Abstract
The adrenergic receptor (adrenoceptor) family genes have been extensively studied as candidate genes in hypertension but the results of individual genetic association studies (GAS) are controversial and inconclusive. To clarify these data, a systematic assessment of GAS for adrenoceptor family genes in hypertension was conducted.
Data from 163 GAS involving 7 genes and 37 distinct genetic variants were analyzed and cataloged in CUMAGAS-HYPERT (Cumulative Meta-analysis of Genetic Association Studies–HYPERTension; a web-based information system, which allows the retrieval and synthesis of data from GAS in hypertension, available at http://biomath.med.uth.gr). Data from genome-wide association studies involving the adrenoceptor family genes were also systematically searched.
Individual GAS reported inconsistent associations and had limited power to detect modest genetic effects, with only 1.2% having power >80%. Thirteen variants were investigated by three or more studies and their results were subject to meta-analysis. In the main meta-analyses, significant results were shown for five variants (ADRB1 p.Arg389Gly, ADRB1 p.Ser49Gly, ADRB2 g.9368308A>G, ADRB3 p.Trp64Arg, and ADRA1A p.Cys347Arg) under the allelic contrast and/or the dominant model. Subgroup analyses by ethnicity and gender detected significant associations for three variants (ADRB1 p.Arg389Gly in east Asians, ADRB2 p.Gln27Glu in whites, and ADRB3 p.Trp64Arg in whites and in males). Heterogeneity ranged from none to high. No significant associations were recorded from genome-wide studies.
There is evidence to implicate adrenoceptor genes in hypertension, although future studies designed to investigate epistatic and gene–environment interactions would allow more solid conclusions to be drawn about the role of these genes in hypertension.
Essential hypertension is a major public health problem due to its high prevalence (20–30% of the adult population in western societies) and its causal relationship with cardiovascular morbidity and mortality. Hypertension is considered as a multifactorial disorder, the onset and severity of which are influenced by both genetic and environmental factors.1,2 Genetic association and linkage studies have been used to identify hypertension susceptibility genes, although with limited success.1,3 The adrenergic receptor (adrenoceptor) family genes have emerged as logical candidate genes for hypertension based on experimental evidence showing involvement of the sympathetic nervous system in hypertension and on positional cloning findings from genome-wide linkage studies.1,2,3,4,5
The adrenoceptors belong to the G-protein-coupled receptors superfamily, which are integral membrane proteins with seven transmembrane helices, responsible for the signal transduction of a variety of extracellular signals.6 Neuronally released and circulating catecholamines bind to adrenoceptors to stimulate the intracellular signal transduction cascade and finally exert their biologic effect. The adrenoceptor family is subclassified into α1-, α2-, or β-adrenoceptors, although each of these classes has multiple subtypes so that a total of nine subtypes have been characterized: α1A-, α1B-, α1D-, α2A-, α2B-, α2C-, β1-, β2-, and β3-adrenoceptors. Each of these adrenoceptor subtypes is coded by a separate gene (ADRA1A, ADRA1B, ADRA1D, ADRA2A, ADRA2B, ADRA2C, ADRB1, ADRB2, and ADRB3, respectively) and has a different tissue distribution and function.7,8,9
Overall, 2,250 genetic variants have been annotated10 to the adrenoceptor family genes (1,147 to ADRA1A, 286 to ADRA1B, 324 to ADRA1D, 49 to ADRA2A, 69 to ADRA2B, 34 to ADRA2C, 92 to ADRB1, 192 to ADRB2, and 57 to ADRB3). Of these variants, 37 have been investigated in association with hypertension. A presentation of the standardized nomenclature of these variants according to the dbSNP identifiers (rs-numbers)10 and the Human Genome Variation Society nomenclature recommendations,11 as well as evidence on functional relevance of these candidate variants12,13,14,15,16,17,18,19,20 is provided in Supplementary Table S1 online.
