. The 1F–1F genotype and the 2 allele have been associated with increased and decreased risk for COPD, respectively (14). These associations may be due to the effects of VDBP on neutrophil chemotaxis (15) and/or macrophage activation (16).
Polymorphisms in the tumor necrosis factor-α (TNF-α) and TNF-β (lymphotoxin) genes (TNF-α G-308A and TNF-β A252G) have been associated with increased in vitro production of TNF-α and TNF-β (17). An association between the TNF-α-308A allele and COPD was found in a Taiwanese (18) but not in a white population (19).
All of the study subjects were white and had continued to smoke during the period of the LHS. Spirometry was performed annually over a period of 5 yr as previously described (7). The lung function of the subjects was assessed as FEV1 %predicted, that is, FEV1 adjusted for age, height, and sex (20). The rapid decliners had a decrease in FEV1 ⩾ 3.0% predicted/yr and the nondecliners had an increase in FEV1 ⩾ 0.4%/yr.
The subjects were genotyped for the α1-AT Z and 3′ mutations by a mutiplex polymerase chain reaction (PCR). Primers for the Z allele were as described previously (21). Primers for the 3′ mutation were 5′-GAC ACA GCA GCC AGG AAG TCC-3′ and 5′-CTC TCA GGT CTG GTG TCA TCC-3′. Digestion with TaqI restriction enzyme produced an uncut 144-bp fragment from the Z allele, 123- and 21-bp fragments from the M allele, an uncut 373-bp fragment from the mutant 3′ allele, and 191- and 182-bp fragments from the wild-type 3′ allele.
Primers for the S allele were 5′-GAG GGG AAA CTA CAG CAC CTC G-3′ and 5′-TGT GGG CAG CTT CTT GGT CAC CCT CAG GT-3′. Digestion with TaqI produced an uncut 117-bp fragment from the S allele and 97- and 20-bp fragments from the M allele. Genotyping for the VDBP isoforms was as described previously (22).
The subjects were genotyped for the mEH mutations by a mutiplex PCR. Primers for the exon 3 mutation were 5′-CAG GTG GAG ATT CTC AAC AGG-3′ and 5′-CAC ATT GTG GAA GAA GGC TGT T-3′. Primers for the exon 4 mutation were as described previously (11). Digestion with RsaI produced an uncut 115-bp fragment from the exon 3 slow allele (His113), 94- and 21-bp fragments from the exon 3 wild-type allele (Tyr113), an uncut 210-bp fragment from the exon 4 wild-type allele (His139), and 164- and 46-bp fragments from the exon 4 fast allele (Arg139).
The subjects were genotyped for the TNF-α and TNF-β mutations by a mutiplex PCR. Primers for the TNF-α mutation were as previously described (23). Primers for the TNF-β mutation were 5′-TGG TGG GTT TGG TTT TGG T-3′ and 5′-AGA GAA GGG GAC AAG ATG CAG T-3′. Digestion with NcoI produced an uncut 143-bp fragment from the TNF-α mutant allele (TNF-α-308A), 123- and 20-bp fragments from the TNF-α wild-type allele (TNF-α-308G), an uncut 175-bp fragment from the TNF-β wild-type allele (TNF-β-252A), and 88- and 87-bp fragments from the TNF-β mutant allele (TNF-β-252G).
The products of each PCR reaction were resolved by electrophoresis on 2% agarose gels stained with ethidium bromide. Template-free controls and known genotype controls were included in each experiment. Genotypes were scored without knowledge of the phenotypes by two independent observers. The samples were regenotyped if there was any disagreement concerning the genotyping.
The frequencies of the alleles and genotypes between groups were initially compared by χ2 analyses for 2 × 2 contingency tables. Odds ratios and 95% confidence intervals were calculated as previously described (24). The associations were also analyzed by binary logistic regression to adjust for potential confounding factors. The outcome was a dichotomous variable, that is, rapid decliner or nondecliner. Potential confounding factors included in the analysis were smoking history (expressed as mean number of cigarettes/d over the course of the LHS), age, sex, initial level of lung function (prebronchodilator FEV1 %predicted), and responsiveness to methacholine. The latter variable was expressed as a two-point dose–response slope as previously described (25). Only significant main effects were included in the final models and these were age, smoking history, and methacholine responsiveness. No significant two-factor interactions were found in any of the models. All tests were performed using the JMP Statistics software package (SAS Institute Inc.).
