4.4 Results

Of over 3000 men participating in the Caerphilly study, there is follow-up data available on 2010 subjects. Two groups were formed from this cohort:

• The 347 men who are in the top 20% of the diastolic distribution at either phase II or phase III, and in the top 40% at the other phase
• The 284 men who are in the bottom 20% of the diastolic distribution at either phase II or phase III, and in the bottom 40% at the other phase. Additionally, these men are not on any antihypertensive drug at either phase II or phase III.

Phase II data was collected 1983–1985, phase III data in the years 1989–1991 (demographic data are summarised in Table 4.1). Evolution of age, BMI, diastolic and systolic blood pressure are visualised in Figure 4.4. Aside from blood pressure measurements and body mass index, the two quintiles did not show any significant differences in both phases. Genomic DNA from these subjects was genotyped for the presence of the Gly16Arg, Glu27Gln as well as the 5’-LC alleles using direct sequencing of both strands (for protocol details see section 4.2.2). Genotyping of the β₂ adrenergic receptor variant was cross-checked by a second individual also unaware of the subject’s phenotype. For a representative example of Dye Primer sequencing, see Figure 4.5. Dye Terminator sequencing examples are shown in Figure 4.6 for the Arg/Gly16 variant and Figure 4.7 for the Glu/Gln27 variant.

(a) Age (y) (b) Body mass index (kg/m2) (c) Systolic blood pressure (mmHg) (d) Diastolic blood pressure (mmHg) Figure 4.4: Age, Body mass index, Diastolic and Systolic blood pressure trends for phase II and phase III of the Caerphilly data set

(a) 1309A (b) 1309G (c) 1309R Figure 4.5: Dye Primer sequencing (forward strand) of the Arg/Gly16 (A1309G) mutation

(a) 1309T (b) 1309C (c) 1309Y Figure 4.6: BigDye Terminator sequencing (reverse strand) of the Arg/Gly16 (A1309G) mutation

(a) 1342C (b) 1342G (c) 1342S Figure 4.7: BigDye Terminator sequencing (forward strand) of the Gln/Glu27 (C1342G) mutation

The distribution of β₂ adrenoceptor alleles did confirm to Hardy-Weinberg equilibrium in this study. There was marked disequilibrium in the distribution of the Glu27→Gln alleles between the top and bottom 20% groups (χ²= 5.5, p=.021), with the frequency of the Gln27 allele markedly increased in the hypertensive subjects. There was no difference in allele frequencies for the Arg16→Gly variant (Table 4.9).

The mutations at positions 1217 (5’-LC Cys/Arg19), 1244 (C1244T) and 1342 (Glu/Gln27) were found to be in strong linkage disequilibrium (χ² = 903.8, p<0.0001), leading to three different main haplotypes (Table 4.10). Haplotype frequencies were estimated by maximum likelihood estimation, using the best state of haplotype composition, under assumption of equal prior probabilities, as starting point for the expectation maximisation (EM) algorithm. In order to get an impression of the significance of the estimated frequencies, a bootstrap algorith was used, simulating the sample data on the basis of the estimated frequencies and re-estimating them for 100 times, calculating mean and variance.

Figure 4.8 shows the correlation between both, systolic and diastolic, blood pressure and BMI. As age and BMI are known predictors of blood pressure, we performed multiple regression analyses on systolic and diastolic blood pressure as dependent variables (Table 4.11) in both phases of the Caerphilly data set. All available quantitative and qualitative factors were entered in the analysis to investigate their possible influence on blood pressure. BMI (p < 0.0005), age (p < 0.0005), alcohol consumption (p<0.006) and the Gln27Gln polymorphism (p<0.0005) remained as significant variables in the model for systolic blood pressure whereas BMI (p < 0.0005), family history of heart attack in first degree relative before 55yrs (p=0.020), smoking (p=0.026) and the Gln27Glu polymorphism (p=0.079) remained as explanatory variables in the diastolic blood pressure model. These effects matched with multiple regression analysis for pulse pressure as dependent variable, resulting in age (p<0.0005), the Gln27Gln polymorphism (p<0.0005), alcohol consumption (p<0.001), family history of heart attack in first degree relative before 55yrs (p=0.027) and BMI (p=0.091) as independent variables.

(a) Systolic blood pressure (b) Diastolic blood pressure Figure 4.8: Scatter diagram of body mass index (kg/m2) against systolic (a) and diastolic (b) blood pressure (mmHg) in phase II of the Caerphilly data set

ANOVA analysis using BMI, age and alcohol consumption as covariants of phase II data showed a significant association with systolic blood pressure (p=.0011) and pulse pressure (p=.0007) for the Glu27Gln polymorphism (Figure 4.9). Gln27 homozygotes showed a significantly higher systolic mean blood pressure than Glu27 homozygotes (SBP difference = 14.6 mmHg, Table 4.15). Results for diastolic blood pressure showed the same tendencies, but were no longer significant (p=.1269). ANOVA for the 5’ LC polymorphism (Arg19Cys) was highly significant for pulse pressure (p=.0008) and systolic blood pressure (p=.0018) in phase II (Figure 4.11). Mean systolic blood pressure difference between Arg and Cys19 homozygotes was 12.3 mmHg (Table 4.18). Diastolic blood pressure showed the same trends, but was not significant (p=.1833). Due to linkage disequilibrium the variant at position 1244 showed similar results. The Arg16Gly variant did not show any significant association with diastolic, systolic or pulse pressure (Figure 4.13) in phase II or phase III.

