β₂‐Adrenergic Receptor Promoter Haplotype Influences Spirometric Response During an Acute Asthma Exacerbation

Abstract

Genetic variants in the β₂‐adrenergic receptor (ADRB2) coding block have been associated with different parameters of asthma severity, but there is no consensus on which variants are most important. Our objective was to determine whether the genetic variants in the 5′‐ or 3′‐flanking regions of ADRB2 impact the response to therapy. DNA was obtained initially from 72 adults hospitalized for an asthma exacerbation. We sequenced a 5,000 bp region of the ADRB2 gene that spanned the flanking regions and identified 31 single nucleotide polymorphisms (SNPs). Nonresponders to asthma therapy were defined as patients whose forced expiratory volume in 1 second (FEV₁) worsened by >10% at 24 hours after admission. We then evaluated the relationship between the 19 common SNPs and response to asthma‐specific therapy during acute disease exacerbations. Our results showed a significant association between nonresponders and a haplotype of five promoter SNPs in a nearly complete linkage disequilibrium. An analysis of the promoter and coding block polymorphisms in an extended cohort of 99 patients confirmed that promoter haplotype was the genetic component most strongly associated with asthmatic nonresponders, which was statistically significant among whites (p < 0.05). An identification of this promoter haplotype may provide an alternate explanation for the variation in the asthma responses observed with ADRB2 coding block polymorphisms.

Introduction

A number of clinical studies over the past decade have examined whether genetic variation in the β₂‐adrenergic receptor (ADRB2) gene is associated with asthma. Two polymorphisms, corresponding to amino acid changes Arg16Gly and Gln27Glu, have been studied extensively because of their high prevalence in the population and because the original in vitro studies suggested that these polymorphisms had a functional significance. ¹ , ² , ³ No ADRB2 polymorphism has been found to occur more commonly in asthma patients than in normal subjects, and there has also been no consensus on which ADRB2 polymorphism influences the response to therapy. ³ , ⁴ , ⁵ , ⁶ , ⁷ , ⁸ , ⁹ , ¹⁰ , ¹¹ , ¹² , ¹³ , ¹⁴ Several recent studies from the Asthma Clinical Research Network report that homozygosity at the Arg16 allele is associated with a worsening lung function in patients who regularly use short‐or long‐acting β‐agonists. ¹⁵ , ¹⁶ , ¹⁷ However, other recent studies have failed to validate this association. ¹⁸ , ¹⁹ Two recent reviews of ADRB2 pharmacogenetics highlight the major challenges in the field, ²⁰ including the differences observed in African American asthma patients. ²¹

In contrast to the large number of studies looking at the ADRB2 coding block polymorphisms, there have been only a handful of studies exploring the relationship of polymorphisms in the regulatory regions of the ADRB2 gene with asthma responses. ²¹ , ²² , ²³ In total, 22 polymorphisms with a minor allele frequency >3% have been identified in the 3.47‐kB region 5′ to the coding block of the ADRB2 gene, and a tight linkage disequilibrium (LD) across this region has resulted in only four ADRB2 promoter haplotypes. Drysdale et al. observed an 8% to 20% variation in the in vivo response to β‐agonist administration based on the haplotype pair, ²² and Hawkins et al. performed an extensive analysis of these haplotypes and observed an association of the baseline FEV₁/forced vital capacity (FVC) ratio with specific haplotypes in African Americans. ²³ This group also identified a poly‐C‐repeat of variable length in the 3′‐untranslated (UTR) region of ADRB2 that is associated with the baseline FEV₁/FVC ratio. ²³ Although these results provide the baseline allele frequencies for whites and African Americans in asthmatic and nonasthmatic populations, there have been no studies of the functional consequences of these polymorphisms during an acute asthma exacerbation.

Our hypothesis was that the genetic variants in the regulatory regions of the ADRB2 gene could influence the response to asthma therapy during an asthma exacerbation. To test this hypothesis, we sequenced the flanking and coding regions of the ADRB2 gene in a cohort of adults who were hospitalized for an asthma exacerbation and characterized by the response to therapy, and then evaluated the relationship between the most common polymorphisms identified and the response to asthma‐specific therapy.

