INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The identification of asthma patients at risk for clinical instability or death remains an important challenge. Studies have associated fatal or near-fatal asthma and worsening of asthma with excessive use of inhaled ₂-agonists (1, 2). Furthermore, airways from fatal asthmatics demonstrate decreased functional responses to ₂-agonists (3, 4). The basis of these observations could be genetically dysfunctional ₂-adrenergic receptors (₂AR) in asthmatics who die of their disease (5). Similarly, we and others have hypothesized that asthma severity may be related to ₂AR genotype (6, 7).

Several polymorphisms of the ₂AR gene have been identified (8). In particular, the amino acid substitutions from Arg Gly at position 16, from Gln Glu at position 27, and from Thr Ile at position 164 are known to affect receptor function in vitro. The Gly16/Gln27 and Gly16/Glu27 variants show significantly more agonist-promoted receptor downregulation than cells expressing the Arg16/Gln27 variant in vitro (11, 12). Cells expressing the Ile164 polymorphism display significantly less coupling of the ₂AR receptor to adenylyl cyclase than wildtype cells (13).

Further support for a role for ₂AR variants in contributing to the asthmatic phenotype comes from a series of clinical studies of genotype-phenotype associations. The Gly16 allele was associated with both steroid-dependent (10) and nocturnal asthma (6), and the Gln27 allele was associated with enhanced airway hyperresponsiveness among asthmatics (7).

In this study we sought to determine whether these polymorphisms are risk factors for fatal and near-fatal asthma or for asthma severity. Results of the previous in vitro and in vivo studies led us to propose that asthmatics with the Gly16 or Gln27 allele may experience greater agonist-induced receptor downregulation in vivo. Similarly, individuals with the Ile164 mutation are more likely to have severe asthma due to decreased responsiveness to ₂-agonists. We therefore considered that the prevalences of the Gly16, Gln27, or Ile164 alleles might be greater among fatal and near-fatal asthmatics than among nonfatal asthmatics of the same age. We also hypothesized that these allele prevalences will be greater among asthmatics with moderately severe disease than among mild asthmatics and nonasthmatic control subjects. To test these hypotheses, we used polymerase chain reaction (PCR) methods to genotype the ₂AR variants in these subject groups. This is the largest collection of severe asthmatics yet genotyped for these polymorphisms, the first study to examine their prevalence in fatal and near-fatal asthma, and the first to compare ₂AR allele frequencies between different ethnic backgrounds.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Study Subjects

Ethics approval was obtained from the University of British Columbia institutional review board. All subjects were classified as follows: Nonasthmatics had no personal history of asthma or atopy and had negative skin tests (< 1 mm wheal greater than saline) to common aeroallergens (from cat, dog, house dust mite, grass, molds, trees). Fatal asthma cases were younger than age 50 and had a history of asthma with recent worsening of symptoms leading to the fatal asthmatic episode. In addition, classical asthmatic histopathology was confirmed on microscopic study of the lung by two pathologists. Near-fatal asthmatics had a history of asthma and were intubated in a hospital emergency room (n = 22) or had hypercapnic respiratory failure (Pa_CO2 > 45 mm Hg) during an acute asthmatic episode (n = 1). All of the near-fatal asthmatics had reversible airflow obstruction (14). Nonfatal asthmatics were defined as asthmatics who were not taking oral steroids. This group of nonfatal asthmatics was further subdivided into those with mild asthma (taking  400 µg of inhaled beclomethasone or equivalent doses of other inhaled steroid preparations per day and/or having an FEV₁ > 75% of predicted) and those with moderate asthma (taking > 400 µg of inhaled beclomethasone or equivalent per day and/or having an FEV₁ < 75% of predicted). This subdivision of nonfatal asthmatics was done without prior knowledge of the subjects' genotypes. Medication history and lung function data were not available for the fatal asthmatics. All of the above subjects were Caucasian. In addition, 15 Asian (10 Chinese, 4 East Indian, and 1 Polynesian) asthmatics and 62 Black (African-American) nonasthmatics were genotyped to assess the allele frequencies of these polymorphisms in different ethnic groups.

Genotyping Methods

All samples were genotyped using DNA extracted from blood (15) except for samples from the fatal asthmatics, whose DNA was extracted from frozen lung (15) or paraffin-embedded tissue (16). Genotyping of the polymorphism at position 164 was performed as previously described (17). Genotyping of the polymorphisms at positions 16 and 27 was performed by one of two PCR-based methods: dot blot analysis or restriction enzyme digestion.