ANOVA of phase III data resulted in the same trends for the Glu27Gly polymorphism as in phase II, but lost significance (Figure 4.10). Association with systolic blood pressure (p=.06) and diastolic blood pressure (p=.08) were marginally significant, pulse pressure was no longer significant (Table 4.15). Analysis of variance of the 5’ LC Arg19Cys variant in phase III lost significance for blood pressure completely (Figure 4.12). Multiple range tests showed a remaining significant difference for mean systolic blood pressure in homozygotes only (Table 4.15). The method being used to discriminate among the means is Fisher’s least significant difference (LSD, 95% confidence intervals) procedure. Age, body mass index, alcohol consumption and smoking habits did not exhibit any association with 5’ LC-Arg19Cys, Arg16Gly or Glu27Gln variants.

The multivariate analyses revealed that genotype of Gln27Glu, current age, and BMI were predictors of both systolic and diastolic blood pressure. According to the estimation of R² (adjusted), the Gln27Glu polymorphism could account for 10.8% and 15.2% of variances in systolic and diastolic blood pressure, respectively.

Haplotype analysis, using the 3 main haplotypes (Table 4.10), revealed significant differences for mean systolic blood pressure (p=.0083) and pulse pressure (p=.0196) in phase II of the Caerphilly data set (Figure 4.14). Genotype 11 (5’ LC-Arg19 Gly16 Glu27) showed the lowest mean pressure, significantly lower than combinations of the other two 5’ LC Cys19 genotypes (Table 4.22). This effect was present in phase III data as well, although significance was lost for systolic blood pressure and pulse pressure (Figure 4.15).

(a) Pulse pressure (mmHg), p=.0007 (b) Systolic blood pressure (mmHg), p=.0011 (c) Diastolic blood pressure (mmHg), p=.1269 Figure 4.9: ANOVA with relevant covariates as emerged in multiple regression analysis of Pulse pressure, Systolic and Diastolic blood pressure for the Gln27Glu variant of the human β2 adrenergic receptor in phase II of the Caerphilly data set (means and 95% LSD intervals)

(a) Pulse pressure (mmHg), p=.3440 (b) Systolic blood pressure (mmHg), p=.0638 (c) Diastolic blood pressure (mmHg), p=.0826 Figure 4.10: ANOVA with relevant covariates as emerged in multiple regression analysis of Pulse pressure, Systolic and Diastolic blood pressure for the Gln27Glu variant of the human β2 adrenergic receptor in phase III of the Caerphilly data set (means and 95% LSD intervals)

(a) Pulse pressure (mmHg), p=.0008 (b) Systolic blood pressure (mmHg), p=.0018 (c) Diastolic blood pressure (mmHg), p=.1899 Figure 4.11: ANOVA with relevant covariants as emerged in multiple regression analysis of Pulse pressure, Systolic and Diastolic blood pressure for the 5’ LC-Arg19Cys polymorphism of the human β2 adrenergic receptor in phase II of the Caerphilly data set (means and 95% LSD intervals)

(a) Pulse pressure (mmHg), p=.3475 (b) Systolic blood pressure (mmHg), p=.1224 (c) Diastolic blood pressure (mmHg), p=.1692 Figure 4.12: ANOVA with relevant covariants as emerged in multiple regression analysis of Pulse pressure, Systolic and Diastolic blood pressure for the 5’ LC-Arg19Cys polymorphism of the human β2 adrenergic receptor in phase III of the Caerphilly data set (means and 95% LSD intervals)

(a) Pulse pressure (mmHg), n.s. (b) Systolic blood pressure (mmHg), n.s. (c) Diastolic blood pressure (mmHg), n.s. Figure 4.13: Simple ANOVA of Pulse pressure, Systolic and Diastolic blood pressure for the Arg16Gly polymorphism of the human β2 adrenergic receptor in phase II of the Caerphilly data set

(a) Pulse pressure (mmHg), p=.0196 (b) Systolic blood pressure (mmHg), p=.0083 (c) Diastolic blood pressure (mmHg), p=0.1068 Figure 4.14: ANOVA of Pulse pressure, Systolic and Diastolic blood pressure for the main haplotypes of the human β2 adrenergic receptor in phase II of the Caerphilly data set (Haplotype 1: 5’ LC-Arg19/1242C/Gly16/Glu27, haplotype 2: 5’ LC-Cys19/1242T/Arg16/Gln27, haplotype 3: 5’ LC-Cys19/1242T/Gly16/Gln27)

(a) Pulse pressure (mmHg), n.s. (b) Systolic blood pressure (mmHg), n.s. (c) Diastolic blood pressure (mmHg), n.s. Figure 4.15: ANOVA of Pulse pressure, Systolic and Diastolic blood pressure for the main haplotypes of the human β2 adrenergic receptor in phase III of the Caerphilly data set (Haplotype 1: 5’ LC-Arg19/1242C/Gly16/Glu27, haplotype 2: 5’ LC-Cys19/1242T/Arg16/Gln27, haplotype 3: 5’ LC-Cys19/1242T/Gly16/Gln27)

© 2001 Alexander Binder