Methods

Study design

All patients of age 18 years and older who were hospitalized for an asthma exacerbation between December 1999 and December 2002, during weekdays at a single university teaching medical center were considered for enrolment. Review of hospital charts and prior medical records and consultation with the currently treating hospital physician(s) were used to confirm the diagnosis of asthma and exclude other diagnoses, such as chronic obstructive pulmonary disease. The Vanderbilt Institutional Review Board approved the study, and an informed consent was obtained for all patients.

Data collection

The subjects were enrolled following admission, and the demographic characteristics and quantitative information on asthma medication utilization were collected from each study participant. The daily evaluations included spirometry (including FEV₁) and quantified asthma medication utilization. Blood was drawn, and DNA was extracted for genetic analysis.

Phenotype determination

The response status to asthma‐specific medication was determined using change in FEV₁ at 24 hours after admission as the primary outcome variable. Nonresponders were defined as patients whose FEV₁ worsened by >10% in the first 24 hours after admission because we wanted to identify a cohort that had a physiologic worsening despite maximal asthma management.

Genotype determination

The DNA from the first 72 participants was sequenced using 15 sets of overlying primers spanning a 5,000 bp region of the ADRB2 gene from approximately 3,000 bp upstream and extending up to approximately 600 bp of the 3′‐UTR region. Additional detail is provided in the online supplementary material. In 27 additional patients, genotyping at five specific loci (−2387, −2274, −1343, +46, and +79) was determined using TaqMan SM(c) Genotyping Assays (Applied Biosystems, Foster City, CA, USA) with predetermined primers.

Statistical analysis

A pairwise LD was estimated for the SNPs in ADRB2 using the standardized summary statistics D′ and r ², ²⁴ calculated by the HaploView program (Whitehead Institute for Biomedical Research, Cambridge, MA, USA). The haplotype blocks were assigned to our unphased genotype data using the D′ confidence interval of Gabriel et al. ²⁵ We used the modifcation of Florez et al., ²⁶ where adj acent blocks with a multiallelic D′ > 0.90 were merged. The haplotype frequencies for each block were estimated using the expectation‐maximization (EM) algorithm, again using the HaploView program.

Age (continuous) was described as the mean and standard deviation and compared using Students 2‐sided t‐test. Categorical variables were presented as proportions and compared using Fisher's exact tests. All p values were derived with respect to the 2‐sided alternative hypotheses, and the Statistical Analysis Sofware (SAS)^tm 8.12 (Cary, NC, USA) was used for the statistical analysis. The allele frequencies of nonresponders and responders in each population were compared using the methods developed by Raymond and Rousset, ²⁷ as implemented in the sofware package — Tools for Population Genetic Analysis and RXC (TFPGA, http://bioweb.usu.edu/mpmbio/). The associations between haplotype and response status were assessed using the haplotype trend regression test, as implemented in the sofware package PowerMarker (http://statgen.ncsu.edu/powermarker/). ²⁸ A determination of the genetic association was made if the frequencies of either alleles or haplotypes differed significantly between nonresponders and responders. For all tests performed, a significance level of 5% was used for inference. All reported p values are unadjusted for multiple tests.

Results

The first 120 patients enrolled to date in this prospective cohort were studied; of these 120, 17 did not have blood samples available for DNA analysis and another 4 patients did not have FEV₁ measured at the baseline. The baseline demographics for the remaining 99 patients are listed in Table 1 . Twenty patients had a >10% worsening in FEV₁ in the 24 hours after admission and were defined in this study as nonresponders to asthma‐specific therapy, with the remaining 79 patients designated as responders to β‐agonist therapy. Among responders, 56% were black compared to 20% among nonresponders (p= 0.005). No significant differences in age, gender, education level, insurance level, and smoking status were observed between responders and nonresponders.

Table 1.

