ADRB2 gene variants, DEXA body composition, and hypertension in Tobago men of African descent

Abstract

Objective

Classic tissue effects of β₂ adrenergic receptor activation include skeletal muscle glycogenolysis and vascular smooth muscle relaxation, factors relevant to obesity and hypertension, respectively. In a population-based study, we examined two common amino acid substitutions in the β₂ adrenergic receptor gene (ADRB2) in relation to body composition and blood pressure.

Design and subjects

Cross-sectional analysis of 1893 African-descent men living in Tobago and participating in a prostate cancer screening study.

Measurements

Body mass index (BMI), waist circumference, blood pressure, dual energy X-ray absorptiometry (DEXA) body composition, and ADRB2 (Arg16Gly; Gln27Glu) genotype.

Results

Twenty-six percent were obese (BMI ≥ 30 kg/m²) and 50% hypertensive. ADRB2 Arg16Gly and Gln27Glu alleles were in linkage disequilibrium (D′=0.96, r²=0.15). ADRB2 16Gly-containing and 27Glu-containing genotypes were equally frequent in low, medium, and high tertiles of percentage body fat mass (16Gly-containing genotypes: 73.4%, 74.4%, 74.5%, p_trend=0.66; 27Glu-containing genotypes: 27.6%, 23.8%, 25.4%, p_trend=0.39) and in normal blood pressure, pre-hypertensive, and hypertensive men (16Gly-containing genotypes: 73.4%, 72.8%, 74.4%, p_trend=0.61; 27Glu-containing genotypes: 25.6%, 24.1%, 26.7%, p_trend=0.50).

Conclusion

In a high obesity and high hypertension risk population with ancestry in common with African-Americans, genetic variation defined by two common ADRB2 amino acid substitutions was not associated with body composition or hypertension.

Introduction

Interacting with diet and physical activity, genetic factors contribute to obesity and hypertension.[1–3] Frequently studied candidate genes include the leptin,[4] catecholamine,[5] and peroxisome proliferator-activated receptors [6] and genes in the renin-angiotensin system.[7] Stimulating β adrenergic receptor function promotes lipolysis in fat cells.[2, 8] The β2 adrenergic receptor gene (ADRB2) attracts interest because catecholamine activation of the β2 adrenergic receptor regulates skeletal muscle glycogenolysis and vascular smooth muscle relaxation.[9, 10] The ADRB2 single nucleotide polymorphisms most often studied in human populations substitute glycine for arginine at codon 16 (Arg16Gly) and glutamic acid for glutamine at codon 27 (Gln27Glu).[11]

Despite numerous published studies, the connection, if any, between ADRB2 Arg16Gly or Gln27Glu genotype and metabolic phenotypes, including obesity and hypertension, is tentative. Results from early recombinant cell culture experiments,[12] for example, launched a long standing hypothesis asserting that the ADRB2 16Gly variant promoted obesity and hypertension by causing receptor down regulation and physiologic desensitization in response to chronic agonist stimulation.[13] However, subsequent human experiments showed agonist-induced desensitization only in persons with the ADRB2 16Arg variant.[13–15] A 2008 meta-analysis examined obesity risk in relation to ADRB2 codon 16 variation (13 studies with 6825 subjects) and in relation to ADRB2 codon 27 variation (28 studies with 14,450 subjects).[11] Random effects meta-regression analysis showed statistically insignificant obesity risks in relation to 16Arg-containing genotype (odds ratio 1.02, 95% confidence interval 0.89–1.18) and in relation to 27Glu-containing genotype (odds ratio 1.11, 95% confidence interval 0.98–1.27). However, meta-analysis restricted to Asian (7 studies with 3575 subjects), Pacific Island (1 study with 1020 subjects), and native South American populations (1 study with 149 subjects) showed statistically significant association between 27Glu-containing genotype and obesity (odds ratio 1.46, 95% confidence interval 1.02–2.10). One study [16] with African-American subjects, but not a second,[17] showed association between obesity measures and ADRB2 genotype.

We defined a large population on the Caribbean island of Tobago with ancestry in common with African-Americans, determined ADRB2 Arg16Gly and Gln27Glu genotypes, measured blood pressure, and used dual energy X-ray absorptiometry (DEXA) to estimate body composition. Analyses examined ADRB2 genotypes in relation to body composition and hypertension.

