Author + information
- Received August 27, 2002
- Revision received December 3, 2002
- Accepted December 26, 2002
- Published online June 4, 2003.
- Stefan Blankenberg, MD*,‡,* (, )
- Hans J Rupprecht, MD*,
- Christoph Bickel, MD*,
- Xian-Cheng Jiang, PhD§,
- Odette Poirier, PhD‡,
- Karl J Lackner, MD†,
- Jürgen Meyer, MD, FACC*,
- François Cambien, MD‡,
- Laurence Tiret, PhD‡,
- ↵*Reprint requests and correspondence:
Dr. Stefan Blankenberg, INSERM U 525, Faculté de Médecine Pitié-Salpêtrière, 91 bld de l’Hôpital, 75634 Paris cedex 13, France.
Objectives We sought to evaluate the association between cholesteryl ester transfer protein (CETP) genotypes and the risk of future cardiovascular mortality in patients with coronary artery disease (CAD).
Background Polymorphisms of the CETPgene influence CETP activity and high-density lipoprotein (HDL) cholesterol concentration and might affect the long-term prognosis and response to statin therapy in patients with CAD.
Methods We used serum samples and deoxyribonucleic acid collected at baseline from a prospective cohort of 1,211 patients with CAD prospectively followed up (median follow-up of 4.1 years), 82 of whom experienced a fatal cardiovascular event. The CETP/C-629Aand I405Vpolymorphisms, CETP activity, and HDL cholesterol were determined.
Results Patients carrying the −629Aallele had significantly lower CETPactivity and higher HDL cholesterol levels. There was a significant association between this polymorphism and the risk of future cardiovascular death. Mortality decreased from 10.8% in CChomozygotes to 4.6% in CAheterozygotes and 4.0% in AAhomozygotes (p < 0.0001). This association was independent of potential confounders, particularly HDL cholesterol and CETP activity levels. The clinical benefit of statin therapy was restricted to CChomozygotes, in whom cardiovascular mortality was divided by half (p = 0.01 for treatment × genotype interaction). Similar trends were observed with the CETP/I405Vpolymorphism, but these effects seemed to be mainly the consequence of linkage disequilibrium with the CETP/C-629Apolymorphism.
Conclusions In patients with CAD, the CETP/−629Aallele had a strong protective effect on future mortality from cardiovascular causes, independent of its role on HDL cholesterol and CETP activity levels. Additionally, this common polymorphism appeared to predict which patients with CAD will experience a survival benefit from statin therapy.
The inverse relationship between serum high-density lipoprotein (HDL) cholesterol and ischemic heart disease has increasingly been recognized in numerous clinical and epidemiologic studies (1). Cholesteryl ester transfer protein (CETP) distributes cholesteryl esters and triglycerides between lipoproteins and is a key enzyme in reverse cholesterol transport (RCT) and HDL metabolism (2,3). However, whether CETP acts in a pro- or anti-atherogenic fashion still remains controversial (4–6).
A major deficiency of CETP activity due to rare genetic mutations has been reported to be associated with an increased prevalence of coronary artery disease (CAD) in Japanese men, despite high levels of HDL cholesterol (7). A common polymorphism of the CETPgene located in intron 1 (TaqIB) has consistently been shown to influence CETP activity and HDL cholesterol levels (8–12). However, this polymorphism is unlikely to be functional by itself and rather reflects the effect of one or several functional variants of the CETPgene (13). In particular, the TaqIB polymorphism was found to be in nearly complete association with the CETP/C-629Apolymorphism located in the promoter (13), which was further shown to influence gene expression and CETP activity (14).
Patients with low plasma HDL cholesterol receive a clinical benefit from statin therapy (15). Because of the strong impact of CETPgene variations on HDL cholesterol, it might be hypothesized that CETPpolymorphisms modulate the effect of statin therapy on HDL cholesterol and possibly on CAD risk. Supporting this hypothesis, it was recently shown that the TaqIB polymorphism was associated with different effects of pravastatin treatment on progression of coronary atherosclerosis (16).
In light of the previous findings, the present study was primarily aimed at investigating whether CETPgene polymorphisms influence CETP activity, HDL cholesterol concentrations, and long-term prognosis in a large, prospective cohort of German patients with CAD. A secondary hypothesis investigated was whether CETPpolymorphisms might modulate the clinical benefits of statin therapy on future CAD events. Among the different polymorphisms that were previously described in the CETPgene (14), we selected for this study the functional promoter CETP/C-629Apolymorphism and the exonic CETP/I405Vpolymorphism, because they had been previously shown to have the strongest influence on CETPactivity and HDL cholesterol (14).