The results of candidate-gene studies of adrenoceptors and hypertension are controversial and inconclusive, possibly due to methodological limitations, including inadequate sample size, patient selection, ethnicity of the populations studied, and lack of adjustments for confounders.21 In order to explore the involvement of adrenoceptor family gene polymorphisms in hypertension susceptibility, we systematically searched for all available genetic association studies (GAS) of adrenoceptor family genes and hypertension and created the CUMAGAS-HYPERT (Cumulative Meta-analysis of Genetic Association Studies–HYPERTension) information system. In this project, we cataloged all retrieved articles and estimated the risk effects of all individually investigated variants. Finally, the available data were synthesized using meta-analytic techniques in order to increase the power for detecting significant results and to decrease the uncertainty of the estimated genetic risk effects.22
Methods
Selection of studies. A systematic search of the PubMed and HuGE PubLit databases from their inception (1,950 and 2,000, respectively) through June 2009 was conducted. The search criterion in the PubMed database included combination of the following terms: “adrenergic”, “adrenoreceptor”, “ADRA1A”, “ADRA1B”, “ADRA1D”, “ADRA2A”, “ADRA2B”, “ADRA2C”, “ADRB1”, “ADRB2”, “ADRB3”, “hypertension”, “hypertensive”, “blood pressure”, “gene”, “polymorphism”, “allele”, “variant”, “mutant”. Bibliographies in articles were searched for further references. The HuGE PubLit database23 was searched for the disease term “hypertension” and for the gene terms “ADRA1A”, “ADRA1B”, “ADRA1D”, “ADRA2A”, “ADRA2B”, “ADRA2C”, “ADRB1”, “ADRB2”, “ADRB3”.
The eligible studies fulfilled the following inclusion criteria: (i) providing cases with clinically diagnosed hypertension and controls free of hypertension, (ii) providing information on genotype frequency or risk estimates, (iii) using DNA-based analysis methods for genotyping, and (iv) including subjects who were human. Studies investigating progression, severity, phenotype modification, response to treatment, or survival were excluded from our study. Case reports, editorials, review articles, and non-English articles were also excluded. Finally, family-based studies were excluded because of different design settings. Abstracts of studies retrieved were independently read by two investigators (G.D.K., E.Z.) to assess their appropriateness for this study. The results were compared, and disagreements were resolved by consensus. Full-text articles of the studies were evaluated according to the inclusion criteria.
Additionally, the full-texts and supplementary materials of the published genome-wide association studies (GWAS) of hypertension in HuGE PubLit23 and the NHGRI Catalog of Published Genome-Wide Association Studies24 were screened for findings of variants annotated to the adrenoceptor family genes. Supplementary searches were also performed in the open-access database for GWAS.25 Then, we identified whether each of the variants tested in candidate-gene studies had been included or tagged by proxy variants26 in the genotyping platforms used in the GWAS for hypertension and any significant results were recorded.
Data abstraction. From each article, the following information was extracted (see Supplementary Table S1 online): first author, year of publication, ethnicity of the study population, study design, demographics, and number of cases and controls for each genotype. The frequencies of the alleles and the genotypic distributions were extracted or calculated for both the cases and the controls. The dbSNP identifiers (rs-numbers) and the Human Genome Variation Society nomenclatures (showing nucleotide base and amino acid changes) for all genetic variants were identified by extended searches of bioinformatics databases (see Supplementary Table S1 online).10,23,25,26
Data analysis and synthesis. Before meta-analysis, the risk effect of gene variants for the allele contrast and the dominant models were evaluated for each study separately. For a variant with two alleles (A* and a*), where one of which is thought to be associated with a disease (e.g., A*), the allele contrast compares the number of alleles A* with the number of alleles a*, whereas the dominant model combines the AA and Aa genotypes and compares AA + Aa with aa.27 All associations were indicated as odds ratios (ORs) with the corresponding 95% confidence intervals (CI).
