ACE Gene Polymorphism as a Risk Factor for Ischemic Cerebrovascular Disease

Methods

Participants

Case-Referent Study 1

Case-patients were 35 women and 38 men from the greater Copenhagen area in whom focal neurologic symptoms due to ischemic cerebrovascular disease developed suddenly before 50 years of age (Figure 1). The case-patients were consecutively referred for outpatient examination of defects in the coagulation system at Rigshospitalet in Copenhagen from 1989 to 1995. The criteria for diagnosis of ischemic cerebrovascular disease were ischemic stroke (focal neurologic symptoms that lasted for more than 24 hours and for which computed tomography excluded intracerebral and extracerebral hemorrhage), transient ischemic attack (focal neurologic symptoms that lasted less than 24 hours), or amaurosis fugax (transient blindness in one eye only). Seventy-one case-patients had Danish parents, one was from France, and one was not white. The referent group comprised 1454 women and 1737 men from the general population sample who were within the same age range as the case-patients (women >30 years and <54 years of age; men >24 years and <56 years of age). In a case-referent study, case-patients are compared with a sample of the general population; the sample is roughly within the same age range as the case-patients, but participants are not exactly matched for age.

Case-Referent Study 2

Case-patients were 82 women and 137 men from the greater Copenhagen area who had sudden onset of focal neurologic symptoms due to ischemic cerebrovascular disease (Figure 1). The case-patients were consecutively referred for outpatient ultrasonography of the carotid artery at Rigshospitalet from 1994 to 1996. The overlap in time between the two groups of case-patients (1989 to 1995 and 1994 to 1996) reflects the different periods in which we could recruit the two groups of case-patients. It is unlikely that the small difference in recruitment period is important for the questions asked in our study: Neither genotype frequencies nor the context in which the genotypes act would change significantly in such a short period.

The criteria for a diagnosis of ischemic cerebrovascular disease were ischemic stroke, transient ischemic attack, or amaurosis fugax, all diagnosed according to the criteria described in the preceding section. Only patients with carotid stenosis greater than 40% on the symptomatic side were included in this study; thus, case-patients had ischemic cerebrovascular disease and clinically significant atherosclerosis. All patients were white, and more than 98% had Danish parents. The referent group comprised 4273 women and 3091 men from the general population sample who were within the same age range as the case-patients (women >36 years and <80 years of age; men >42 years and <80 years of age).

Cross-Sectional Study

A general population sample comprising 9203 persons was assembled during the Copenhagen City Heart Study from 1991 to 1994. The sample, which consisted of an almost equal number of women and men stratified for age (from 20 to ≥80 years), was drawn randomly from the Copenhagen Central Population Register with the aim of obtaining a representative sample of the adult Danish general population (Figure 1) [16]. Less than 1% of the sample was nonwhite, and 98.8% had Danish citizenship (that is, they were essentially of Danish parentage). Persons with previous sudden onset of focal neurologic symptoms due to ischemic cerebrovascular disease were identified by an experienced neurologist on the basis of history and by review of all hospital admissions and diagnoses (obtained from the Danish National Hospital Discharge Register) and, if necessary, medical records from hospitals or general practitioners.

The criteria for diagnosis of ischemic cerebrovascular disease were ischemic stroke, transient ischemic attack, or amaurosis fugax [17]; the diagnostic criteria for these conditions were almost identical to those described in the section on case-referent study 1. Of 107 patients with stroke, 70 had definite ischemic stroke, 9 had intracerebral hemorrhage, and 28 had strokes that could not be classified because of the lack of a computed tomographic scan. However, given the ratio of cases of intracerebral hemorrhage to all cases with a definite diagnosis (9 of 79), 3 ([9/79] × 28) of the 160 patients with ischemic cerebrovascular disease probably had intracerebral hemorrhage. Information on ischemic cerebrovascular disease was available for 7393 persons; of these, 67 women and 93 men had ischemic cerebrovascular disease and 4077 women and 3156 men did not have disease.