Haplotype frequencies were estimated using the expectation-maximization (EM) algorithm (26), as haplotypes could not be discerned directly from double heterozygotes. The Arlequin software package was used to compute haplotype frequencies. Linkage disequilibrium (D) between pairs of loci was calculated and expressed as a percentage of its maximal value (D/Dmax) as described previously (28).
The characteristics of the study groups are shown in Table 1. All loci were in Hardy–Weinberg equilibrium except for the amino acid substitution at position 416 of VDBP (p = 0.03).
|Phenotype||Rapid Decliners||Nondecliners||p Value Fast versus Slow|
|Mean (± SE) age, yr||49.5 (± 0.4)||47.6 (± 0.4)||0.0005|
|Mean (± SE) pack years||42.9 (± 1.1)||38.3 (± 1.1)||0.003|
|Mean (± SE) cigarettes/d*||25.6 (± 0.6)||22.3 (± 0.6)||0.0002|
|Mean (± SE) methacholine response†||−23.2 (± 1.5)||−7.8 (± 1.4)||< 0.0001|
|Mean (± SE) FEV1, L‡||2.47 (± 0.03)||2.71 (± 0.03)||< 0.0001|
|Mean (± SE) FEV1, %predicted‡||72.7 (± 0.5)||75.7 (± 0.5)||< 0.0001|
|Mean (± SE) ΔFEV1, ml/yr||−153.8 (± 2.6)||14.8 (± 1.5)||—|
|Mean (± SE) ΔFEV1, %predicted/yr||−4.1 (± 0.06)||1.1 (± 0.04)||—|
The α1-AT MZ genotype was more prevalent in the rapid decliners than in the nondecliners (Table 2). In a previous study, MZ heterozygotes were shown to have an increased rate of decline of lung function only if they also had a family history of lung disease (asthma, bronchitis, or emphysema) (29). In our study, subjects with a combination of the MZ genotype and a family history of COPD were more prevalent among rapid decliners (9/283 = 3.3%) than nondecliners (1/ 308 = 0.3%) (adjusted OR = 9.7, 95% CI 1.7–184.8, p = 0.009). Family history of COPD was defined as a first-degree relative who had chronic bronchitis or emphysema. Family history alone was not associated with rate of decline of lung function (OR = 1.3, p = 0.16). The prevalences of the S and 3′ polymorphisms were not significantly different between the groups (Table 2).
|Mutation||Genotypes||Rapid Decliners||Nondecliners||Odds Ratio (95% CI)*||p Value Slow versus Fast*|
|Z||MM||265 (94%)||300 (97%)||2.8 (1.2, 7.3)||0.02|
|MZ||18 (6%)||8 (3%)|
|S||MM||260 (92%)||272 (88%)||0.7 (0.3, 1.2)||0.19|
|MS||23 (8%)||36 (12%)|
|3′||wt/wt||233 (82%)||259 (84%)||1.2 (0.8, 1.9)†||0.39†|
|wt/mut||49 (17%)||46 (15%)|
|mut/mut||1 (< 1%)||3 (1%)|
There was a significant difference in mEH haplotype frequencies between the rapid decliners and nondecliners (Table 3). In addition, homozygosity for the His113–His139 haplotype was significantly increased in the rapid decliners (19/282 = 6.7%) versus the nondecliners (9/307 = 2.9%, OR = 2.4, 95% CI 1.1–5.4, p = 0.03). Subjects with both His113–His139 homozygosity and a family history of COPD were more prevalent among rapid decliners (11/282 = 3.9%) than nondecliners (2/307 = 0.6%) (OR = 6.2, 95% CI 1.4–28.0, p = 0.007). However, only the combined phenotype of both His113–His139 homozygosity and a family history of COPD remained significant after adjustment for confounding variables (adjusted OR = 4.9, 95% CI 1.1–34.9, p = 0.04). There was no significant linkage disequilibrium between the two loci (D/Dmax = 1.4%, p = 0.33) despite the polymorphisms being only 6,775 bp apart.
