Genetic factors play a major role in susceptibility to coronary atherosclerosis, the leading cause of death in the Western Hemisphere (1). Initial clues implicating genes in susceptibility to coronary artery disease (CAD) and myocardial infarction (MI) were derived from epidemiological and twin studies (2). Slack and Evans (2) were among the first to show that the risk of death from CAD was five- to seven-fold greater in the first-degree relatives of individuals with CAD. Similarly, Marenberg et al. (3) showed that the relative hazard of death from CAD was 8 times higher in a monozygotic twin and 3.8 times higher in a dizygotic twin if one’s twin died from premature CAD. The basis for the interindividual differences in the genetic susceptibility to atherosclerosis is variations in the nucleotide sequence of the genes, which are referred to as polymorphisms. A polymorphism can influence the steady state plasma or tissue levels of the gene product (4) or affect the response of a phenotype to therapeutic interventions (5- 6). It is estimated that one-fourth to one-half of the total variance in plasma total cholesterol, high density lipoprotein-cholesterol (HDL-C) and triglyceride levels are due to genetic polymorphisms (7).

Despite the well-established role of genes in susceptibility to coronary atherosclerosis, identification of the responsible genes has been problematic. The commonly used technique of case-control polymorphism (allelic) association, whereby the frequency of the variants (alleles) of a candidate gene are compared in cases and controls, is notorious for spurious results (8). This is exemplified by the conflicting data on the possible role of a common insertion/deletion (I/D) polymorphism in the angiotensin-1 converting enzyme (ACE) gene in susceptibility to CAD and MI (4). A better approach is a longitudinal study of a well-characterized cohort, which does not have the inherent weaknesses of case-control allelic association studies (8).

Interest in the role of ACE I/D genotypes in the susceptibility to CAD stems from observations that the I/D genotypes account for half of the variability of the plasma, cellular and tissue levels of ACE (9- 11) and from compelling experimental data implicating ACE in vascular homeostasis and atherosclerosis (12). To examine further the role of ACE in atherosclerosis, we analyzed data from a longitudinal study of a well-characterized cohort, the Lipoprotein and Coronary Atherosclerosis Study (LCAS) population (13), for possible associations of the I/D genotypes with baseline demographic, biochemical and coronary angiographic variables and the response of biochemical and angiographic variables to treatment with the cholesterol lowering agent fluvastatin, a 3-hydroxy-3-methylglutaryl coenzyme A (HMG-CoA) reductase inhibitor.

All subjects provided written informed consent and the study was approved by the Institutional Review Board. The study population was comprised of 429 subjects (349 men and 80 women). The design and primary results of LCAS have been published (13- 15). In brief, patients between 35 and 75 years who had at least one coronary lesion causing 30% to 75% diameter stenosis and low density lipoprotein-cholesterol (LDL-C) of 115–190 mg/dl despite diet were randomized to 40 mg fluvastatin daily or placebo. Total cholesterol, LDL-C, HDL-C, triglyceride and apolipoprotein (apo) levels were measured in all subjects at baseline and throughout the study. Subjects underwent quantitative coronary angiography upon enrollment (n = 429) and 2.5 years following randomization (n = 340). The primary end point was within-subject per-lesion change in the minimum lumen diameter (MLD) of qualifying lesions, defined by MLD ≥25% of the reference lumen diameter at baseline and MLD ≥0.8 mm less than the reference lumen diameter at either baseline or follow-up. Subjects were also categorized as having definite progression, definite regression or mixed angiographic change. Definite progression was defined as ≥1 qualifying lesion with MLD decrease ≥0.4-mm, including new total occlusions and no qualifying lesion with MLD increase ≥4 mm. Definite regression was defined as ≥1 qualifying lesion with MLD increase ≥0.4 mm, no qualifying lesion with MLD decrease ≥0.4 mm and no new total occlusion. Subjects that showed neither definite progression nor definite regression were classified as having mixed change. Clinical events monitored were definite or probable MI, unstable angina requiring hospitalization, percutaneous transluminal coronary angioplasty, coronary artery bypass grafting, all-cause mortality and cardiovascular mortality.