Laboratory Methods

Cholesterol and triglyceride levels were determined enzymatically (CHOD-PAP, GPO-PAP, Boehringer Mannheim, Mannheim, Germany). High-density lipoprotein (HDL) cholesterol was measured in the supernatant after precipitation of apolipoprotein B-containing lipoproteins (Boehringer Mannheim). Lipoprotein(a), apolipoprotein A1, and apolipoprotein B levels were measured by using end-point turbidimetry with commercially available antisera (rabbit antihuman lipoprotein[a], DAKO A/S, Glostrup, Denmark; sheep antihuman apolipoprotein A1 and apolipoprotein B, Boehringer Mannheim). Plasma fibrinogen levels were measured kinetically (Fibrinogen Kinetic, Boehringer Mannheim). All analyses in all patients were performed by use of identical diagnostic kits and very similar autoanalyzers and were done at the examination facility of the Copenhagen City Heart Study, Rigshospitalet; the Department of Clinical Biochemistry, Rigshospitalet; or the Department of Clinical Biochemistry, Herlev University Hospital. Lipoprotein(a), apolipoprotein A1, apolipoprotein B, and fibrinogen levels were measured at only one of the departments on only one autoanalyzer. Precision and accuracy of all analyses were continually tested by using internal departmental controls; accuracy of cholesterol, triglyceride, and HDL cholesterol levels was also monitored by a nationwide external quality control program. Persons who read the biochemical results were blinded to the disease status of all participants.

Other Analyses

Body mass index was calculated as weight in kilograms divided by height in squared meters. Blood pressure was considered elevated if 1) the systolic blood pressure in mm Hg was larger than both 145 mm Hg and the combination of 110 and the participant's age, 2) the diastolic blood pressure (phase 5) was greater than 100 mm Hg regardless of age [16], or 3) the participant was receiving treatment with at least one antihypertensive drug. Participants were classified as diabetic if they received treatment for diabetes. Smokers were considered to be persons who currently smoked; former smokers had stopped smoking at least the day before the examination. The presence of hypertension and diabetes mellitus and data on smoking habits were based on cross-sectional information and measurements at the time of examination; no information on the duration of these risk factors was used.

ACE Gene Polymorphism

Total genomic DNA was extracted from whole blood, as described elsewhere [18]. The insertion-deletion polymorphism of 287 base pairs in intron 16 of the ACE gene was identified by conventional polymerase chain reaction (PCR) by using two primers flanking the site of the insertion [19]. Fragments of about 190 base pairs (D allele) and 490 base pairs (I allele) were separated on a 2% agarose gel. All samples that seemed homozygous for the D allele were subjected to a second PCR amplification with an insertion-specific primer to check for misclassification resulting from a potential preferential amplification of the smaller D allele [10]. Four percent to 5% of persons with the ID genotype were initially misclassified as having the DD genotype, but this error was corrected before statistical analysis. All PCR analyses were performed at the Department of Clinical Biochemistry, Herlev University Hospital, by the same scientist and two technicians, both of whom were supervised by the scientist. Persons who read the PCR results were blinded to the disease status of the participants from the Copenhagen City Heart Study.

Statistical Analysis

Statistical analyses were done by using the SPS program [20, 21]. To test for differences in various characteristics, we used the Student t-test for continuous variables and the Pearson chi-square test and the likelihood ratio statistic for categorical variables. To approach a normal distribution, plasma triglyceride and lipoprotein(a) levels were transformed logarithmically (log₁₀) before the t-tests were conducted. Relative frequencies of alleles were estimated by gene counting. We calculated odds ratios comparing relative genotype frequencies of case-patients with frequencies of all persons in the general population sample and of persons with and without ischemic cerebrovascular disease in the general population sample. The likelihood ratio statistic was used as a test of independence. Logistic regression analysis with forced entry was performed to describe the predictive value of ACE genotype and conventional cardiovascular risk factors for ischemic cerebrovascular disease. Because apolipoprotein A1 and B levels were closely associated with HDL cholesterol and plasma cholesterol levels, respectively, only the latter two covariates, together with age, diabetes mellitus, hypertension, and smoking habits, were entered into these models. For the cross-sectional study, body mass index and plasma levels of triglycerides, plasma lipoprotein(a), and fibrinogen were also entered into the models. To approach linearity on the logit scale [22], HDL cholesterol levels were transformed logarithmically (log₁₀). When age and all other independent covariates were controlled for, the contribution of ACE genotype in predicting ischemic cerebrovascular disease was expressed as an odds ratio with 95% CIs. Homogeneity of the ability of conventional cardiovascular risk factors to predict ischemic cerebrovascular disease among ACE genotype classes was tested for all cardiovascular risk factors combined; in addition, all possible two-way interaction terms between ACE genotype and each of the conventional cardiovascular risk factors were introduced, one at a time, into the logistic regression model. The likelihood ratio test was used to determine whether covariates or interaction terms contributed significantly to the logistic regression models. A two-sided P value less than 0.05 was considered statistically significant. Correction for multiple comparisons was not performed in any of the analyses.