Demographic and clinical characteristics of patients with acute asthma exacerbations requiring hospitalization, defined as responders and nonresponders.

	All patients (n= 99)	Responders (n= 79)	Nonresponders (n= 20)	p value
Age in years (mean ± SD)	42.6 ± 12.3	42.0 ± 11.4	45.0 ± 15.3	0.33
	% (n)	% (n)	% (n)
Gender (%)
Male	20 (20)	20 (16)	20 (4)	1.00
Female	80 (79)	80 (63)	80 (16)
Race (%)
White	51 (51)	44 (35)	80 (16)	0.005
Black	49 (48)	56 (44)	20 (4)
Education
<High school	6 (6)	5 (4)	10 (2)	0.67
High school	45 (45)	47 (37)	40 (8)
College+	47 (47)	47 (37)	50 (10)
Insurance level
TennCare	45 (45)	49 (39)	30 (6)	0.08
Commerical	36 (36)	37 (29)	35 (7)
Other	18 (18)	14 (11)	35 (7)
Smoking status
Never smoked	51 (50)	51 (40)	50 (10)	1.00
Former smokers	15 (15)	14 (11)	20 (4)
Current smokers	34 (34)	35 (28)	30 (6)
Medication use prior to hospitalization
Inhaled corticosteroids	56 (55)	56 (44)	55 (11)	0.74
Oral corticosteroids	25 (25)	24 (19)	30 (6)	0.59
Long‐acting β‐agonists	44 (44)	44 (35)	45 (9)	0.88
Leukotriene modifiers	29 (29)	30 (24)	25 (5)	0.73

We reviewed the baseline clinical data to determine whether there were other clinical factors associated with those nonresponders who had a >10% worsening in FEV₁ in the 24 hours after admission. Despite the dramatic differences in FEV₁ 24 hours after admission between nonresponders and responders, as defined by our model, the baseline FEV₁ did not differ between these two groups ( Figure 1A ). β‐agonist use 3 months prior and in the immediate 24 hours prior to hospitalization or during hospitalization also did not differ among nonresponders and responders ( Figure 1B ). In addition, the use of other medications (inhaled corticosteroids, oral corticosteroids, long‐acting β‐agonists, and leukotriene modifiers) prior to hospitalization did not differ among nonresponders and responders ( Table 1 ), and >90% of patients in each group received systemic corticosteroids at admission (data not shown).

Figure 1.

(A) Forced expiratory volume in 1 second (FEV1) as percent predicted at admission and at 24 hours after admission in non‐responders (filled squares) and responders (open squares) to asthma‐specific therapy. (B) Median number of daily meter dose inhaler (MDI) actuations at 3 time points in non‐responders (filled squares) and responders (open squares): 3‐month period prior to admission, 24‐hour time period prior to admission, 24‐hour time period after admission.

DNA from the initial 72 subjects enrolled was used for a complete sequencing of the ADRB2 gene. We sequenced a 5,000 bp region of the ADRB2 gene from approximately 3,000 bp upstream, through the coding region, and extending up to approximately 600 bp through the 3′‐UTR region. Using this approach, 31 SNPs were identified ( Table 2 ). We then evaluated which haplotype combinations existed for the 19 SNPs with greater than 5% frequency. Among 2 ¹⁹ (524,888) possible combinations, only 31 haplotypes were identified (see Table S4 in online supplementary material), which suggests a strong LD (see Figure S1 in online supplementary material).

Table 2.

Single nucleotide polymorphisms identified in the ADRB2 gene of subjects with asthma.