|Haplotype||Enzyme Activity||Rapid Decliners||Nondecliners||p Value|
|Tyr113–His139||Wild type||312 (55%)||348 (57%)||0.03|
|Tyr113–Arg139||Fast||91 (16%)||91 (15%)|
|His113–Arg139||Slow||17 (3%)||39 (6%)|
|His113–His139||Very slow||144 (25%)||136 (22%)|
The prevalence of VDBP genotypes was not different between the fast decliners and nondecliners (Table 4). The different VDBP isoforms are caused by the Glu416→ Asp and Thr420→ Lys amino acid substitutions and there was strong linkage disequilibrium between the alleles (D/Dmax = 100%, p < 0.0001). The prevalence of TNF haplotypes was also not different between the fast decliners and nondecliners (Table 5). The TNF-α and TNF-β genes are located 2,718 bp apart and substantial linkage disequilibrium (D/Dmax = 97.5%, p < 0.0001) was observed between the alleles.
|Genotypes||Rapid Decliners||Nondecliners||p Value|
|1F–1F||3 (1%)||6 (2%)||0.26|
|1F–1S||51 (18%)||57 (19%)|
|1F–2||23 (8%)||18 (6%)|
|1S–1S||80 (29%)||85 (28%)|
|1S–2||93 (33%)||120 (39%)|
|2–2||29 (10%)||19 (6%)|
|Haplotype||Rapid Decliners||Nondecliners||p Value|
|TNF-α-308G–TNF-β-252A||376 (66.7%)||406 (66.1%)||0.83*|
|TNF-α308A–TNF-β-252A||1 (0.2%)||2 (0.4%)|
|TNF-α-308G–TNF-β-252G||83 (14.7%)||97 (15.9%)|
|TNF-α-308A–TNF-β-252G||104 (18.4%)||109 (17.7%)|
The LHS cohort provides a unique opportunity to test susceptibility genes for rapid decline in lung function. There are very few studies of genetic factors and the rate of decline of lung function in the literature. This is due to the difficulties and expense associated with prospective study designs. Cross-sectional studies have generally defined COPD in terms of symptoms together with a certain degree of airflow obstruction (e.g., FEV1 < 70% predicted). Although cross-sectional studies are easier to perform they have certain drawbacks. First, there could be several mechanisms that lead to a low level of lung function. A patient with COPD may have experienced an abnormally high rate of decline in lung function, may not have attained the normal maximal level of lung function, or the age of onset of decline may have been abnormally early. Genetic factors may affect only one of these mechanisms. Therefore, the use of rate of decline of lung function rather than COPD as a phenotype for genetic studies may reduce the phenotypic heterogeneity and increase the power of the study. Second, cross-sectional case–control studies are more susceptible to systematic ascertainment bias and therefore to type I error. Finally, study designs using individuals from each extreme of the distribution of the phenotype of interest often have more power than random samples, depending on the mode of inheritance (30).
To our knowledge only three genetic factors (α1-AT type, ABO blood type, and ABH secretor status) have been investigated using rate of decline of lung function as the outcome variable (29, 31). Beatty and coworkers found that non-Z α1-AT variants and the A blood type were associated with lower rates of decline of FEV1 in women (31).
In this study we have demonstrated that the α1-AT MZ genotype and the mEH His113/His139 haplotype were associated with increased rate of decline of lung function. Both of these associations were stronger when the subjects had a family history of COPD, suggesting an interaction with other familial, possibly genetic, risk factors. However, the nature of these other risk factors remains to be determined. Regression analysis was used to show that the α1-AT and mEH associations were independent of known risk factors such as smoking history. This analysis also showed that there was no significant interaction of the α1-AT and mEH polymorphisms on the rate of decline of lung function.
The results of previous studies of the α1-AT MZ genotype and rate of decline are contradictory. Although some investigators concluded that the MZ genotype is associated with an accelerated rate of decline of lung function (29), others found no association (31). In one study, there was evidence that MZ heterozygotes had a lower rate of decline in lung function (32). These inconsistencies may be due to different selection criteria for the study subjects, different lengths of follow-up, and inclusion of different proportions of nonsmokers.