When more than two studies investigated the same variant, a meta-analysis of the published results was conducted. In the meta-analysis, the heterogeneity between studies was tested using the Q-statistic and it was quantified with the I2 metric.22 The pooled OR was estimated using random effect (DerSimonian and Laird) model. Random effect modeling assumes a genuine diversity in the results of various studies, and it incorporates a between-study variance to the calculations. When there is lack of heterogeneity, the random effect model coincides with the fixed effect model. The differential magnitude of effect in large vs. small studies (of variants included in meta-analysis) was also checked using the test proposed by Harbord et al.,28 when the meta-analysis involved four or more studies. The meta-analysis consisted of the main (overall) analysis, which includes all available data, subgroup analyses by ethnicity and gender, and sensitivity analysis, which examines the effect of excluding specific studies.22
The distribution of each variant in the control group was tested for Hardy–Weinberg equilibrium. As lack of Hardy–Weinberg equilibrium indicates possible genotyping errors and/or population stratification, a sensitivity analysis was carried out for these studies.22 The power of each study for the allele contrast was calculated assuming an OR of 1.2 (modest effect), a significance level of 0.05, a 0.3 disease prevalence and a disease allele frequency equal to the one of the study population.29,30
Analyses were performed using CUMAGAS-HYPERT and Compaq Visual Fortran90 with the International Mathematics and Statistics Library. Power was calculated using CaTS Power Calculator for Genetic Studies (Center for Statistical Genetics, University of Michigan).
Information system. CUMAGAS-HYPERT is a web-based database and an information system for cumulative meta-analysis of GAS located at http://biomath.med.uth.gr.29,30 CUMAGAS-HYPERT performs meta-analysis for all genetic models (allele contrast, dominant, recessive, and codominant) and provides information on study design and gene polymorphisms characteristics. CUMAGAS-HYPERT has the capacity of continuous updating (we currently aim to update the system on an annual basis and to include all GAS in the field of hypertension), and authors of published studies have the privilege of entering their data into the system after a request.
Results
Eligible articles
The literature review identified 944 titles that met the search criteria. In total, 190 articles remained after abstract selection. Sixty nine articles that investigated the association between genetic variants from the adrenoceptor family genes and hypertension fulfilled the inclusion criteria.31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99Figure 1 presents a flowchart of retrieved studies and studies excluded, with specification of reasons. Overall, 7 genes and 37 distinct variants investigated in 163 gene–disease association studies were identified. The studies were published between 1992 and 2009. A list of all the details abstracted from these studies is provided in Supplementary Table S1 online.
Flow chart of studies retrieved and studies excluded, with specification of reasons.
Studies' characteristics
The characteristics of each study and the association results of variants are shown in Supplementary Table S1 online. Studies were conducted in various populations of different racial descent: 76 studies involved solely whites, 45 studies recruited east Asians, 25 studies involved blacks, and 15 studies were conducted in ethnically mixed populations. The distribution of genotypes in the control group departured from Hardy–Weinberg equilibrium in 13 studies, whereas there was not enough information to check for Hardy–Weinberg equilibrium in 77 studies. In 6.7% of the studies, the statistical power was >50%, and in 11.6% studies, the power ranged from 25 to 50%. Only 1.2% of studies had power >80%.
Meta-analysis results
In total, 13 variants were investigated in three or more studies and their results were subject to meta-analysis. Table 1 shows the meta-analysis results for the association between the different variants and the risk of developing hypertension.
Meta-analysis results, the odds ratios (ORs) with the corresponding 95% confidence intervals (CIs), the heterogeneity metrics (PQ, I2), and the differential magnitude of effect in large vs. small studies (PH) are shown for the allele contrast and the dominant model
Meta-analysis results, the odds ratios (ORs) with the corresponding 95% confidence intervals (CIs), the heterogeneity metrics (PQ, I2), and the differential magnitude of effect in large vs. small studies (PH) are shown for the allele contrast and the dominant model
In the main analyses, significant results were shown for the variants of the ADRB1 and ADRA1A genes under both the allelic contrast and dominant model, for the p.Trp64Arg variant of the ADRB3 gene under the dominant model and for the g.9368308A>G variant of the ADRB2 gene under the allelic contrast. However, in general, these results were based on a relative small number of studies (3–15) and therefore they should be interpreted with caution. The main analyses for the commonly investigated p.Arg16Gly and p.Gln27Glu variants of the ADRB2 gene (n = 27 and n = 22 studies, respectively) were negative.