Location	Nucleotide	SNPs	Allele frequency Caucasian Americans		Allele frequency African Americans
5′‐flanking	−2,633	C/T	0.47	0.53	0.62	0.38
	−2,547	G/T	0.99	0.01	1.00	0.00
	−2,387	C/T	0.51	0.49	0.40	0.60
	−2,274	C/T	0.53	0.47	0.40	0.60
	−2,047	T/C	1.00	0.00	0.98	0.02
	−1,818	A/T	0.70	0.30	0.66	0.34
	−1,531	C/T	0.81	0.19	0.78	0.22
	−1,429	T/A	0.78	0.22	0.72	0.28
	−1,343	A/G	0.51	0.49	0.38	0.62
	−1,283	G/T	1.00	0.00	0.98	0.02
	−1,261	A/G	0.99	0.01	1.00	0.00
	−1,023	A/G	0.50	0.50	0.39	0.61
	−722	C/T	0.98	0.02	1.00	0.00
	−654	G/A	0.74	0.26	0.73	0.27
	−468	C/G	0.46	0.54	0.72	0.28
	−406	C/T	1.00	0.00	0.95	0.05
	−367	C/T	0.49	0.51	0.20	0.80
	−262	G/A	1.00	0.00	0.90	0.10
5′‐UTR	−75	G/A	1.00	0.00	0.97	0.03
	−47	C/T	0.50	0.50	0.23	0.77
	−20	C/T	0.51	0.49	0.25	0.75
Coding block	46	G/A	0.73	0.27	0.52	0.48
	66	C/T	0.99	0.01	0.97	0.03
	79	G/C	0.50	0.50	0.23	0.77
	252	G/A	0.80	0.20	0.78	0.22
	523	C/A	0.87	0.13	0.72	0.28
	1,053	G/C	0.82	0.18	0.71	0.29
	1,237	G/A	0.72	0.28	0.59	0.41
3′‐UTR	1,268	C/G	1.00	0.00	0.92	0.08
	1,537	T/C	1.00	0.00	0.98	0.02
	1,626	C/T	0.99	0.01	1.00	0.00

We then evaluated the relationship between the 19 SNPs with greater than 5% prevalence and response to asthma‐specific therapy during acute disease exacerbations. No single haplotype was associated with nonresponders to asthma‐specific therapy (data not shown). However, as shown in Table 3 , the allele frequency of each of the five individual SNPs (−2633C, −2387T, −2274T, −1343G, and −1023G) was more prevalent in the nonresponders than in the responders (p < 0.05).

Table 3.

Allele frequencies of the 19 most common SNPs in nonresponders and responders among the initial 72 subjects with asthma hospitalizations enrolled in the cohort.

Locus	Allele frequency		p value
	Nonresponders	Responders
−2633C	0.75	0.51	0.035
−2387T	0.75	0.51	0.035
−2274T	0.77	0.50	0.014
−1818T	0.42	0.30	0.258
−1531T	0.29	0.19	0.311
−1429A	0.36	0.22	0.154
−1343C	0.75	0.51	0.035
−1023C	0.73	0.51	0.049
−654G	0.35	0.24	0.317
−468C	0.29	0.45	0.143
−367T	0.75	0.62	0.258
−47T	0.75	0.62	0.207
−20T	0.75	0.57	0.107
46G	0.39	0.36	0.831
79C	0.75	0.62	0.180
252G	0.29	0.19	0.310
523C	0.29	0.17	0.170
1053G	0.32	0.20	0.213
1237G	0.43	0.30	0.251

The haplotype comparison in this initial cohort of 72 patients was limited by the small number of samples in each racial group. We therefore genotyped the next 27 patients enrolled, at five specific loci of the ADRB2 gene: at −2387, −2274, and −1343 in the promoter region and at +46 and +79 because these coding block SNPs have been so extensively studied and linked to various asthma outcomes. The analysis of this expanded cohort of 99 patients confirmed our prior findings; the allele frequency of each of the ADRB2 promoter SNPs was increased in the nonresponders when compared to the responders ( Table 4 ). Because the allele frequency of the known ADRB2 SNPs differs by race, we segregated these results by race. The association between nonresponders and each ADRB2 promoter SNP was statistically significant only in whites. There was also a similar trend in blacks, but the small number of black nonresponders (n= 4) in this particular cohort limits any conclusions.

Table 4.

Allele frequencies of the promoter and coding block SNPs in the cohort of 99 asthmatics requiring hospitalization for an acute asthma exacerbation, segregated by race and spirometric responder status.