Laboratory personnel who had no knowledge of the clinical data performed the genotyping and two individuals read the genotypes. Angiotensin-1 converting enzyme genotypes were determined using polymerase chain reaction (PCR) (16) in 364/429 (85%) subjects. In 54 subjects DNA samples were not available, and in the remaining 11 subjects, the PCR reactions did not work. Each PCR reaction contained 100 ηg of DNA template, 0.125 μM of each primer, 200 uM of 4dNTPs, 1 U of Taq DNA polymerase and 1.5 mM of MgCl₂ and 5% dimethyl sulfoxide (DMSO). DNA was amplified for 30 cycles, each comprised of denaturation at 94°C for 1 min, annealing at 58°C for 1 min, extension at 72°C for 1 min with a final extension time of 3 min. To avoid possible mistyping of the genotypes, in addition to using DMSO (17), those with the DD genotype were reanalyzed by PCR using a set of insertion-specific primers (18). The PCR products were separated by electrophoresis on 2% agarose gel and identified by ethidium bromide staining.

Continuous variables were expressed as mean ± SD (with the exception of lesion-specific MLD which was expressed as mean ± SE) and their differences among genotypes were compared by analysis of variance (ANOVA). Variables that were unsuited for ANOVA because of inequality of variance were analyzed by the Kruskal-Wallis test. Distribution of categorical variables among genotypes was compared using chi-square or Fisher’s exact tests. To determine the association of ACE genotypes with response to fluvastatin treatment, mean changes in lipid levels and MLD among the genotypes were compared using ANOVA. The ANOVA performed in this study utilizes Yates’ weighted squares of means technique, which has the ability to obtain the proper sums of squares for testing main and interaction effects on data with disproportional unequal subclass sizes. In addition, analyses were repeated using the following assigned weights to the genotypes according to their known biological association with plasma, cellular and tissue levels of ACE (9- 11): DD = 2, ID = 1.5, II = 1. The results of weighted analysis are reported when they did not differ substantially from the results of unweighted analysis, otherwise both results are reported. Statistical analysis was performed using STATA, version 5.0, (Stata Corporation, College Station, Texas).

The products of D and I alleles were identified by the presence of 190 and 490 base pair bands on agrose gel electrophoresis, respectively. Repeat PCR using insertion-specific primers did not identify genotype mistyping, which may be due to the routine use of DMSO in the PCR reaction (17). The frequencies of D and I alleles were 0.52 and 0.48, respectively. The frequencies of genotypes are shown in (Table le1).

The baseline characteristics of subjects according to genotype are shown in (Table le1). The mean systolic and diastolic blood pressure, mean MLD and the number of subjects with ≥1 qualifying lesion or total occlusions were not significantly different among the genotypes. Subjects with the ID genotype had a lower proportion of previous MI (33%) as compared with DD (50%) and II (51%) genotypes (p = 0.006). However, this observation is not in accord with the genetic gradient or the known biological association of the ACE I/D genotypes with plasma and tissue levels of ACE (9- 11).

(Table le2)shows the baseline and final values and percent change of lipid levels in the placebo and fluvastatin treatment groups according to genotype. There was a significant genotype-by-treatment interaction for total cholesterol (p = 0.018), LDL-C (0.005), and apo B (p = 0.045). Subjects with the DD genotype, as compared with subjects with the ID and II genotypes, had a greater reduction in plasma levels of total cholesterol (19% vs. 15% vs. 13%), LDL-C (31% vs. 25% vs. 21%) and apo B (23% vs. 15% vs. 12%), respectively, in response to fluvastatin therapy. The magnitude of the reduction in total cholesterol, LDL-C and apo B followed the genetic gradient and the biological effect of the I/D genotypes on tissue, cellular and plasma levels of ACE (DD>ID>II) (9- 11).