Dot blot analysis: Each 50 µl PCR reaction contained 100 ng DNA, 20 mM TRIS buffer (pH 8.4), 50 mM KCl, 1 mM MgCl₂, 200 µM deoxyribonucleoside triphosphates (dNTPs) and 0.5 µM each primer (5'-CTTCTTGCTGGCACCCAAT-3' and 5'-GTGATGAAGTAGTTGGTGAC-3'). The reaction conditions were 4 min 94° C, hold at 80° C for the addition of 1 unit of Taq DNA polymerase, and proceed with 35 cycles of 94° C for 30 s, 54 ° C for 30 s, and 72° C for 60 s. The 154 base pair (bp) PCR product contained the polymorphic sites at both nucleotide 46 (encoding amino acid 16) and nucleotide 79 (encoding amino acid 27). The PCR product was dot blotted equally onto four nylon membranes as described by the manufacturer, Amersham (Oakville, ON, Canada). Membranes were hybridized overnight at 55° C and the signal was detected with allele-specific oligonucleotide probes end-labeled with digoxigenin as described by the manufacturer, Boehringer-Mannheim (Laval, PQ, Canada). Probe sequences were: Arg16 GCACCCAATGGAAGCCATG; Gly16 GCACCCAATAGAAGCCATG; Gln27 GTCACGCAGCAAAGGGACG and Glu27 GTCACGCAGGAAAGGGACG. Final washing conditions were 15 min in 1× standard sodium citrate (SSC)/0.1% sodium dodecyl sulfate (SDS) at 55° C for Arg16 and Gly16 and at 58° C for Gln27 and Glu27.

Restriction enzyme analysis. PCR conditions were modified from Martinez and coworkers (18). Briefly, 30 µl of a 60-µl PCR product was digested with NcoI at an introduced restriction site to determine the genotype encoding amino acid 16. Twenty-five microliters of the same PCR product was digested with BbvI to identify the presence or absence of a naturally occurring restriction site at the locus encoding amino acid 27.

Haplotype Analysis

Haplotype analysis of the Arg16Gly and Gln27Glu loci included only subjects who were homozygous for at least one locus to unequivocally identify pairs of alleles on the same chromosome.

Statistical Analysis

A power analysis indicated that our study design could detect a difference in allele frequencies of 0.2 with 84% power. The Gly16 and Gln27 allele frequencies and allelic associations were analyzed with the chi-squared test for 2 × 2 contingency tables. The Gly16/Glu27 and Arg16/Gln27 allelic associations were compared with a 2-sample test for equality of proportions. The Ile164 allele frequency was analyzed by Fisher exact test. The haplotype frequencies in mild and moderate asthmatics were analyzed with the chi-squared test for both 3 × 2 and 2 × 2 contingency tables. Mean FEV₁ and beclomethasone dose were compared by a Wilcoxon rank-sum test.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Association Studies

The prevalences of the alleles in all Caucasian subjects combined were 0.61 for Gly16 (n = 244), 0.57 for Gln27 (n = 239), and 0.01 for Ile164 (n = 151). The prevalences of the 16 and 27 alleles in the Caucasian asthmatics were significantly different from the Asian asthmatics (p = 0.02 for Gly16 and p = 0.01 for Gln27, Table 1). Similarly, the prevalence of the Gln27 allele was significantly lower in the Caucasian nonasthmatics than in the Black nonasthmatics (p = 0.02, Table 1). The allele frequencies were not different between Asian and Black subjects, despite their different phenotypes. Because of these differences in allele frequencies between ethnic groups, we limited the study of asthmatic subsets to the Caucasians in whom we had the largest number of fatal/near-fatal subjects.

The frequencies of the Gly16, Gln27, and Ile164 alleles did not differ between fatal and near-fatal asthmatics compared with nonfatal asthmatics or with nonasthmatics (Table 2). The mean ages of these three groups were not different although the subgroup of near-fatal asthmatics was significantly older than the subgroup of fatal asthmatics (55.0 versus 25.2 yr, respectively). Both dot blot analysis and restriction enzyme digestion resulted in identical genotypes on samples which were genotyped by both methods (n = 18). In some cases, the DNA extracted from paraffin-embedded tissue did not yield a genotype for a given locus (n = 3 for Arg16Gly and n = 12 for Gln27Glu).