The association of the mEH slow haplotype with rate of decline is consistent with a previous study in which the highest odds ratio for COPD (4.1) was found with this haplotype. The data in the present study are the first to suggest that variants in the mEH gene predispose to COPD via an accelerated decline in lung function. This association has good biological plausibility. There are in vitro (10) and in vivo (33) data to suggest that the polymorphisms at amino acid positions 113 and 139 influence the function of the mEH protein. Thus, an individual with the His113–His139 haplotype would have a slow activity enzyme and would detoxify epoxides in cigarette smoke less readily. The presence of epoxides in the lung for longer periods could lead to greater tissue damage and inflammation.
However, the influence of these polymorphisms on the mEH protein is controversial. The results of one study indicated that the mEH polymorphisms did not account for all the variability in enzyme activity (34). In another study, the authors concluded that none of the variation in mEH activity could be attributed to the two polymorphisms (35). It has been suggested that mEH disease associations are due to linkage disequilibrium between these polymorphisms and polymorphisms in the mEH gene regulatory regions (36). Several polymorphisms have been identified in the 5′ region of the mEH gene and some of these may affect the level of gene transcription (37).
There was no association of VDBP genotypes with rate of decline of lung function. The reason for the previous associations between VDBP polymorphisms and level of lung function is unknown. However, VDBP has effects on neutrophil chemotaxis and macrophage activation, two processes known to be important in the pathogenesis of COPD. In a previous study we were unable to show any difference in rate of neutrophil chemotaxis stimulated by the different VDBP isoforms (22). This suggests that the role of VDBP in macrophage activation may underlie the association found in this study. Interestingly, differences in the glycosylation of the VDBP isoforms suggest that less than 10% of the 2 isoform can be converted into a macrophage-activating factor (38). These data are consistent with a protective role for the VDBP 2 isoform.
We found no association of the TNF haplotypes with rate of decline of lung function. This result suggests that these polymorphisms in the TNF-α and TNF-β genes do not contribute to susceptibility to a rapid rate of decline of lung function. This conclusion does not support the findings of a previous study (18). The lack of association may reflect genetic heterogeneity in the pathogenesis of COPD. The presence of several risk alleles for rapid decline of lung function would make replicating linkage and association studies difficult. Different risk alleles may be important in different populations. Alternatively, the lack of association may be due to differences between populations in amount or type of exposure to environmental factors. These factors may be particularly pertinent in this case as our population and the population of Huang and coworkers (18) were from different ethnic groups and different countries.
In this study we have attempted to control for potential confounders that might influence the rate of decline of lung function (age, smoking history, etc.). Another potential confounder is that some of the study subjects used an inhaled anticholinergic bronchodilator in the LHS. The main objective of the LHS was to determine whether this treatment could slow the rate of decline of lung function. However, use of a bronchodilator did not influence the rate of decline and therefore was not included as a covariate in our analyses. Another potential confounding factor that affects genetic association studies is population stratification. Although we confined our analyses to the white individuals in the study, there could be genetically distinct subgroups in this population because the LHS was a multicenter study. The α1-AT Z polymorphisms and the two mEH polymorphisms were all in Hardy–Weinberg equilibrium. This provides some evidence that the associations of these polymorphisms with rate of decline were not confounded by population structure. However, this possibility cannot be completely excluded.
In this study we have analyzed genotypic data from four separate loci and therefore it might be argued that correction for multiple comparisons is appropriate. However, for each of the polymorphisms investigated we had an a priori hypothesis of association with rate of decline of lung function. This was based on the previous associations of the polymorphisms with COPD and with changes in gene expression. Therefore, we believe that adjustment for multiple comparisons would be inappropriate and we have simply presented the statistical tests that were done.
In summary, we have demonstrated that the α1-AT MZ genotype and mEH His113/His139 haplotype were associated with increased rate of decline of lung function. This association was stronger when the subjects also had a family history of COPD. There was no association of VDBP or TNF genotypes with rate of decline of lung function.
