Context-Dependent Associations of the ACE I/D Polymorphism With Blood Pressure
Abstract—The objective of the present study was to assess whether the influences of gender, age, or measures of body size on blood pressure are homogeneous among genotypes of the insertion/deletion (I/D) polymorphism of the gene that codes for angiotensin-converting enzyme (ACE). We studied a sample of 1875 non-Hispanic white individuals (988 female and 887 male subjects) between 5 and 90 years of age from the general population of Rochester, Minn. When statistical interactions between effects associated with the I/D polymorphism and age, height, and weight were not considered, there was no evidence of a significant relationship between variation in blood pressure level or diagnostic category (hypertension versus normotension) and variation in ACE genotype in either gender. However, in females 5 to 29.9 years of age, the linear regression relationships of systolic blood pressure level with age and weight and of diastolic blood pressure level with age were significantly heterogeneous among ACE genotypes. For these concomitant traits, the rank order of expected blood pressure levels associated with each genotype reversed from low values of the concomitant, in which blood pressure was lower for I/D heterozygotes than for II or DD homozygous, to high levels of the concomitant, in which blood pressure was higher for I/D heterozygotes than for II or DD homozygotes. In male subjects 50 to 90 years of age, the logistic regression relationship of the probability of having hypertension with height was also heterogeneous among ACE genotypes; it was statistically significant in II homozygotes but not statistically significant in either I/D heterozygotes or DD homozygotes. Findings of this study are consistent with the conclusion that the influence of variation in the ACE gene on interindividual variation in blood pressure is dependent on contexts that are indexed by gender, age, and measures of body size.
Two studies of large population-based samples recently presented evidence that variation in the gene coding for angiotensin-converting enzyme (ACE) influences interindividual differences in blood pressure.1 2 Both studies found evidence of linkage between a highly polymorphic microsatellite marker in the regulatory region of the gene that codes for the human growth hormone (hGH) and a gene that influences the diastolic blood pressure level in males but not females. The ACE gene was implicated because markers of it were in strong linkage disequilibrium with the hGH marker.3
Association studies with biallelic markers can also be helpful in indicating whether variation in a candidate gene may influence blood pressure. Most association studies that involved the ACE gene have relied on an insertion/deletion (I/D) polymorphism in intron 16, with the D allele associated with an increase in plasma ACE activity in a codominant fashion.4 Several previous studies reported associations between interindividual variation in blood pressure level5 6 or the probability of having hypertension7 8 9 and the I/D polymorphism. In the recent study of O’Donnell and colleagues,1 increased diastolic blood pressure and increased odds of having hypertension were associated with the D allele in men but not women. Although numerous other studies reported no association between measures of blood pressure and the I/D polymorphism,10 11 12 13 only one of these latter studies considered women and men separately.11
Despite universal recognition that blood pressure is influenced by interactions between the effects of many genetic and environmental factors, the possibility of detecting statistically significant interactions between genetic effects measured by the ACE I/D polymorphism and effects of other genetic and environmental factors has received little attention. Because such interactions may give rise to different relationships between genotypic variation and phenotypic variation in different environments,14 15 some previous studies may have falsely concluded that interindividual variation in blood pressure is not associated with the I/D polymorphism because the context dependency of the association was not considered.
The objective of the present study was to assess whether the influences of gender, age, or measures of body size on blood pressure are homogeneous among genotypes of the ACE I/D polymorphism. We studied a sample of non-Hispanic white individuals from the same 3-generation pedigrees that provided the young male sibling pairs in which Fornage and colleagues2 reported evidence of linkage between the hGH microsatellite marker and a gene that influenced diastolic blood pressure. We found evidence that the associations between measures of blood pressure and variation in genotypes of I/D polymorphism were not only dependent on gender but also on age and measures of body size.