A total of 63 subjects (25 in the placebo group and 38 in the fluvastatin group) were on ACE inhibitor therapy at baseline or during the study. The distribution of subjects taking ACE inhibitors among the ACE genotypes was not significantly different (Table le1). When controlled for treatment with ACE inhibitors, significant interactions between ACE genotypes and response of the plasma cholesterol (p = 0.003), LDL-C (p = 0.014) and apo B (p = 0.041) to treatment with fluvastatin persisted. Similarly, exclusion of subjects who were on ACE inhibitors did not affect the overall results. In addition, the distribution of genotypes did not differ significantly in this group as compared with the remainder of the subjects, and treatment with ACE inhibitors had no significant effect on the baseline plasma lipid levels among the subjects with different ACE genotypes (data not shown).

Angiographic results (n = 308) are shown in (Table le3). Concordant with the influence of ACE I/D genotype on the response of plasma LDL-C levels to fluvastatin, subjects with the DD genotype were less likely to have definite progression (14%) and more likely to have definite regression (24%) compared with those with ID (32% progression and 17% regression) or II (33% progression and 3% regression) genotypes (p = 0.023). Mean MLD also increased in those with the DD genotype (increase: 0.05 ± 0.20 mm) indicative of regression, whereas it decreased in those with the ID (decrease: 0.04 ± 0.26 mm) or II (decrease: 0.01 ± 0.26) genotypes, which indicate progression. Lesion-specific MLD also followed the same trend. These differences did not reach statistical significance.

Of the 364 subjects, 52 (14%) had clinical events. The distribution of the clinical events among the genotypes was not significantly different (Table le3).

The main finding of this study is that ACE I/D genotypes influence the response of plasma lipids to treatment with fluvastatin. Subjects with the DD genotype had a greater reduction in plasma levels of LDL-C than those with ID or II genotypes. Similarly, the reductions in plasma levels of total cholesterol and apo B were also greater in subjects with the DD genotype. The interaction between the ACE I/D genotypes with the response of plasma lipids to fluvastatin therapy exhibited a gene-dose effect and was in accord with the known biological association of the I/D genotypes with plasma, cellular and tissue levels of ACE (DD>ID>II) (9- 11). Consistent with a greater reduction in LDL-C in response to fluvastatin therapy, subjects with the DD genotype also exhibited a higher rate of regression and a lower rate of progression of coronary atherosclerosis as determined by quantitative coronary angiography. Another main finding of this study is that the ACE genotypes were not associated with baseline severity of CAD, progression or regression of coronary atherosclerosis in the placebo group or clinical events.

In the entire LCAS population, treatment with fluvastatin reduced the mean LDL-C and increased the mean HDL-C by 27% and 6%, respectively (13). The mean MLD of qualifying lesions showed significantly less progression in the fluvastatin group. There were also fewer subjects with the progression and more subjects with the regression of coronary atherosclerosis in the fluvastatin treatment group as compared with the placebo group (13). A similar trend, which was not statistically significant, was also observed in clinical event rates (13). The greatest angiographic and clinical benefit was observed in subjects with low HDL-C (15).

The observed association between ACE I/D genotypes and the response of plasma lipids to treatment with fluvastatin was unexpected and requires further validation in additional data sets. Nonetheless, it is likely a true association for the following reasons. The strength of the observed association (p = 0.005) is considered strong (8,19). The concordant associations with several variables, namely, three related plasma lipid components and angiographic indexes of progression/regression also favor a true association (8). In addition, the presence of a gene-dose effect (genetic gradient), which is consistent with the association of the ACE I/D genotypes with plasma, cellular and tissue levels of ACE (biological gradient) (9- 11) further supports a true association. Furthermore, the prospective nature of this longitudinal study and the detailed characterization of the subjects reduce the chance of a spurious association (8).

A total of 108 (including 79 subjects in whom angiographic data are available) in the LCAS were also treated with cholestyramine. To exclude the possible confounding effect of cholestyramine therapy on the results of this study, the distribution of ACE genotypes among subjects treated with or without cholestyramine was compared and did not differ significantly (data not shown). In addition, possible genotype-by-treatment (with cholestyramine) interactions were analyzed and no significant interactions were detected (data not shown). Regarding the possible confounding effect of patient compliance, we note that the compliance rate with fluvastatin treatment, defined as medication compliance of at least 80%, in the entire LCAS population was 95%. Therefore, it is unlikely that the overall results would be significantly affected by the presence of possible unexpected differences in the compliance rates among the subjects with different ACE genotypes.