Materials and Methods

Study sample

Between September 1997 and September 2007, the Tobago Prostate Study used public service announcements, flyers, local health care workers, and word of mouth to solicit 40–79 year-old men in Tobago for participation in a study of periodic prostate cancer screening.[18] The study excluded non-ambulatory, terminally ill, or cognitively impaired men. Participants signed written informed consent. The Institutional Review Boards of the University of Pittsburgh and Tobago Division of Health and Social Services approved the research protocol.

The Tobago Prostate Study enrolled 3837 men. Standardized entry questionnaires provided demographic and basic cancer risk factor information. To support studies of obesity, hypertension, diabetes, and musculoskeletal health, men enrolling or returning after January 2004, also completed a detailed standardized staff-administered health history questionnaire. We defined, for the current analysis, a race-eligible study group that consisted of 3363 (87.6% of 3837) men reporting four grandparents of African descent (when data not missing) or self-reporting black or African race (when grandparents’ data missing).

Procedures included periodic musculoskeletal tests (DEXA and/or quantitative computed tomography) and, for the subset of men who completed the health history questionnaire and accepted musculoskeletal testing, genetic tests. Information from DEXA was available for 2766 (82.2%) and health history questionnaires for 2170 (64.5%) of 3363 race-eligible men and ADRB2 Arg16Gly and/or Gln27Glu genotypes for 1890 of 2049 (92.2%) race-eligible men with both health history questionnaires and DEXA. Data analyses included 1893 (1888 with DEXA) ADRB2 genotyped men with health history questionnaires affirming four grandparents of African descent. When compared against the group of men not available for the health history questionnaire (n=1193), n=12 missing grandparents’ race on health history questionnaire, or not otherwise genotyped (n=265), the 1893 men included in analyses enrolled at younger age (median 52 vs. 56 years, p(Wilcoxon) < 0.0001). Genotyped and non-genotyped race-eligible men available for the health questionnaire had statistically similar body mass index (BMI), waist circumference, hypertension prevalence, and body composition (percent body fat).

ADRB2 Arg16Gly (rs1042713) and Gln27Glu (rs1042714) genotype determinations used a TaqMan allelic discrimination assay performed on an Applied Biosystems (Foster City, CA) 7900HT Fast Real-Time PCR System. Call rates for the ADRB2 Arg16Gly and Gln27Glu polymorphisms were 97.0% and 94.5%, respectively. ADRB2 Arg16Gly, Gln27Glu, and both Arg16Gly and Gln27Glu genotypes were available for 1790 (94.6%), 1800 (95.1%), and 1697 (89.6%) men included in data analyses, respectively. Genotype distributions for ADRB2 Arg16Gly and Gln27Glu satisfied Hardy-Weinberg equilibrium conditions (p>0.05). Allele frequencies (16Gly 49.6%, standard error 0.8% and 27Glu 13.9%, standard error 0.6%) estimated for the 1893 men included in data analyses agreed with values reported by the International HapMap Project (www.hapmap.org) for the Yoruba population (50.0% and 17.5%, respectively). ADRB2 Arg16Gly and Gln27Glu alleles were in linkage disequilibrium (D′=0.96, r²=0.15).

Outcome measurements

Outcome measurements included height (measured without shoes to the nearest 0.1 cm on a wall-mounted stadiometer), weight (measured without shoes to the nearest 0.1 kg on a balance beam scale), BMI (calculated as weight in kg divided by height in meters-squared), and waist circumference (measured in cm at the umbilicus with an inelastic tape measure). Percent body fat was acquired on a Hologic QDR 4500W DEXA operated in array beam mode and analyzed with QDR software version 8.26a (Hologic Inc. Bedford, MA). After seating and resting subjects for 5 minutes, technicians selected an appropriate cuff size and used an automated OMRON HEM705CP sphygmomanometer (Omron Healthcare, Inc., Vernon Hills, IL) to obtain three consecutive blood pressure measurements.