Role of Funding Source

The sponsors of the study are public or nonprofit organizations and support science in general. They had no role in gathering, analyzing, or interpreting the data and had no right to approve or disapprove the submitted paper.

Results

Participant Characteristics

In case-referent study 1, plasma cholesterol levels and presence of diabetes mellitus were significantly increased among women and men who developed ischemic cerebrovascular disease before 50 years of age compared with persons in the general population sample (Table 1 , Table 5). In addition, HDL cholesterol levels were lower and the number of smokers was greater in female case-patients compared with persons in the referent group.

In case-referent study 2, age, plasma cholesterol levels, and prevalence of hypertension were increased and HDL cholesterol levels were decreased in women and men who had ischemic cerebrovascular disease and carotid stenosis greater than 40% compared with persons in the general population sample. The presence of diabetes mellitus was increased and the number of smokers was decreased in male case-patients compared with referents.

In the cross-sectional study, age, plasma triglyceride levels, and presence of diabetes mellitus were increased in men and women who had ischemic cerebrovascular disease compared with those who did not have this disease. In male case-patients, levels of HDL cholesterol and apolipoprotein A1 were decreased and plasma levels of fibrinogen and presence of hypertension were increased compared with persons who did not have ischemic cerebrovascular disease. Apolipoprotein B levels in female case-patients were increased compared with levels in persons who did not have disease.

Relative Frequencies of Genotypes and Alleles

Relative frequencies of genotypes and alleles are listed in Table 2 and (Table 6); with one exception, these distributions did not differ significantly from those predicted by the Hardy-Weinberg equilibrium. The exception was seen with female case-patients in case-referent study 2 (P < 0.001). Almost no statistically significant differences in the relative frequencies of genotypes or alleles were seen between male or female case-patients and the referents or persons without disease in the three studies separately or combined. One exception was the relative frequency of the ACE genotype in women in case-referent study 2 (P = 0.002). This difference occurred because there were fewer heterozygous case-patients and more case-patients homozygous for both alleles compared with persons in the referent group; however, relative frequencies of the ACE allele did not differ between case-patients and referents (P > 0.2).

Risk for Ischemic Cerebrovascular Disease

In none of the studies (alone or combined) did calculation of odds ratios show that ACE genotype significantly contributed to the prediction of ischemic cerebrovascular disease (Table 2 , Table 6). The borderline-significant odds ratios among women in case-referent study 2 in the DD compared with ID and II model (odds ratio, 1.59 [95% CI, 1.01 to 2.50]; P = 0.05) and DD and ID compared with II model (odds ratio, 0.63 [CI, 0.40 to 1.01]; P = 0.06) are the result of the higher frequency of both types of homozygous persons relative to that of heterozygous persons among case-patients compared with referents. The erratic pattern of positive and negative trends shown in Table 2 and Table 6 indicate that there is little or no association between ACE gene polymorphism and ischemic cerebrovascular disease.

Mean age did not differ among persons with DD, ID, or II genotypes in any of the groups shown in Table 1 and (Table 5) (data not shown). Furthermore, in both men and women, the relative frequency of the DD genotype as a function of age did not change in patients 20 years of age to 80 years of age or older in the Copenhagen City Heart Study cohort [9], in patients 24 to 56 years of age who had ischemic cerebrovascular disease before 50 years of age (data not shown), or in patients 36 to 80 years of age who had ischemic cerebrovascular disease and carotid stenosis greater than 40% (data not shown). Accordingly, ACE genotype was not associated with age and did not seem to affect longevity [9]; it is therefore unlikely that age is a confounder in the relation between ACE genotype and ischemic cerebrovascular disease. Adjustment for age and other cardiovascular risk factors did not substantially change the general lack of an association between ACE gene polymorphism and ischemic cerebrovascular disease; this adjustment also provides evidence that age is probably not a confounder.

To examine the contribution of ACE genotype in predicting ischemic cerebrovascular disease with adjustment for the contribution of age and known cardiovascular risk factors, multiple logistic regression analysis was performed separately for the two case-referent studies, the cross-sectional study, and the three studies combined (Table 3). With one exception, the odds ratio for ACE genotype as a predictor of ischemic cerebrovascular disease was not significant in any study separately or combined, regardless of the coding scheme used for genotypes. The exception was seen in female case-patients in case-referent study 2 for the DD compared with ID and II model (odds ratio, 1.77 [CI, 1.04 to 3.02]; P = 0.04) and the DD compared with ID compared with II model (odds ratio for DD compared with II, 1.12 [CI, 0.60 to 2.09]; odds ratio for ID compared with II, 0.46 [CI, 0.24 to 0.88]; P value for likelihood ratio test = 0.01). This finding occurred because of the relatively few heterozygotes among case-patients relative to the number of heterozygotes among referents (Table 3) (odds ratio for ID compared with II, 0.46 [CI, 0.24 to 0.88]).