The meta-analyses for the ADRB1 gene showed significant associations for both variants examined. Carriers of the Arg* allele of the p.Arg389Gly polymorphism had a 16% reduced risk of hypertension (dominant model OR = 0.84 (0.73–0.97)), an effect that was marginally significant in sensitivity analysis and showed significant heterogeneity between the studies (PQ = 0.06, I2 = 50%). Subgroup analysis by race detected a marginally significant effect only in east Asians (allelic contrast OR = 0.91 (0.83–0.99)). Regarding the p.Ser49Gly polymorphism, carriers of the Ser* allele had a 24% increased risk for hypertension. Heterogeneity was not significant (PQ = 0.91) for the dominant model and subgroup analyses by ethnicity did not detect any significant effects.
The meta-analysis for the ADRB3 p.Trp64Arg was significant only for the dominant model (OR = 1.36 (1.12–1.64)) and revealed significant heterogeneity (PQ = 0.01, I2 = 0.49). This association remained significant in sensitivity analysis and in subgroup analysis for whites and for males. The allele contrast comparison included a smaller number of studies (because of unavailability of data for the full genotypic distribution, given the low frequency of the variant allele in whites) and was not significant overall, with the exception of subgroup analysis in males (OR = 1.37 (1.14–1.63)).
The main analysis for the nonsynonymous variant ADRA1A p.Cys347Arg detected a significant protective effect for the Cys* allele under both the allelic contrast and dominant model, with no significant heterogeneity of results (PQ = 0.55). Finally, for the ADRB2 gene, the meta-analysis of three studies for the promoter polymorphism g.9368308A>G showed a marginally significant association for the allelic contrast (allelic contrast OR = 0.79 (0.63–0.99)).
Among variants with negative main analyses, a positive subgroup analysis by ethnicity was recorded only for the variant ADRB2 p.Gln27Glu in whites. The association was significant for the allelic contrast (OR = 0.92 (0.86–0.97)) and the dominant model (OR = 0.89 (0.82–0.98)), and the heterogeneity analysis was not significant (PQ = 0.82 and PQ = 0.93, respectively).
In the main meta-analyses, no differential magnitude of effect in large vs. small studies was detected for all variants examined (all PH > 0.05).
Genome-wide association studies
In the eight available GWAS that examined the phenotype hypertension (as a dichotomous trait),100,101,102,103,104,105,106,107 no gene from the adrenoceptor family showed association at a genome-wide level of significance (see Supplementary Table S2 online). Seven of the 13 variants examined in the meta-analyses are captured by commercial genotyping platforms (see Supplementary Table S3 online). A nominal association has been observed only for two proxies of the variant ADRB2 p.Gln27Glu,103 although the statistical signal was rather weak (p = 10−2).
Discussion
In this project, our primary scope was to synthesize the currently available data on the GAS of human adrenoceptor family genes in hypertension and assess comprehensively the involvement of these gene variants in the development of the disease. Data from 163 GAS described in 69 published articles were cataloged in a publicly available web-based database and information system called CUMAGAS-HYPERT (located at http://biomath.med.uth.gr). With the implementation of the CUMAGAS-HYPERT, summary effect estimates were calculated in the context of 13 meta-analyses for adrenoceptor genetic variants and risk of hypertension. The resulting evidence provided insights regarding the role of these candidate genes on hypertension susceptibility.
Most of the published GAS were underpowered in terms of detecting the minor contributing role of common alleles. As the most realistic genetic association between a polymorphic locus and a complex disease has been claimed to yield an OR between 1.1 and 1.5 (ref. 108), a sample size of 10,000 subjects would be needed to achieve a satisfactory power (>80%). Meta-analysis clearly has a role in offering an analysis with the potential for higher power by pooling the results of independent analyses.22
The overall meta-analyses showed a significant role for five variants in the ADRB1, ADRB2, ADRB3, and ADRA1A genes. However, these positive associations resulted from pooling a small number of studies and therefore these results must be interpreted with caution. The two associated variants of ADRB1 (p.Arg389Gly and p.Ser49Gly) are likely functional and in strong linkage disequilibrium between them, thus pinpointing a potential locus of causative, functional variation.15,16 The positive association for the ADRB3 p.Trp64Arg polymorphism is also supported by evidence from functional studies.19 Synthesized data from white populations show a significant association with hypertension whereas data from east Asian populations, which present a higher frequency of the variant allele, do not support an association.109 Of interest, the main analysis was negative for the most extensively studied polymorphisms (p.Arg16Gly and p.Gln27Glu) belonging to the ADRB2 gene, which according to a bioinformatics application110 ranks in the top positions of genes related to hypertension. The p.Gln27Glu variant showed significant association in whites and not in east Asians, supporting an ethnicity-specific effect.