Statistical analysis

Using the second and third measurements, if available, or any two available measurements, otherwise, analyses calculated average values for systolic (SBP) and diastolic blood pressure (DBP). Using these measurements, we classified subjects into three mutually exclusive hypertension categories, hypertension (SBP > 140 mmHg or DBP > 90 mmHg or current antihypertensive medication use), pre-hypertension (SBP 120–139 mmHg or DBP 80–89 mmHg and current antihypertensive medication nonuse), and normal (SBP < 120 mmHg and DBP < 80 mmHg and current antihypertensive medication nonuse). Sixty-two (3.3%) men were not classified because of missing blood pressure measurements (n=6) and/or missing or inconsistent self-reports of antihypertensive medication use (n=61). BMI classification used World Health Organization cutpoints (underweight <18.5 kg/m², normal weight 18.5–24.9 kg/m², overweight 25–29.9 kg/m², and obese ≥30 kg/m²).

Statistical analyses (SAS 9.2; SAS Institute Inc., Cary, NC) used conventional methods (chi-square or Fisher exact tests for discrete data and t-tests, ANOVA, or Wilcoxon rank-sum tests for continuous data) to evaluate the significance of group differences. Haplotype-based analyses used the Expectation-Maximization (EM) algorithm implemented in SAS Genetics (PROC HAPLOTYPE) to estimate group-level haplotype frequencies and to generate subject-level haplotype probability weights.¹ EM algorithm refers to a statistical genetic method commonly used to estimate haplotype frequencies from genotype data where gametic phase is ambiguous for individuals who are heterozygous at more than one loci.[19] We then used simple linear regression (implemented in SAS PROC GENMOD with normal probability distribution and identity link function) to estimate independent associations between the haplotype probability weights and continuous outcomes (BMI and total body percent body fat) and logistic regression (implemented in SAS PROC LOGISTIC) to estimate independent associations between the haplotype probability weights and binary outcomes (hypertension or pre-hypertension vs. normal blood pressure).

We used Quanto (Version 1.2.4; http://hydra.usc.edu/GxE) to estimate the statistical power of studies designed to detect genetic effects on a continuous outcome with 6.3 unit population standard distribution (corresponding to the value observed for percent body fat) under an additive model tested at alpha=0.05 (two-sided). For polymorphisms with 0.50 and 0.14 minor allele frequencies (corresponding to the values observed for Arg16Gly and Gln27Glu), a study genotyping 1850 unrelated subjects detects (at 80% power) per allele effects of 0.6 and 0.9 units, respectively.

Results

Study men were mean age 58.9 years (standard deviation (SD) 10.4 years). Over half (50.5%) were hypertensive, 32.4% pre-hypertensive, 43.9% overweight, and 25.6% obese. One in 6 (16.5%) self-reported a physician diagnosis of diabetes.

Genotype frequencies were 26.0% Arg/Arg, 48.9% Arg/Gly, and 25.1% Gly/Gly at codon 16 (n=1790) and 74.4% Gln/Gln, 23.4% Gln/Glu, and 2.2% Glu/Gly at codon 27 (n=1800). The prevalence of obesity (BMI≥30 kg/m²) did not vary according to genotype (28.0% of 465, 24.6% of 873, 24.5% of 449 men with the codon 16 Arg/Arg, Arg/Gly, and Gly/Gly genotype, p=0.3562, and 26.1% of 1338, 25.7% of 420, and 23.1% of 39 men with the codon 27 Gln/Gln, Gln/Glu, and Glu/Glu genotype, p= 0.9085). Height, weight, and waist circumference were slightly higher in men with a 16Arg allele (Table 1). Arg16Gly associations with height, weight, and waist circumference reached statistical significance (p=0.0181, 0.0363, and 0.0488, respectively). Otherwise, weight, waist circumference, BMI, and percent body fat were unrelated to ADRB2 genotype (Table 1). In the obese subset, body measurements did not vary in any systematic or meaningful way with respect to ADRB2 genotype (Table 1).

Table 1.

Body composition measures (means ± standard deviations) according to ADRB2 genotype, restricted to subjects with four grandparents of African descent.