Locus	Overall		p value
	Nonresponders (20)	Responders (79)
−2387T	0.69	0.53	0.02
−2274T	0.74	0.53	0.003
−1343G	0.69	0.53	0.02
46G	0.62	0.63	0.77
79c	0.72	0.67	0.40

Locus	Black		p value
	Nonresponders (4)	Responders (44)
−2387T	0.75	0.51	0.20
−2274T	0.75	0.51	0.16
−1343G	0.75	0.52	0.22
46G	0.63	0.53	0.62
79c	1.00	0.78	0.14

Locus	White		p value
	Nonresponders (16)	Responders (35)
−2387T	0.68	0.54	0.04
−2274T	0.73	0.56	0.007
−1343G	0.68	0.55	0.03
46G	0.62	0.76	0.14
79C	0.65	0.53	0.09

The longstanding, ongoing interest in the ADRB2 coding block SNPs led us to examine the haplotypes of the promoter, the coding block, and a combination of the promoter and coding block ( Table 5 ). The racial distribution of the haplotype frequencies is similar to what has previously been reported. ²² , ²³ Using the haplotype trend regression, only the promoter haplotype was found to be associated with nonresponders to asthma therapy in whites, although the small number of black nonresponders limits any conclusions. Neither the coding block haplotype nor the combined haplotype was associated with a lack of response to asthma treatment. The LD mapping using the HaploView indicates that there are racial differences in LD among the five SNPs that comprise this extended haplotype ( Figure 2 ). There is a near‐complete LD of the promoter SNPs both in blacks ( Figure 2A ) and in whites ( Figure 2B ). However, there are significant racial differences in LD with respect to the coding block SNPs. While both the +46 and +79 SNPs are in near‐complete LD with each other and with the promoter SNPs in whites, only the +79 SNP is in linkage in blacks, and there is no apparent linkage with the +46 SNP (which encodes amino acid 16).

Table 5.

Haplotype frequencies of the promoter and coding block SNPs among 99 hospitalized asthmatics, segregated by race and spirometric responder status.

	Overall (48)	Black		p value (HTR vs. response)
		Nonresponders (4)	Responders (44)
Promoter haplotype
TTG	0.52	0.75	0.49	0.494
CCA	0.44	0.25	0.47
Coding block haplotype
Gly16Gln27	0.27	0.37	0.26	0.562
Gly16Glu27	0.20	0.00	0.22
Arg16Gln27	0.53	0.63	0.52
Combined haplotype
TTGGC	0.27	0.37	0.26	0.693
CCAGG	0.18	0.00	0.19
TTGAC	0.25	0.37	0.24
CCAAC	0.26	0.25	0.26

	Overall (51)	White		p value (HTR vs. response)
		Nonresponders (16)	Responders (35)
Promoter haplotype
TTG	0.52	0.68	0.44	0.048
CCA	0.45	0.29	0.53
Coding block haplotype
Gly16Gln27	0.23	0.26	0.23	0.394
Gly16Glu27	0.47	0.35	0.53
Arg16Gln27	0.29	0.38	0.24
Combined haplotype
TTGGc	0.23	0.26	0.21	0.176
CCAGG	0.44	0.29	0.51
TTGAC	0.28	0.38	0.24
CCAAC	0.00	0.00	0.00

Figure 2.

Linkage disequilibrium (LD) mapping using the HaploView program in Blacks (A) and in Whites (B) at 5 loci of the B2‐adrenergic receptor gene: −2387, −2274, −1343 (5 of the translation start site), and +46 and −79 (in the coding block).