The sample of 1875 individuals was composed of 988 non-Hispanic white female and 887 non-Hispanic white male subjects who remained after the following exclusion criteria were applied to 2326 members of 598 three-generation pedigrees that participated in the Rochester Family Heart Study between 1984 and 1991.16 These exclusion criteria were membership in a pedigree with inconsistencies in the Mendelian inheritance of I/D alleles of the ACE gene (n=219); no physical examination data (n=79); no laboratory data (n =21); not fasting at the time of blood drawing (n=47); missing diastolic blood pressure (n=9); factors causing low blood pressure in a person categorized as normotensive (see diagnostic categories described below) (n=41); factors causing elevated blood pressure in a person categorized as hypertensive (n=23); and age <50 years in a person categorized as hypertensive and treated with blood pressure lowering medication (n=12).
Sampled individuals of each gender were stratified into 3 age groups: 5 to 29.9 years; 30 to 49.9 years; and 50 to 90 years, which corresponded to the age ranges of children, parents, and grandparents. Among subjects <50 years of age, systolic and diastolic blood pressure levels were analyzed as continuously-distributed, quantitative traits because individuals with factors that alter blood pressure were excluded (see exclusion criteria above). Among subjects 50 to 90 years of age, blood pressure diagnostic category (hypertension versus normotension) was considered a discrete trait because individuals treated with antihypertensive medications were included.
The study protocol was approved by the Institutional Review Board of the Mayo Clinic and all procedures were followed in accordance with institutional guidelines. Each participant signed a consent form; afterward, a physician reviewed the participant’s medical history and performed a brief physical examination including measurement of height, weight, and blood pressure as previously described.16 Three blood pressure readings, ≥2 minutes apart, were obtained in the right arm after the participant sat quietly for ≥5 minutes. Averages of the 3 readings were used in the analyses.
Blood Pressure Diagnostic Categories
Each participant was assigned to 1 of the blood pressure diagnostic categories defined by the following criteria.
These individuals had never been treated with medication for hypertension, and their systolic and diastolic blood pressures were <140 and <90 mm Hg, respectively, at the clinic visit. None of these individuals had concomitant illnesses or were being treated with medications known to lower blood pressure.
These individuals reported a previous diagnosis of hypertension and were being treated with antihypertensive medications or their systolic blood pressure was ≥140 mm Hg or their diastolic blood pressure was ≥90 mm Hg at the study visit.
Blood samples drawn before blood pressure measurement were anticoagulated with EDTA, and buffy coat was separated and stored at −80°C. Genomic DNA was extracted from thawed buffy coat by a salting-out procedure. Aliquots of 100 ng of genomic DNA were transferred to 96-well plates (Costar) for amplification of the I/D polymorphism of the ACE gene by polymerase chain reaction as described by Rigat and colleagues.4 Polymerase chain reaction products were electrophoresed on 2% agarose gels; the 190-bp deletion allele and the 490-bp insertion allele were identified by ethidium bromide staining.
For quantitative traits, means and standard deviations were calculated for each ACE genotype within each age and gender group; a 1-way ANOVA was used to assess differences in means among genotypes within the age and gender groups. Relative frequencies of ACE genotypes and alleles were calculated for each age and gender group; in subjects 50 to 90 years of age, relative frequencies of blood pressure diagnostic categories were calculated for each gender and genotype group. A χ2 contingency test was used to assess differences in relative frequencies between genders within each age group and among age groups and genotype groups within each gender.
In age and gender groups in which the subjects were <50 years of age, we used linear regression models to assess whether variation in ACE genotype made a statistically significant contribution to the prediction of systolic or diastolic blood pressure level. In groups in which the subjects were 50 to 90 years of age, we used logistic regression models to assess whether variation in ACE genotype made a statistically significant contribution to prediction of blood pressure diagnostic category. With each type of regression model, we first estimated the influence of variation among ACE genotypes on variation in measures of blood pressure level (or probability of having hypertension) and ignored variation in age, height, and weight. Second, we estimated the influence of variation among ACE genotypes on variation in measures of blood pressure level (or probability of having hypertension) after adjustment for variation in age, height, and weight. Third, to assess heterogeneity among genotypes in the regression of blood pressure level or diagnostic category on each concomitant trait, we included in the regression model the interaction of ACE genotype with the concomitant being considered. Results were considered statistically significant when an observed test statistic was expected ≤5% of the time if the null hypothesis were true.