The subdivision of the 86 nonfatal asthmatics resulted in 33 subjects who could be classified as moderate asthmatics and 41 subjects who were mild asthmatics. Sixteen of the subjects without FEV₁ data were classified on the basis of drug dose alone and 11 subjects without accurate drug dosage data were classified on the basis of FEV₁ data alone. A further 12 subjects could not be classified. Of the 41 mild asthmatics with drug dosage data, all but nine were taking no inhaled corticosteroids. Contrary to one of our hypotheses, the Gly16 allele frequency was not different between these two subgroups. However, the Gln27 allele frequency was significantly higher in the subgroup that had moderately severe disease compared with the subgroup that had mild disease (p = 0.02, Table 3). Further examination of the data revealed that the Gly16/ Gln27 haplotype was also significantly more prevalent in the subgroup with moderately severe disease compared with the subgroup with mild disease (p = 0.01, Table 4). These data were analyzed based on a 3 × 2 contingency table combining the two Arg16 haplotypes because the number of subjects with Arg16/Glu27 was too small. The analysis indicated that the significant difference between the mild and moderate asthmatics was not due to the Arg16 haplotypes. Therefore, we excluded the Arg16 haplotypes and constructed a 2 × 2 contingency table, thus comparing the Gly16/Gln27 haplotype to only the Gly16/Glu27 haplotype. The result was a higher prevalence of the Gly16/Gln27 haplotype in the group with moderately severe asthma (p = 0.003). The odds ratio of having moderate rather than mild asthma was 3.1 for individuals with the Gly16/Gln27 haplotype versus the three other haplotypes combined (95% CI = 1.2, 8.0).

The Ile164 allele frequency was too low to analyze statistically in these groups. Of the 151 Caucasian subjects genotyped at this locus, only four subjects were heterozygous for the Ile164 allele (three nonasthmatics and one nonfatal asthmatic with moderately severe asthma); no homozygotes were found.

Allelic Association

Strong allelic association was observed between the Gly16 allele and the Glu27 allele and between the Arg16 allele and the Gln27 allele (p < 0.001). The Arg16/Gln27 haplotype occurred more frequently than the Gly16/Glu27 haplotype; 94% of Arg16 alleles associated with the Gln27 allele whereas 67% of Gly16 alleles associated with the Glu27 allele (p < 0.001).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The results of this study suggest that the presence of the Gly16, Gln27, and Ile164 alleles of the ₂AR is not a risk factor for fatal or near-fatal asthma in Caucasians. However, the study also revealed two other previously unknown findings. First, the prevalences of the Gln27 allele and the Gly16/Gln27 haplotype were significantly higher in the subgroup of nonfatal asthmatics who had moderate asthma than in those who had mild asthma. Second, the allele frequencies among Caucasians were significantly different from Asians and Blacks. Finally, the results also confirmed the previously observed allelic association between the Gly16 and Glu27 alleles and between the Arg16 and Gln27 alleles (18, 19). All of these findings have important implications for association studies of this gene.

The allele frequencies found in our Caucasian population were Gly16 = 0.61 and Gln27 = 0.57. These data are similar to allele frequencies reported by Martinez and coworkers in the largest study of unrelated Caucasians (Gly16 = 0.62, Gln27 = 0.61) (18). We are not aware of reports of these allele frequencies in Black or Asian populations.

The lack of association of ₂AR genotype with fatal and near-fatal asthma has several possible explanations. First, it is possible that the ₂AR genotype has no influence on receptor downregulation in vivo and thus it does not affect bronchodilation in asthma as we proposed. Second, it is possible that the ₂AR genotype does affect ₂-agonist-induced receptor downregulation in vivo and does influence the risk of fatal or near-fatal asthma but to a lesser extent than our study could detect. The minimum difference in allele frequencies we could detect with 80% power was 17%. However, we believe that a difference in allele frequencies less than this may yield a relative risk with limited predictive value. For example, we calculated that a difference in allele frequencies of 17% (e.g., for Gly16: 72% for fatal and 55% for nonfatal asthma) would give a relative risk of only 1.3. A third reason for the lack of association in our study could be that the ₂AR genotype does affect ₂-agonist-induced receptor downregulation in vivo and does contribute to the asthmatic phenotype but that this mechanism may not be an important contributor to the risk of a fatal or near-fatal episode. This explanation seems most likely given the association we find between ₂AR genotype and severity among less severe asthmatics and the results of previous studies. The factors that contribute to fatal or near-fatal asthma include the severity of airway inflammation and structural changes, the degree of exposure to allergen, the appropriateness of therapy, the compliance of the patient, access to the health care system, and socioeconomic status. Quite likely, the effects of the ₂AR genotype may be overshadowed by these other influences or by as yet unidentified genetic and environmental modifiers of severity.