Discrepancies in results of individual studies may stem from a series of methodological issues, present in the literature of GAS for hypertension, as previously described.21,111 Substantial variability in terms of study design, inclusion criteria, phenotypic definition and sample sizes was observed in our literature sample analyzed. Failure to account for haplotypic structure or ethnicity/gender-specific interactions between genetic polymorphisms and environmental factors may have also contributed to the pattern of results observed.21,22,111
The GWAS in the field have not highlighted a significant role for the adrenoceptor family genes. However, coverage of the 13 variants included in the meta-analyses may have been suboptimal in commercial genotyping platforms26 (see Supplementary Table S3 online) and the variants identified to date from the GWAS approach explain only a fraction of the disease heritability, thus not excluding a potential role for adrenoceptors or other genes. Furthermore, the main-effects analyses implemented by GWAS may have missed associations of multilocus contributions of genes involved in pathways with strong pathophysiological relevance to disease mechanisms.112 Although the respective methodologies for “pathway-based” analyses have to be refined and provide proofs-of-concept, the adrenoceptor family genes may represent high priority candidates for such analyses.113
By selecting the binary phenotype “hypertension”, the statistical power of our analyses may have been limited, given the exclusion of certain articles presenting associations with blood pressure and the fact that analysis of continuous variables can be more powerful. This selection of endpoint was made because we aimed to identify the role of adrenoceptor genes in determining a clinically relevant condition, as any identification of genetic susceptibility can have important public health implications. Additionally, an analysis of haplotypes instead of single-marker analysis of these genes could be more informative. However, such a meta-analysis of haplotypes was not feasible in this study because individual GAS have used different haplotypes in their analyses and there is no widely accepted methodology available for the synthesis of differentially defined haplotypes across studies.32,114,115
CUMAGAS-HYPERT represents an evidence-based approach combined with an electronic information system to search systematically, review, and synthesize the rapidly emerging body of genetic studies of adrenoceptors in hypertension. Available evidence is cataloged and where appropriate, synthesized with meta-analytic techniques, highlighting the strengths as well as the gaps of research in the field. CUMAGAS is already functioning for additional complex phenotypes29,30 and will be expanded to other pathways of genes investigated in hypertension aiming to include all GAS in the field. The system will incorporate the findings from emerging GWAS and will be updated as evidence accumulates on an annual basis.
In summary, there is evidence to implicate adrenoceptor family genes in hypertension, although future studies designed to investigate epistatic and gene–environment interactions would allow more solid conclusions to be drawn about the role of these genes in hypertension. The CUMAGAS-HYPERT information system may be a useful resource for reviewing and interpreting the findings of accumulating genomic epidemiology research in hypertension.
Acknowledgements
Scientific support for this project was provided through the Tufts Clinical and Translational Science Institute (Tufts CTSI) under funding from the National Institute of Health/National Center for Research Resources (UL1 RR025752). Points of view or opinions in this paper are those of the authors and do not necessarily represent the official position or policies of the Tufts CTSI.
Disclosure
G.D.K. is a Pfizer-Tufts Medical Center Post-Doctoral Fellow in Clinical Research. E.Z. declared no conflict of interest.
References
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
43.
44.
45.
46.
47.
48.
49.
50.
51.
52.
53.
54.
55.
56.
57.
58.
59.
60.
61.
62.
63.
64.
65.
66.
67.
68.
69.
70.
71.
72.
73.
74.
75.
76.
77.
78.
79.
80.
81.
82.
83.
84.
85.
86.
87.
88.
89.
90.
91.
92.
93.
94.
95.
96.
97.
98.
99.
100.
101.
102.
103.
104.
105.
106.
107.
108.
109.
110.
111.
112.
113.
114.
115.