Body composition measure	Arg16Gly				Gln27Glu
	Arg/Arg	Arg/Gly	Gly/Gly	p-value [1]	Gln/Gln	Gln/Glu	Glu/Glu	p-value [1]
All men
n [2]	462–465	871–875	446–449		1335–1339	418–421	39
Height (cm)	175.8 ± 7.1	175.0 ± 6.9	174.5 ± 6.6	0.0181	174.8 ± 6.8	175.5 ± 7.2	175.8 ± 6.6	0.1393
Weight (kg)	85.1 ± 15.2	84.9 ± 17.1	82.7 ± 15.6	0.0363	84.6 ± 16.2	84.8 ± 17.5	84.2 ± 14.6	0.9683
Waist circumference (cm)	93.6 ± 12.8	93.4 ± 11.2	91.9 ± 11.3	0.0488	93.3 ± 11.6	93.0 ± 11.4	92.3 ± 9.0	0.7723
Body mass index (BMI; kg/m2)	27.5 ± 4.5	27.7 ± 5.2	27.1 ± 4.7	0.1678	27.7 ± 4.9	27.5 ± 5.2	27.2 ± 4.0	0.6617
Total body fat (%) [3]	20.9 ± 6.4	21.2 ± 6.3	20.6 ± 6.2	0.2247	21.2 ± 6.3	20.9 ± 6.5	21.5 ± 6.1	0.6474
Obese men (BMI ≥ 30 kg/m2)
n [2]	128–130	213–215	109–110		345–349	107–108	9
Height (cm)	174.8 ± 6.6	175.4 ± 6.3	174.1 ± 6.8	0.2514	174.6 ± 6.2	175.3 ± 7.4	177.8 ± 7.0	0.2361
Weight (kg)	101.2 ± 12.4	105.0 ± 17.4	101.4 ± 13.0	0.0369	103.0 ± 14.8	104.3 ± 18.2	103.0 ± 8.1	0.7522
Waist circumference (cm)	105.5 ± 9.8	106.5 ± 9.1	104.9 ± 8.2	0.3028	106.0 ± 9.4	105.9 ± 8.3	102.5 ± 3.1	0.5298
Body mass index (BMI; kg/m2)	33.1 ± 3.1	34.1 ± 5.4	33.4 ± 3.4	0.0863	33.8 ± 4.5	33.9 ± 5.0	32.6 ± 1.4	0.7302
Total body fat (%) [3]	25.8 ± 5.2	26.6 ± 4.7	26.0 ± 4.3	0.3408	26.2 ± 4.7	26.5 ± 4.9	26.2 ± 3.8	0.8005

¹Statistical significance (ANOVA) of differences in mean values across genotype categories

²Sample numbers vary because of missing attribute data

³Total body (except head) fat mass expressed as a percentage of total body (except head) mass

ADRB2 16Gly-containing and 27Glu-containing genotypes were equally frequent in low, medium, and high tertiles of percent body fat (Table 2; 16Gly-containing genotypes: 73.4%, 74.4%, 74.5%, p_trend=0.66; 27Glu-containing genotypes: 27.6%, 23.8%, 25.4%, p_trend=0.39) and in normal blood pressure, pre-hypertensive, and hypertensive men (Table 3; 16Gly-containing genotypes: 73.4%, 72.8%, 74.4%, p_trend=0.61; 27Glu-containing genotypes: 25.6%, 24.1%, 26.7%, p_trend=0.50). In men not taking antihypertensive medication, systolic, diastolic, and mean arterial blood pressure values were unrelated to ADRB2 genotype (data not shown).

Table 2.

ADRB2 genotype and body composition, restricted to subjects with four grandparents of African descent.

ADRB2 genotype	Percent Total Body Fat [1]
	Low tertile N=629		Middle tertile N=630		High tertile N=629
	N	%	N	%	N	%
Arg16Gly
Arg/Arg	158	26.6	153	25.6	151	25.5	Pglobal=0.76
Arg/Gly	279	47.0	294	49.2	301	50.7
Gly/Gly	157	26.4	151	25.2	141	23.8
Arg/Gly or Gly/Gly	436	73.4	445	74.4	442	74.5	Ptrend=0.66
Gln27Glu
Gln/Gln	424	72.4	454	76.2	457	74.6	Pglobal=0.24
Gln/Glu	146	24.9	135	22.6	140	22.8
Glu/Glu	16	2.7	7	1.2	16	2.6
Gln/Glu or Glu/Glu	162	27.6	142	23.8	156	25.4	Ptrend=0.39

¹Percent total body fat (total body (except head) fat mass expressed as a percentage of total body (except head) mass) -- Low tertile: 4.769–18.764; Middle tertile: 18.765–23.745; High tertile: 23.746–44.352

Table 3.