ACE Genotypes and Alleles
For all female subjects, relative frequencies of the II, I/D, and DD genotypes were 0.201, 0.479, and 0.320, respectively; relative frequencies of the I and D alleles were 0.441 and 0.559, respectively (Table 1⇓). For all male subjects, relative frequencies of the II, I/D, and DD genotypes were 0.178, 0.494, and 0.328; relative frequencies of the I and D alleles were 0.425 and 0.575, respectively. The genotype and allele frequencies did not differ significantly between genders or among age groups within either gender (Table 1⇓). In each age and gender group, the observed genotype frequencies did not differ significantly from those predicted by the Hardy-Weinberg law.
Means and variances for age, height, and weight did not differ significantly among ACE genotypes except in 2 of the age and gender groups (Table 2⇓). In female subjects 50 to 90 years of age, mean weight and its variance differed significantly among genotypes (P=0.030 for the contrast of means, and P=0.011 for the contrast of variances); and in female subjects 30 to 49.9 years of age, the variance of weight differed significantly among genotypes (P=0.044).
Associations of Blood Pressure with ACE Genotype
In age and gender groups with subjects <50 years of age, means for systolic and diastolic blood pressure did not differ significantly among ACE genotypes (Table 3⇓). Variation in ACE genotype also did not make a statistically significant contribution to prediction of blood pressure level after adjustment by linear regression for variation in age, height, and weight. In these latter models (not shown), the percentage of interindividual variation in adjusted systolic or diastolic blood pressure level (R2×100%) explained by variation in ACE genotype was <2%. In contrast, the percentage of interindividual variation in systolic blood pressure level explained by the combined effects of variation in age, height, and weight ranged from 6.7% (males 30 to 49.9 years, P=0.003) to 34.3% (males 5 to 29.9 years, P<0.001), and the percentage of interindividual variation in diastolic blood pressure level explained ranged from 2.0% (females 5 to 29.9 years, P=0.03) to 6.6% (females 30 to 49.9 years, P<0.001).
In females and males 50 to 90 years of age, relative frequencies of the blood pressure diagnostic categories did not differ significantly among genotypes (Table 4⇓). Consistent with this observation, ACE genotype did not make a statistically significant contribution to the prediction of blood pressure diagnostic category in females or males when it was the only predictor included in logistic regression models (not shown). This was also true when ACE genotype was entered into the logistic regression model after entering age, height, and weight. In contrast, in these latter models, the combined effects of variation in age, height, and weight made statistically significant contributions to predicting blood pressure diagnostic category in each gender (P<0.001).
Heterogeneity of Regression Relationships
We conducted additional analyses to assess whether the regression relationships of blood pressure level or diagnostic category with age, height, or weight differed significantly among ACE genotypes (Table 5⇓ and Figures 1⇓ and 2⇓). These analyses addressed whether genetic effects on blood pressure measured by the I/D polymorphism might be dependent on levels of the concomitant traits. (In contrast, the analyses presented in Tables 3⇑ and 4⇑ assumed that these effects were independent of variation in age and measures of body size.) In female subjects 5 to 29.9 years old, we found that the regressions of systolic blood pressure level on age and on weight and of diastolic blood pressure level on age were significantly heterogenous among genotypes (Figure 1⇓ and Table 5⇓). In each instance, the slope of the regression relationship (Figure 1⇓) and the percentage of interindividual variation in blood pressure explained by the regression relationship (Table 5⇓) was greater in I/D heterozygotes than in II or DD homozygotes. Moreover, the rank order of expected blood pressure level reversed from low values of the concomitant trait, in which blood pressure level was lower for I/D heterozygotes than for II or DD homozygotes, to high values of the concomitant trait, in which blood pressure level was higher for I/D heterozygotes than for II or DD homozygotes (Figure 1⇓). In male subjects 5 to 29.9 years old and in both females and males 30 to 49.9 years old, there was no evidence of statistically significant heterogeneity in the linear regression relationships of systolic or diastolic blood pressure level with age, height, or weight (not shown).