The association of the Gln27 allele and the Gly16/Gln27 haplotype with asthma severity in nonfatal asthmatics but not with fatal/near fatal asthma is unexpected. While fatal asthma is a clearly defined endpoint, this definition may not represent a homogeneous phenotype. It is possible that some individuals within the fatal asthma group would be classified prospectively as mild despite eventually having a fatal or near-fatal event (20). It has been suggested (21) that there are at least two major forms of fatal asthma. Some individuals may have intrinsically severe disease and die after progressive deterioration and unrelenting airway obstruction ("slow onset") whereas others may be perceived to have mild-moderate disease but die relatively quickly during a single overwhelming episode ("sudden onset"). The analysis of subgroups of our fatal asthmatic subjects was not possible because medication histories, lung function data, and full details of the terminal events were not available for many subjects. Finally, it is possible that potentially fatal asthmatics with the Gly16/Gln27 haplotype may be less likely to have their symptoms controlled by ₂-agonists so that they are treated with higher doses of inhaled steroids that protect them from fatal exacerbations.

Fatal asthma has been strongly associated with excessive ₂-agonist usage, and airway tissues from fatal asthmatics show decreased response to ₂-agonists. However, it is not clear whether excessive ₂-agonist use has a detrimental effect on airway function or is simply a marker of severity. A detrimental effect could be related to ₂AR desensitization, and interindividual variation in ₂AR desensitization could be influenced by genotype. We therefore hypothesized that individuals with the genotypes that have been shown to downregulate excessively could be at increased risk for fatal or near-fatal asthma. It was our hope that by identifying genotypes associated with excessive downregulation of the ₂AR, individuals at increased risk for fatal asthma could be identified prospectively. If a strong association had been found, genotyping of asthmatic subjects to assess their risk for fatal or near-fatal episodes could be used clinically. Our results do not support this course of action.

The positive association between the Gln27 allele and moderately severe asthma supports previous work by Hall and coworkers (7) who found that the Gln27 allele was associated with enhanced bronchial hyperresponsiveness among asthmatics. These investigators have also shown an association of the Gln27 allele and elevated IgE levels in asthmatic families (19). Our findings that the Gln27 allele and particularly, the Gly16/Gln27 haplotype, are significantly more prevalent among moderately severe asthmatics than among mild asthmatics are consistent with the results of these two studies.

The observation of strong allelic association between Gly16 and Glu27 and between Arg16 and Gln27 has important implications. Take, for example, the hypothesis that severe asthmatics would have an increased prevalence of both the Gly16 and Gln27 alleles. We now know that it is unlikely that both alleles would be increased in the same group of individuals. In our study, because the Gln27 allele was associated with both asthma severity and the Arg16 allele, the Gly16 allele could not also be associated. It is possible that this is why we do not see an association between the Gly16 allele and asthma severity in our data set.

Because of the allelic association between these loci, it is difficult to draw conclusions from an allele frequency at a single locus. For this reason, we analyzed our data set by haplotypes. The data revealed that the Arg16/Glu27 haplotype is quite rare, accounting for only 3% of informative haplotypes. The data also showed that the Gly16/Gln27 haplotype was significantly more prevalent in the moderately severe asthmatics than in the mild asthmatics. Interestingly, this is the same haplotype that displayed the greatest receptor downregulation in vitro (9, 10). Our allelic association data from the entire Caucasian study population indicate that this is the recombinant haplotype because the Gly16 allele occurred 63% of the time with the Glu27 allele and only 37% of the time with the Gln27 allele. Nevertheless, analysis of the nonfatal asthmatics revealed that 20 of 28 Gly16/Gln27 haplotypes were from subjects who had moderate rather than mild asthma. The odds ratio of having moderate rather than mild asthma was 3.1 for individuals with the Gly16/Gln27 haplotype versus the three other haplotypes combined (95% CI = 1.2, 8.0).

The distinct differences in the allele frequencies of the ₂AR polymorphisms between different ethnic groups are significant for several reasons. First, these differences underscore the necessity of accounting for ethnic background in clinical studies of these ₂AR polymorphisms. Futhermore, one might hypothesize that these differences may contribute to differences in asthma severity or the response to ₂-agonists observed between different ethnic groups. For example, the Gln27 allele which is associated with greater bronchial hyperresponsiveness, higher IgE levels, and more severe asthma (this report) is more common in Blacks than in Caucasians, with the former group representing a portion of the U.S. population that is at greatest risk for asthma morbidity (22).

In summary, the results of this study have identified the Gln27 allele and the Gly16/Gln27 haplotype as risk factors for severity among asthmatics. However, neither the Gly16/Gln27 haplotype nor the Gly16, Gln27, and Ile164 polymorphisms were more prevalent in fatal and near fatal asthma, pointing toward additional factors that supersede ₂AR genotype in the expression of this extreme asthma phenotype. We also identified both allelic association and ethnic variability in allele frequencies as two potential confounders which must be considered in the design and interpretation of ₂AR association studies.