ADRB2 genotype and hypertension, restricted to subjects with four grandparents of African descent.

ADRB2 genotype	Normal blood pressure N=314		Pre- hypertensive N=593		Hypertensive N=924
	N	%	N	%	N	%
Arg16Gly
Arg/Arg	79	26.6	153	27.2	223	25.6	Pglobal=0.94
Arg/Gly	143	48.2	274	48.8	423	48.6
Gly/Gly	75	25.2	135	24.0	225	25.8
Arg/Gly or Gly/Gly	218	73.4	409	72.8	648	74.4	Ptrend=0.61
Gln27Glu
Gln/Gln	218	74.4	432	75.9	646	73.3	Pglobal=0.73
Gln/Glu	71	24.2	125	22.0	217	24.6
Glu/Glu	4	1.4	12	2.1	18	2.1
Gln/Glu or Glu/Glu	75	25.6	137	24.1	235	26.7	Ptrend=0.50

Estimated ADRB2 Arg16Gly Gln27Glu haplotype frequencies were Arg-Gln 50.0% (standard error 0.8%), Gly-Gln 36.0% (standard error 0.8%), Gly-Glu 13.6% (standard error 0.6%), and Arg-Glu 0.3% (standard error 0.1%). The Figure uses box plots to summarize BMI and percent body fat distributions in men classified according to most probable haplotype combination. In linear models, the less common Gly-Gln and Gly-Glu haplotypes associated with lower BMI and lower percent body fat (Table 4). In obese men, one or both haplotypes associated with higher percent body fat and lower waist circumference (Table 4). However, none of these associations, including the apparent interaction between haplotype and obesity grouping (data not shown), was statistically significant. Finally, logistic regression did not show statistically significant ADRB2 haplotype associations with either pre-hypertension or hypertension (data not shown).

Figure.

Box plots showing body mass index (BMI; upper figure) and DEXA-determined percent body fat (lower figure) in men sub-grouped according to diplotype (most probable ADRB2 Arg16Gly Gln27Glu haplotype combination). The number (N) in each subgroup appears above the diplotype label. Analyses exclude six men with an Arg-Glu haplotype-containing diplotype. The line segment and diamond symbol within each box identify the median and mean, respectively. The lower and upper borders of each box identify the 25^th and 75^th percentiles, respectively. The lower whisker extends to the minimum observation greater than or equal to the 25^th percentile minus 1.5 times the inter-quartile range. The upper whisker extends to the maximum observation less than or equal to the 75^th percentile plus 1.5 times the inter-quartile range. Open circle symbols identify individual observations less than the 25^th percentile minus 1.5 times the inter-quartile range or greater than the 75^th percentile plus 1.5 times the inter-quartile range.

Table 4.

Estimated difference (Δ; and 95% confidence interval) in body composition measure between men homozygous for a specified haplotype relative to men homozygous for the Arg-Gln haplotype.

Body composition measure	n	Gly-Gln			Gly-Glu
		Δ	95% CI	p-value	Δ	95% CI	p-value
BMI (kg/m2)	1890	−0.36	−1.06, 0.34	0.3109	−0.59	−1.56, 0.39	0.2370
Total body fat % [1]	1888	−0.29	−1.19, 0.60	0.5198	−0.54	−1.79, 0.70	0.3923
Total body fat % [1], in obese men	479	0.00	−1.31, 1.31	0.9989	0.51	−1.35, 2.37	0.5916
Waist circumference (cm), in obese men	481	−0.42	−2.93, 2.10	0.7458	−1.29	−4.88, 2.30	0.4819

¹Total body (except head) fat mass expressed as a percentage of total body (except head) mass

Discussion

Mediating fat cell lipolysis[2, 8] and vascular smooth muscle relaxation,[10] ADRB2 gene variants that code for functionally altered receptors could promote weight gain or high blood pressure. In a study of 1893 Tobago men of African ancestry, however, we observed no significant body composition or blood pressure associations with the two most commonly studied ADRB2 gene variants.