On multiple logistic regression analysis that combined all women or men, increase in age and plasma cholesterol level, decrease in HDL cholesterol level, and presence of diabetes mellitus and hypertension (but not smoking) significantly contributed to the prediction of ischemic cerebrovascular disease (Table 4; data on age not shown). By comparing the likelihood value for the model that included all participants (n = 9361) with the combined likelihood value for the three genotype models (persons with the DD, ID, and II genotypes, respectively), we tested for an interaction between ACE genotype and the five covariates shown in Table 4 plus age. No statistically significant evidence showed an interaction between genotype and the covariates or age in women (P > 0.10) or men (P > 0.2). When each of the five covariates shown was tested individually for interaction with genotype in predicting ischemic cerebrovascular disease in the logistic regression model, no interaction terms were significant in women or men.

Discussion

We tested the hypothesis that the ACE gene polymorphism may represent a susceptibility mutation for ischemic cerebrovascular disease. No statistically significant evidence to support this hypothesis was provided by the two case-referent studies, the cross-sectional study, or the three studies combined (constituting a total of 452 persons with ischemic cerebrovascular disease and 9043 “controls”). Predictors of ischemic cerebrovascular disease in our study were either well-established (age, diabetes mellitus, and hypertension) or suspected (cholesterol and HDL cholesterol levels) risk factors for ischemic stroke.

Previous reports on ACE gene polymorphism and ischemic cerebrovascular disease have shown some positive associations. For example, among 210 Italian persons, the DD genotype was most common in survivors of ischemic stroke compared with controls [13]; among 499 British persons, the D allele was associated with an increased risk for early death from acute cerebral infarction [14]; among 332 Japanese persons, the D allele or DD genotype was more common among hypertensive patients who had ischemic stroke than among hypertensive patients who did not have stroke or among normotensive controls [15]; and among 238 British persons, the D allele or DD genotype was more common among persons who had lacunar stroke than among controls [12]. Nevertheless, two of the studies [12, 14] found no evidence that the DD genotype or D allele was more common in the total group of patients who had stroke than in controls.

Several factors may explain the discrepancies between the four previous studies and our study. First, the ACE D allele may be in linkage disequilibrium, with a mutation acting as a susceptibility mutation for ischemic cerebrovascular disease in the Italian, Japanese, and British populations but not in the Danish population. Linkage disequilibrium may result from a short physical distance on chromosome 17 between the ACE D allele and a potential susceptibility mutation, such that the two mutations are inherited together in persons with ischemic cerebrovascular disease. This would have been a likely explanation if the association was found in Japanese persons but not in any of the three studies done in white persons.

Second, the lack of an association between the D allele and ischemic cerebrovascular disease in our study could be a chance observation, and an association might really exist. It is hard to believe that our negative finding is caused by a lack of statistical power because our study is by far the largest study published to date on this subject. Alternatively, selection bias could be important in the three case-patient groups, the referent groups, or the persons in the cross-sectional study without ischemic cerebrovascular disease. Because the three patient groups (who were categorized according to essentially identical diagnostic criteria) consisted of different types of patients (young survivors of ischemic cerebrovascular disease, patients referred for carotid ultrasonography after manifestations of ischemic cerebrovascular disease, and survivors of ischemic cerebrovascular disease in a general population sample studied cross-sectionally), it is difficult to imagine that the same type of selection bias would operate in all three case-patient groups to cause the consistently negative results in both men and women. Theoretically, the referent or nondiseased groups could also be biased; this is unlikely, however, because the sample was drawn randomly from the Danish general population. In the two case-referent studies, a few case-patients (relative to the number of referents) were part of the referent group because of the study design; the inclusion of some case-patients in the referent groups in these studies dilutes the associations between ACE gene polymorphism and ischemic cerebrovascular disease. However, the small percentage of case-patients in the referent groups should not cause much of a decrease in these associations. In the cross-sectional study, case-patients were compared only with persons who did not have disease. In this latter study, as well as in the combined studies, misclassification of case-patients as persons without disease cannot be completely ruled out. This potential misclassification bias could reduce an association between ACE genotype and ischemic cerebrovascular disease. All participants in our study were of similar ethnic background; more than 98% had Danish citizenship and thus were essentially of Danish parentage.