In males 50 years of age or older, we found that the logistic regression of the probability of having hypertension on height was significantly heterogenous among genotypes (Figure 2⇑). The relationship between the probability of having hypertension and height was inverse and significantly different from zero in II homozygotes (P<0.001) but was not significantly different from zero in I/D heterozygotes (P=0.243) or in DD homozygotes (P=0.562).
This study was prompted by recent reports that suggested that the relationship between interindividual variation in blood pressure and variation in the ACE gene may be gender-specific.1 2 Most previous studies did not consider this possibility nor did they assess whether genetic effects measured by variation in the ACE gene are dependent on other measures of genetic and environmental variation. In this study, we only found evidence of associations between measures of blood pressure and the ACE I/D polymorphism that were gender-specific and dependent on levels of other concomitant traits including age, height, and weight. Although further analyses in other samples will be required to confirm the validity of our findings, several considerations argue against their dismissal as type 1 error.
First, the regulation of blood pressure level through interactions of multiple biochemical, physiological, and anatomical traits with redundant, counterbalancing vasopressor and depressor actions17 makes it unlikely that variation in any single etiologic factor (genetic or environmental) will have a large effect on interindividual differences in blood pressure. Moreover, a considerable body of evidence from studies in experimental animals and plants indicates that the relationship between phenotype and genotype is usually not constant over a broad range of environments. Each genotype may have a different “norm of reaction” that determines the distribution of possible phenotypes associated with a distribution of environments.14 15 The regression lines in Figures 1⇑ and 2⇑ indicate different norms of reaction for the expected blood pressure phenotype associated with each ACE genotype over the observed range of age or measures of body size. In this sample, not only the magnitude but also the direction of blood pressure differences among ACE genotypes changed across the range of environments indexed by measures of age, height, and weight. The greater slope and range of expected blood pressure levels in I/D heterozygotes than in II or DD homozygotes (Figure 1⇑) suggests that the latter are more “homeostatic” than I/D heterozygotes. These figures illustrate how differences between ACE genotypes may be apparent in particular environments, but, when effects are averaged over all environments, the I/D polymorphism appears not to influence blood pressure level or diagnostic category (Figures 1⇑ and 2⇑).
To the best of our knowledge, the few previous studies that have investigated have found evidence of interactions between effects measured by the ACE I/D polymorphism and other measures of genetic and environmental variation. In one study, the DD genotype was associated with higher blood pressure levels only in a subset of untreated hypertensive patients who did not have additional risk factors for cardiovascular disease (smoking, hypercholesterolemia, or diabetes).18 The authors proposed that such context dependencies may in part account for inconsistencies in the effects attributed to the ACE polymorphism among studies that have sampled geographically, racially, and medically diverse populations. In a sample from the Framingham Heart Study, the DD genotype was a significant predictor of increased odds of having hypertension only in individuals who were also homozygous for a hypertension-associated allele at the angiotensinogen locus.19 Finally, in a sample of untreated hypertensive patients, systolic and diastolic blood pressure levels were influenced by interaction between effects associated with the ACE I/D polymorphism and a polymorphism of the angiotensin II receptor type 1.5
Because overlap of the sample with our previous linkage study2 was limited to the 5- to 29.9-year-old male and female subjects in the present study, it is appropriate to ask why we found no evidence of association between the ACE I/D polymorphism and blood pressure levels in 5- to 29.9-year-old male subjects in this study, although our previous linkage study was positive only in young male siblings. One possible explanation is that the ACE I/D polymorphism, which itself may have no effect on blood pressure, is in linkage equilibrium with functional variants elsewhere in the ACE gene that are responsible for the blood pressure effects detected in the previous linkage study.20 Moreover, the previous linkage study used a highly polymorphic microsatellite marker in the regulatory region of the hGH gene and cannot exclude the possibility that variation in another closely-linked gene, instead of ACE, was responsible for the positive linkage result.2 In addition, for the previous linkage analyses, blood pressure levels were “adjusted” in each gender to remove variation attributable to differences in age and measures of body size ignoring genotypic classification. Thus, the possibility was not considered that the relationship between interindividual variation in blood pressure and variation in the ACE gene may be dependent on age or measures of body size, such as what we found in the young female subjects in the present study.