Supported by the Medical Research Council. A.J.S. is supported by a Parker B. Francis fellowship.
|1.||Murray CJL, Lopez ADEvidence-based health policy—lessons from the Global Burden of Disease Study. Science2741996740743|
|2.||Fletcher C, Peto R, Tinker C, Speizer FE. The natural history of chronic bronchitis: an eight-year study of chronic obstructive lung disease in working men in London. Oxford: Oxford University Press; 1976.|
|3.||Beck GJ, Doyle CA, Schachter ENSmoking and lung function. Am Rev Respir Dis1231981149155|
|4.||Givelber RJ, Couropmitree NN, Gottlieb DJ, Evans JC, Levy D, Myers RH, O'Connor GTSegregation analysis of pulmonary function among families in the Framingham Study. Am J Respir Crit Care Med157199814451451|
|5.||Redline S, Tishler PV, Lewitter FI, Tager IB, Munoz A, Speizer FEAssessment of genetic and nongenetic influences on pulmonary function: a twin study. Am Rev Respir Dis1351987217222|
|6.||Tager IB, Segal MR, Speizer FE, Weiss STThe natural history of forced expiratory volumes: effect of cigarette smoking and respiratory symptoms. Am Rev Respir Dis1381988837849|
|7.||Anthonisen NR, Connett JE, Kiley JP, Altose MD, Bailey WC, Buist AS, Conway WA, Enright PL, Kanner RE, O'Hara P, Owens GR, Scanlon PD, Tashkin DP, Wise RAEffects of smoking intervention and the use of an inhaled anticholinergic bronchodilator on the rate of decline of FEV1: The Lung Health Study. JAMA272199414971505|
|8.||Lieberman J, Winter B, Sastre Aα1-Antitrypsin Pi-types in 965 COPD patients. Chest891986370373|
|9.||Kalsheker NA, Watkins GL, Hill S, Morgan K, Stockley RA, Fick RBIndependent mutations in the flanking sequence of the α1-antitrypsin gene are associated with chronic obstructive airways disease. Dis Markers81990151157|
|10.||Hassett C, Aicher L, Sidhu JS, Omiecinski CJHuman microsomal epoxide hydrolase: genetic polymorphism and functional expression in vitro of amino acid variants. Hum Mol Genet31994421428|
|11.||Smith CA, Harrison DJAssociation between polymorphism in gene for microsomal epoxide hydrolase and susceptibility to emphysema. Lancet3501997630633|
|12.||Yoshikawa M, Hiyama K, Ishioka S, Maeda H, Maeda A, Yamakido MMicrosomal epoxide hydrolase genotypes and chronic obstructive pulmonary disease in Japanese. Int J Mol Med520004953|
|13.||Yim JJ, Park GY, Lee CT, Kim YM, Han SK, Shim YS, Yoo CGGenetic susceptibility to chronic obstructive pulmonary disease in Koreans: combined analysis of polymorphic genotypes for microsomal epoxide hydrolase and glutathione S-transferase M1 and T1. Thorax552000121125|
|14.||Horne SL, Cockcroft DW, Dosman JAPossible protective effect against chronic obstructive airways disease by the GC 2 allele. Hum Hered401990173176|
|15.||Kew RR, Fisher JA, Webster ROCo-chemotactic effect of Gc-globulin (vitamin D binding protein) for C5a: transient conversion into an active co-chemotaxin by neutrophils. J Immunol155199553695374|
|16.||Yamamoto N, Homma SVitamin D-binding protein (group-specific component) is a precursor for the macrophage-activating signal factor from lysophosphatidylcholine-treated lymphocytes. Proc Natl Acad Sci USA88199185398543|
|17.||Bouma G, Crusius JB, Oudkerk Pool M, Kolkman JJ, von Blomberg BM, Kostense PJ, Giphart MJ, Schreuder GM, Meuwissen SG, Pena ASSecretion of tumour necrosis factor α and lymphotoxin α in relation to polymorphisms in the TNF genes and HLA-DR alleles: relevance for inflammatory bowel disease. Scand J Immunol431996456463|
|18.||Huang SL, Su CH, Chang SCTumor necrosis factor-α gene polymorphism in chronic bronchitis. Am J Respir Crit Care Med156199714361439|