The design of LCAS (and hence, the choice of end points) had been determined prior to the decision to perform genetic analyses and therefore, this study is a secondary data analysis. In addition, no adjustment was made for multiple testing as the examination of each variable for the possible existence of a significant genotype-by-treatment interaction is considered a form of multiple unrelated hypotheses (20). According to J.W. Tukey (20), it is only mandatory to adjust the significance level for multiple hypotheses when the hypotheses being tested are not fully independent of one another (20). In view of multiple testing, significant p values (p < 0.05) should be regarded as potential associations.

As it is true for all association studies, the results of this study require confirmation in additional data sets. In this regard, the preliminary analysis of a similar study showed no association between the plasma lipid levels and response to treatment with pravastatin (21). However, patients with the DD genotype showed a blunted effect of pravastatin therapy on coronary atherosclerosis despite identical plasma lipid response (21).

We could not determine the ACE genotypes in 65 of the 429 subjects due to reasons of either unavailability of DNA samples or PCR reactions not working. This might have created unequal treatment sample sizes among the placebo and treatment groups. The loss of genotyping data is essentially random and is unlikely to account for the observed genotype-treatment interactions. In addition, the statistical methods used in this study takes into account disproportional unequal subclasses sizes.

This study shows a gene-treatment interaction without exploring the responsible mechanism(s). It is also not known whether treatment-genotype interaction is specific for fluvastatin and whether it is common to other HMG-CoA reductase inhibitors remains to be explored. Existing data do not provide a clear biologically plausible mechanism to explain the observed association. However, they indicate complex interactions between angiotensin II and bradykinin, the products of ACE enzymatic activity and LDL-C peroxidation as well as the activity of the HMG-CoA reductase inhibitors (22- 23). It is well-established that angiotensin II increases not only the macrophage-mediated oxidation of LDL (24), but also the cellular uptake of oxidized LDL by the scavenger receptors (25- 26). In contrast, fluvastatin inhibits LDL oxidation and stimulates the removal of LDL from the circulation (27). It is intriguing to speculate that the differences in the plasma, cellular and tissue concentrations of ACE (9- 11), angiotensin II and bradykinin in subjects with different genotypes affect LDL peroxidation, removal of the oxidized LDL by the macrophages, the rate of receptor-mediated uptake of the LDL or the pharmacokinetics of fluvastatin itself.

Although this study is the first to show an interaction between ACE I/D genotypes and the response of plasma lipids to therapy, interactions between genetic polymorphisms and drug efficacy have been recognized for many years (28). The classic example is that of cytochrome P450 isozymes, which determine interindividual variations in the metabolism of many drugs in the liver, including the HMG-CoA reductase inhibitors (28). With regard to lipids metabolism, polymorphisms in genes coding for apo B, apo A1/CIII and apo E have been shown to influence the response of plasma lipids to treatment with gemfibrozil (29) and probucol (30), respectively. Similarly, polymorphisms in the genes coding for cholesteryl ester transfer protein, apo E, apo A-1, apo A-IV have been associated with response of plasma lipids to dietary interventions (5- 6,31). Other examples of gene-treatment interaction include polymorphisms in the neurotransmitter sertonin type IIA receptor and alpha-adducin genes which are associated with the clinical response to antipsychotic drugs in schizophrenic patients (32) and to diuretics in hypertensive patients (33), respectively. Thus, genetic polymorphisms, in part, form the basis for interindividual variations in the response to drug therapy.

In summary, the results of this prospective study show that the I/D variants of the ACE gene are associated with the response of plasma lipids and coronary atherosclerosis to treatment with fluvastatin. Subjects the DD genotype exhibited a greater reduction in LDL-C, a higher rate of regression and a lower rate of progression of coronary atherosclerosis.

We would like to express our sincere gratitude to Ms. Kerrie Jara for critical review of the manuscript and helpful suggestions.