Different investigations have reached different conclusions about the significance of ADRB2 genotype in relation to obesity and related phenotypes [11, 20–22]. Variability related to sex may explain inconsistencies in the published literature. Some authors, including Corbalan et al. [23] and González Sánchez et al. [24], suggest that ADRB2 genotype determines not obesity, but obese subtype, perhaps in only one sex group. For example, in a Spanish clinic-based study with 40 men and 199 women, Corbalan et al. [23] compared the ADRB2 Gln27Glu genotypes of subjects with either abdominal obesity (BMI >30 kg/m² and waist-to-hip ratio >0.85) or normal body mass (BMI <25 kg/m² and waist-to-hip ratio <0.85). In men, but not women, 27Glu-containing genotypes were more frequent in the group with abdominal obesity. In a Spanish population-based study with 319 white men and 347 white women, González Sánchez et al. [24] noted a statistically non-significant trend in men (but not women) between ADRB2 Gln27Glu genotype and obesity prevalence (20.0% of 130, 27.7% of 155, and 29.4% of 34 men with the codon 27 Gln/Gln, Gln/Glu, and Glu/Glu genotype, respectively, p= 0.2572). In the obese subset, men with the rare Glu/Glu genotype had significantly higher mean BMI (34.1 kg/m², n=10) than men with either the Gln/Glu (31.9 kg/m², n=46, p=0.013) or Gln/Gln genotype (32.0 kg/m², n=26, p=0.023). To support specific association with abdominal obesity, González Sánchez et al. [24] noted that obese men with the rare Glu/Glu genotype also had significantly higher mean sagittal abdominal diameter (27.8 cm, n=10) than obese men with either the Gln/Glu (24.9 cm, n=46, p=0.037) or Gln/Gln genotype (24.9 cm, n=26, p=0.062). Among obese women, sagittal abdominal diameter was not related to ADRB2 Gln27Gly genotype. In agreement with González Sánchez et al., we observed no statistical association between obesity prevalence and ADRB2 genotype. Among obese Afro-Caribbean men, BMI, waist circumference, and body fat values were unrelated to ADRB2 genotype (Table 1). Simply, our results do not support a relationship, specifically in obese men, between ADRB2 genotype and obesity subtype.

Excluding persons treated for high blood lipids, high blood pressure, or diabetes, Meirhaeghe et al. [25] studied a population-based sample of 836 35–64 year-old urban-dwelling persons from northern France (419 men and 417 women, 14.2% obese overall) and observed sex-specific associations between ADRB2 genotype and body composition. Body weight, BMI, waist circumference, hip circumference, and waist-to-hip ratio mean values were significantly higher in men with the codon 16 Arg/Arg genotype than men with either the Arg/Gly or Gly/Gly genotype and significantly higher in men with the codon 27 Glu/Glu genotype than men with either the Gln/Glu or Gln/Gln genotype. In women, differences were not statistically significant. Observing linkage disequilibrium between 16Arg and 27Gln, Meirhaeghe et al. [25] compared, in men, body composition measures according to combined genotypes at Arg16Gly and Gln27Gly. With the Gly-Glu haplotype serving as reference, higher body weight, BMI, and waist-to-hip ratio associated with the Arg-Gln haplotype, not with the Gly-Gln haplotype. In Afro-Caribbean men, body weight and waist circumference were slightly higher in men with a 16Arg allele (Arg/Arg or Arg/Gly genotype) than men without a 16Arg allele (Gly/Gly genotype; Table 1). However, BMI and percent body fat did not vary according to Arg16Gly genotype. Body weight, waist circumference, BMI, and percent body fat were completely independent of Gln27Glu genotype.