|19.||Higham MA, Pride NB, Alikhan A, Morrell NWTumour necrosis factor-alpha gene promoter polymorphism in chronic obstructive pulmonary disease. Eur Respir J152000281284|
|20.||Crapo RO, Morris AH, Gardner RMReference spirometric values using techniques and equipment that meet ATS recommendations. Am Rev Respir Dis1231981659664|
|21.||Sandford AJ, Chagani T, Spinelli JJ, Paré PDα1-Antitrypsin genotypes and the acute-phase response to open heart surgery. Am J Respir Crit Care Med159199916241628|
|22.||Schellenberg D, Paré PD, Weir TD, Spinelli JJ, Walker BA, Sandford AJVitamin D binding protein variants and the risk of COPD. Am J Respir Crit Care Med1571998957961|
|23.||Chagani T, Paré PD, Zhu S, Weir TD, Bai TR, Behbehani NA, Fitzgerald JM, Sandford AJPrevalence of tumor necrosis factor-α and angiotensin converting enzyme polymorphisms in mild/moderate and fatal/near-fatal asthma. Am J Respir Crit Care Med1601999278282|
|24.||Dawson-Saunders B, Trapp RG. Basic and clinical biostatistics, 2nd ed. Norwalk, CT: Appleton & Lange; 1994.|
|25.||O'Connor GT, Sparrow D, Weiss ST. A prospective longitudinal study of methacholine airway responsiveness as a predictor of pulmonary-function decline: the Normative Aging Study. Am J Respir Crit Care Med 1995;152;87–92.|
|26.||Sham P. Statistics in human genetics. London: Arnold; 1998.|
|27.||Schneider S, Roessli D, Excoffier L. Arlequin ver. 2.000: a software for population genetics data analysis. Geneva, Switzerland: Genetics and Biometry Laboratory, University of Geneva; 2000.|
|28.||Falconer DS, Mackay TFC. Introduction to quantitative genetics. Harlow: Longman Group Ltd; 1996.|
|29.||Madison R, Mittman C, Afifi AA, Zelman RRisk factors for obstructive lung disease. Am Rev Respir Dis1241981149153|
|30.||Allison DB, Heo M, Schork NJ, Wong SL, Elston RCExtreme selection strategies in gene mapping studies of oligogenic quantitative traits do not always increase power. Hum Hered48199897107|
|31.||Beaty TH, Menkes HA, Cohen BH, Newill CARisk factors associated with longitudinal change in pulmonary function. Am Rev Respir Dis1291984660667|
|32.||de Hamel FA, Carrell RWHeterozygous α1-antitrypsin deficiency: a longitudinal lung function study. NZ Med J941981407410|
|33.||Pastorelli R, Guanci M, Cerri A, Negri E, La Vecchia C, Fumagalli F, Mezzetti M, Cappelli R, Panigalli T, Fanelli R, Airoldi LImpact of inherited polymorphisms in glutathione S-transferase M1, microsomal epoxide hydrolase, cytochrome P450 enzymes on DNA, and blood protein adducts of benzo(a)pyrene-diolepoxide. Cancer Epidemiol Biomarkers Prevent71998703709|
|34.||Hassett C, Lin J, Carty CL, Laurenzana EM, Omiecinski CJHuman hepatic microsomal epoxide hydrolase: comparative analysis of polymorphic expression. Arch Biochem Biophys3371997275283|
|35.||Kitteringham NR, Davis C, Howard N, Pirmohamed M, Park BKInterindividual and interspecies variation in hepatic microsomal epoxide hydrolase activity: studies with cis-stilbene oxide, carbamazepine 10,11-epoxide and naphthalene. J Pharmacol Exp Ther278199610181027|
|36.||Pirmohamed M, Kitteringham NR, Park BKPolymorphism in gene for microsomal epoxide hydrolase and lung disease. Lancet350199715531554|
|37.||Raaka S, Hassett C, Omiencinski CJHuman microsomal epoxide hydrolase: 5′-flanking region genetic polymorphisms. Carcinogenesis191998387393|
|38.||Viau M, Constans J, Debray H, Montreuil JIsolation and characterization of the O-glycan chain of the human vitamin-D binding protein. Biochem Biophys Res Commun1171983324331|