Discussion

Using DNA from our initial cohort of 72 patients hospitalized for asthma, we identified 31 SNPs in the ADRB2 gene, which included a 5,000 bp region spanning the promoter, coding, and 3′‐flanking regions. We found a significant relationship between the five promoter region SNPs of the ADRB2 gene and lack of response to asthma‐specific therapy, defined as a >10% worsening in FEV₁ in the 24 hours after hospitalization. Given the racial differences in the allele frequencies of many ADRB2 SNPs, we expanded the analysis to the next group of enrolled subjects, a total of 99 patients hospitalized for asthma. An association between a novel promoter haplotype encompassing these SNPs and the lack of response to asthma treatment was statistically significant in whites and was also observed in blacks, although the small number of black nonresponders limited our conclusions.

Our identification of the novel promoter haplotype (TTG) associated with the lack of response to asthma medications may provide an explanation for the variation in the asthma responses associated with ADRB2 coding block polymorphisms. The Arg16Gln27 coding block haplotype is in linkage with the TTG promoter haplotype (TTGAC) and with the CCA promoter haplotype (CCAAC) in nearly equal numbers in blacks. However, the Arg16Gln27 coding block haplotype is in disequilibrium only with the TTG promoter haplotype (TTGAC) in whites, and the combined CCAAC haplotype is not observed in whites. The absence of this haplotype accounts for a near‐complete LD observed with the ADRB2 promoter SNPs and both coding block SNPs in whites ( Figure 2B ), whereas there is no LD observed with the ADRB2 promoter SNPs and the +46 SNP (which encodes for amino acid 16) in blacks ( Figure 2A ). Thus, if the promoter haplotype TTG is the primary determinant of a decreased responsiveness, the strength of the association between Arg16 and decreased responsiveness in any study may simply be a reflection of the number of patients who have the TTGAC haplotype.

The mechanistic basis by which one of the components of the promoter haplotype alters the responsiveness may relate to the differences in the regulatory elements that alter the basal ADRB2 expression. In the setting of an asthma exacerbation, an increased β‐agonist exposure may lead to an increased desensitization, which could be more pronounced in individuals with lower ADRB2 density at the baseline. Another possible explanation is that ADRB2 haplotype alters the ability of corticosteroids to upregulate ADRB2 levels. Thus, in the setting of an asthma exacerbation, corticosteroids might not reverse the desensitization associated with an increased β‐agonist use.

The strong LD identified in this study is not surprising given the relatively small piece of DNA studied since the likelihood of recombination events increases with DNA length. Of note, there were no SNPs in the immediate (600 bp) 3′‐UTR present in >5% prevalence. More recently, Hawkins et al. identified a poly‐C‐repeat (+1269) in the 3′‐UTR region of ADRB2 that influences the FEV₁/FVC ratio. ²³ The three most common extended haplotypes of the promoter region include the three most common coding block haplotypes (Arg16Gln27, Gly16Gln27, and Gly16Glu27); thus, prior studies that draw conclusions about specific coding block haplotypes might also be explained by the influence of the promoter region haplotypes with which they are in strong LD.

We recognize that the significance of our findings in this study is limited by our small sample size, although even with limited numbers, we were able to establish extended haplotypes linking novel promoter regions SNPs with the infamous coding block SNPs. We recognize that SNP discovery and study of even more distal regulatory regions of the ADRB2 gene may identify the primary regulatory element influencing the ADRB2 function. The poly‐C‐repeat in the 3′‐UTR could influence message stabilization, and thereby, also influence ADRB2 density, ²³ although this study did not measure the in vivo responses. In this cohort of patients hospitalized for an asthma exacerbation, we were not able to study the influence of β‐agonists alone as all patients received a standard therapy that included systemic corticosteroids. Thus, the mechanism by which the ADRB2 variants influence the clinical responses may be mediated by corticosteroids, or by the interplay of β‐agonists and corticosteroids, rather than β‐agonists alone.

Conclusion

In summary, a novel haplotype in the regulatory region of the ADRB2 gene is significantly associated with a decreased response to asthma‐specific therapy, particularly in whites. We speculate that this haplotype may alter ADRB2 promoter activity, and thereby, influence responses to asthma‐specific therapy. An identification of this promoter haplotype may provide an alternate explanation for the variation in the asthma responses observed with ADRB2 coding block polymorphisms.

Supporting information

Supporting info item