Only two studies [16, 17] to date have included a meaningful number of African-Americans. The Insulin Resistance and Atherosclerosis (IRAS) Family Study [16] genotyped 272 African Americans and 720 Hispanic Americans from 18 and 45 families, respectively, and used single-slice computed tomography to measure visceral and subcutaneous fat. Obesity measures associated with Gln27Glu, specifically the Glu/Glu genotype, but not with Arg16Gly genotype. High visceral fat area associated with the Glu/Glu genotype, even after control for BMI. Results for men and women and for African and Hispanic Americans were statistically indistinguishable. The Heritage Family Study included 274 African-Americans (31.8% obese) and 502 whites (19.3% obese) and used under water weighing to measure fat mass, single-slice computed tomography to measure visceral and subcutaneous fat, and skin calipers to measure skin-fold thickness.[17] In white obese (BMI≥30 kg/m²) subjects only, the Heritage Family Study observed lower fat mass in obese white men with 27Glu-containing genotypes and lower fat mass in obese white women with 16Gly-containing genotypes. In African-American subjects, the Heritage Family Study did not report any statistically significant cross-sectional associations between ADRB2 genotype and any body composition measure.

One study of ADRB2 genotype included Afro-Caribbeans recruited from primary care clinics located on St. Vincent [26]. The case group included 136 patients (19.9% men) with high blood pressure (DBP > 95 mmHg or antihypertensive medication use) and family history of hypertension. The control group included 81 patients (46.9% men) with normal blood pressure (DBP < 85 mmHg and antihypertensive medication nonuse) and no family history of hypertension. The 16Gly allele was much more frequent in the case group (84.6% vs. 66.7%; p=0.000014). Our community-based study of 1893 Afro-Caribbean men from Tobago included 318 and 537 persons who satisfied the St. Vincent case and control definitions. In Tobago, 16Gly was also more frequent in cases, 51.9% vs. 48.4%. However, the difference was not statistically significant (p(allele chi-square)=0.1655). In the St. Vincent study, genotype frequencies in both the case and control groups violated Hardy-Weinberg equilibrium. The control frequency (66.7%) of the putative risk allele (16Gly) in St. Vincent was high when compared against reference frequencies in the HapMap Yoruba (50.0%) and Tobago populations (49.6%). Therefore, selection bias and genotyping error plausibly explain the anomalous St. Vincent results.

Our study participants volunteered for prostate cancer screening, in many instances survived a variable duration after study entry, and agreed to extra visits that included measurements, such as DEXA. Therefore, absence of meaningful association between ADRB2 genotype and body composition could occur if these study selection factors preferentially excluded men with certain genotype-phenotype combinations. However, selection bias deriving from factors related to initial study participation can be discounted to the extent that Tobago Prostate Study, as a whole, enrolled a high proportion (60%) of age-eligible men (approximately 5000) living in Tobago. To address possible survival bias, we compared the 1661 and 232 men who had body composition measurements after study entry and coincident with study entry, respectively, and found statistically similar relationships between ADRB2 genotype and body composition (data not shown). Finally, study procedures captured a reasonable proportion (1893 of 3363, 56.3%) of race-eligible Tobago Prostate Study enrollees.

Results from some studies suggest specific ADRB2 association with regional fat distribution [16, 23–25]. Our primary body composition measures, BMI and DEXA-derived percent body fat, do not distinguish between visceral and subcutaneous fat. However, our data set included waist circumference, an indirect measure of central obesity. Waist circumference and ADRB2 genotype were statistically independent (Table 1). When compared against criterion methods, such as total body water measured by isotope dilution, the Hologic QDR 4500W DEXA operated in array beam mode underestimates fat mass and percent body fat by 10%.[27] Unless related to genotype, this systematic measurement error should not bias associations observed between genotype and body composition.

Finally, we evaluated only two ADRB2 genetic variants. These variants change the amino acid sequence of the β₂ adrenergic receptor protein. However, these changes are not generally believed to alter agonist binding or signal transduction [13]. The Arg16Gly and Gln27Glu variants may affect function secondarily, perhaps through agonist-induced receptor protein down regulation or linkage disequilibrium with other less common but more influential variants [13].

In a large racially homogenous population of men with black African ancestry and high prevalence of obesity and hypertension, single variant and haplotype-based analyses did not show meaningful or consistent association between ADRB2 Arg16Gly and Gln27Glu variation and phenotypes related to obesity and hypertension.