Third, all of the associations found in the four previous studies [12-15] may be chance observations. An argument against this interpretation is the fact that all four studies found positive associations. On the other hand, this interpretation is supported by the facts that 1) the studies found that the D allele is associated with different end points [ischemic stroke, early death from acute cerebral infarction, hypertensive stroke, and lacunar stroke], 2) no plausible mechanism exists to explain the association, and 3) a recent meta-analysis of studies of associations between the ACE gene polymorphism and myocardial infarction [11] suggested that publication bias accounted for the positive associations in the smaller studies.

A surprising finding in our study was the high number of both types of homozygotes (DD and II genotypes) relative to the number of heterozygotes (ID genotype) among women with ischemic cerebrovascular disease in case-referent study 2 (P = 0.002). In Table 2 , Table 6, this finding is illustrated by an odds ratio of 1.59 for ischemic cerebrovascular disease when the DD genotype was compared with the ID and II genotypes combined; the odds ratio was 0.63 when the DD and ID genotypes combined were compared with the II genotype. This finding may represent a true effect, but it was not observed in our two other studies of women or among the three groups of men. In addition, no such distribution has been reported previously. We believe that this observation represents a chance occurrence rather than a true effect for two reasons: 1) The finding would imply that persons with the DD genotype and those with the II genotype are at an increased risk for ischemic cerebrovascular disease relative to persons with the ID genotype and 2) the finding contrasts with all previous observations on the ACE gene polymorphism and risk for ischemic cerebrovascular disease. In addition, the pattern of positive and negative association shown in Table 2 and Table 6 is erratic, further suggesting that there is little or no association between ACE gene polymorphism and ischemic cerebrovascular disease.

Although the ACE D allele does not seem to be associated with ischemic cerebrovascular disease, it may still be associated with myocardial infarction [8] or other manifestations of ischemic heart disease. In another study [9], however, we tested this hypothesis by using a sample assembled in the Copenhagen City Heart Study and 947 case-patients with verified ischemic heart disease in a case-referent study (n = 10 150) and by using a cross-sectional study (also called a retrospective cohort study) (n = 7263) of the general population sample. We found no statistically significant difference in the development of myocardial infarction or any other manifestations of ischemic heart disease between genotype classes of the ACE gene polymorphism in women or men in either of these studies. Furthermore, genotype had no effect on longevity. One limitation of the cross-sectional part of this previous study (as well as the cross-sectional part of our present study) was that we performed genotyping only on persons who gave blood at the third examination in 1991 through 1994 and not at the first examination in 1976 through 1978 [9]. However, this was not a problem for the case-referent studies.

In support of our finding, the second largest study to date on this issue-the Physician's Health Study, which prospectively examined approximately 3600 North American male physicians [10] -also found no evidence of an association between the ACE D allele and ischemic heart disease. Nevertheless, a recent meta-analysis [11] (which included the Physician's Health Study [10] but not our previous study [9]) of 15 studies comprising 8873 persons found support for an association between the ACE D allele and myocardial infarction. As mentioned above, however, this meta-analysis also suggested publication bias for the positive results in the smaller studies. Taken together, the combined evidence does not support the notion that the ACE D allele is as a risk factor for ischemic heart disease or ischemic cerebrovascular disease.

Several studies have shown an association between the D allele and higher plasma ACE levels or ACE activity in a codominant pattern [2-7]. It is therefore conceivable that the ACE gene polymorphism (either on its own or through linkage disequilibrium with another mutation) increases the plasma ACE level but that this higher ACE level does not affect the risk for ischemic cerebrovascular disease or ischemic heart disease. Alternatively, such an elevation in plasma ACE levels may contribute to risk for ischemic cerebrovascular disease or ischemic heart disease; however, our present study and our previous study [9] were too small to detect such effects.

In conclusion, we found no evidence of a statistically significant difference in the development of ischemic cerebrovascular disease between genotype classes of the ACE gene polymorphism in women or men, regardless of whether the three studies were examined separately or together.

Dr. Gronholdt: Department of Vascular Surgery, Rigshospitalet, National University Hospital, DK-2100 Copenhagen, Denmark.

Dr. Jensen: Department of Cardiology, Hvidovre University Hospital, DK-2650 Hvidovre, Denmark.