With the mapping and sequencing of all human genes soon to be completed,21 considerable efforts are underway to identify and characterize the subset that influence blood pressure and its cardiovascular disease complications. Although the ACE gene has been one of the most widely and intensively studied candidate genes, a simple yes or no answer to the question of whether variation in the ACE gene influences blood pressure is not possible.22 On the basis of findings in this and other studies, we believe that the ACE gene is likely to be one of many genes whose variation has small effects on blood pressure in some contexts but not in others. For such genes, the more appropriate question may be what are the contexts in which variation in the gene does and does not influence blood pressure? Such an approach emphasizes the need to study not only variants of the genes that are candidates to influence blood pressure but also the genetic and environmental backgrounds in which they are expressed to unravel the complex causes of hypertension.
We are grateful to the staff of the Rochester Family Heart Study, Ken Weiss, and Tracy Fuller.
This study was supported by grants HL-30248, HL-39107, HL-51021, and HL-54464 from the Genetic Epidemiology Network of Arteriopathy (GENOA) of the NHLBI Family Blood Pressure Program and by funds from the Mayo Foundation.
- Received May 8, 1999.
- Revision received July 1, 1999.
- Accepted July 26, 1999.
O’Donnell C, Lindpaintner K, Larson M, Rao V, Ordovas J, Schaefer E, Myers R, Levy D. Evidence for association and genetic linkage of the angiotensin-converting enzyme locus with hypertension and blood pressure in men but not women in the Framingham Heart Study. Circulation. 1998;97:1766–1772.
Fornage M, Amos C, Kardia S, Sing C, Turner S, Boerwinkle E. Variation in the region of the angiotensin-converting enzyme gene influences interindividual differences in blood pressure levels in young white males. Circulation. 1998;97:1773–1779.
Rigat B, Hubert C, Alhenc-Gelas F, Corvol P, Soubrier F. An insertion/deletion polymorphism in the angiotensin I-converting enzyme gene accounting for half the variance of serum enzyme levels. J Clin Invest. 1990;86:1343–1346.
Morise T, Takeuchi Y, Takeda R. Angiotensin-converting enzyme polymorphism and essential hypertension. Lancet. 1994;343:125.
Duru K, Farrow S, Wang J, Lockette W, Kurtz T. Frequency of a deletion polymorphism in the gene for angiotensin converting enzyme is increased in African-Americans with hypertension. Am J Hypertens. 1994;7:759–762.
Chiang F, Lai Z, Chern T, Tseng C, Hsu K, Lo H, Tseng Y. Lack of association of the angiotensin converting enzyme gene polymorphism with essential hypertension in a Chinese population. Am J Hypertens. 1997;10:197–201.
Kiema T, Kauma H, Rantala A, Lilja M, Reunanen A, Kesaniemi Y, Savolainen M. Variation at the angiotensin-converting enzyme gene and angiotensinogen gene loci in relation to blood pressure. Hypertension. 1996;28:1070–1075.
Harrap S, Davidson H, Cornor J, Soubrier F, Corvol P, Fraser R, Foy C, Watt G. The angiotensin I converting enzyme gene and predisposition to high blood pressure. Hypertension. 1993;21:455–460.
Suzuki D, Griffiths A, Miller J, Lewontin R. Quantitative Genetics. In: An Introduction to Genetic Analyses. New York, NY: W. H. Freeman & Co;. 1986: 514–519.
Lewontin R. Genotype and phenotype. In: Fox Keller E, Lloyd E, eds. Keywords in Evolutionary Biology. Cambridge, Mass: Harvard University Press;1992: 137–144.
Celetano A, Mancini F, Crivaro M, Palmieri V, De Stefano V, Ferrara L, Di Minno G, de Simone G. Influence of cardiovascular risk factors on relation between angiotensin converting enzyme-gene polymorphism and blood pressure in arterial hypertension. J Hypertens. 1998;16:985–991.
Soubrier F. Blood pressure gene at the angiotensin I–converting enzyme locus: chronicle of a gene foretold. Circulation. 1998;97:1763–1765.