A detailed description of the design of the AtheroGeneStudy has been outlined previously (17). Briefly, between November 1996 and June 2000, 1,303 patients who underwent coronary angiography at the Department of Medicine II of the Johannes Gutenberg-University Mainz or the Bundeswehrzentralkrankenhaus Koblenz and who had at least one stenosis >30% diagnosed in a major coronary artery were enrolled in a cardiovascular registry. Of the patients, 910 (70%) presented with stable angina and 393 (30%) presented with an acute coronary syndrome; in the latter group, 336 had unstable angina by the Braunwald classification (class B or C) and 57 had an acute myocardial infarction (MI) within the previous 48 h. The present study was based on 1,211 patients (92.9%) who were genotyped for both the CETP/C-629Aand CETP/I405Vpolymorphisms. Missing genotype information was due to technical difficulties or a lack of deoxyribonucleic acid.
Evaluation of end points
A total of 1,206 (99.6%) of 1,211 patients were followed up during a median period of 4.1 years (maximum 5.2 years). Patients either presented at our clinic (87.2%) or a telephone interview was performed by trained medical staff. Follow-up information was obtained on death from cardiovascular causes, including fatal MI, sudden cardiac death, and death from vascular causes (n = 82); death from causes not related to heart disease (n = 25); nonfatal MI (n = 50); and nonfatal stroke (n = 31). Information on the cause of death or clinical events was obtained from hospital or general practitioner charts. Statin prescription status was obtained by a questionnaire at study enrollment and was available for all study subjects. A total of 411 patients (34%) had been receiving treatment with one of six different statins at study entry (immediately before coronary angiography) and as discharge medication and continued to take this medication at the time of follow-up (atorvastatin 24.7%, cerivastatin 19.3%, pravastatin 18.4%, simvastatin 17.3%, lovastatin 17.1%, or fluvastatin 3.2%). All patients who received statin therapy as discharge medication but did not continue to take it during follow-up were identified as “dropouts” and considered in the group of “no statin medication.” Use of statin therapy at the end of follow-up was verified by a questionnaire and general practitioner charts.
Study participants had a German nationality. The study was approved by the local ethics committee. Participation was voluntary, and each subject gave written, informed consent.
Blood was drawn under standardized conditions, samples were stored at −80°C until analysis. Lipid serum levels (total cholesterol, HDL cholesterol, low-density lipoprotein [LDL] cholesterol, and triglycerides) were determined immediately, and apolipoprotein A-I and B-100 concentrations by an immunoturbidometric assay (Tina-quant, Roche Diagnostics, Mannheim, Germany). C-reactive protein was measured by a highly sensitive, latex particle-enhanced immunoassay (Roche Diagnostics). The CETP activity was measured by a fluorescence method (Roar Biomedical, Inc., New York, New York). All information for genotyping can be obtained at our Internet site (http://www.genecanvas.org).
Hardy-Weinberg equilibrium was tested by the chi-square test with one degree of freedom. Linkage disequilibrium between the two polymorphisms was estimated by log-linear analysis (18)and was expressed by the standardized coefficient D’. The association between CETPgenotypes and lipid levels was tested by analysis of variance, comparing the mean values of lipids across genotypes. For the association of prospective outcome with genotype, a first combined cardiovascular end point, including death from cardiovascular causes, nonfatal MI, and nonfatal stroke, was considered. Because the association appeared mainly driven by cardiovascular mortality, further survival analyses focused on cardiovascular mortality and data on patients who died of other causes were censored at the time of death. The cumulative survival plot in relation to CETPgenotypes was estimated by the Kaplan-Meier method with the use of the log-rank test. Hazard risk ratios (HRRs) of future cardiovascular mortality, according to CETPpolymorphisms, were estimated by Cox regression analysis. The interaction between the CETP/C-629Apolymorphism and statin treatment on cardiovascular mortality was tested by introducing a corresponding interaction term in the Cox model. In survival analyses, the p value associated with genotype was obtained assuming a co-dominant allele effect (genotype entered as an ordinal variable 0, 1, or 2), as this genetic model appeared to fit well with the data. A value of p < 0.05 was considered to be significant. All computations were carried out with SPSS, version 10.07 (SPSS Inc., Chicago, Illinois).
Characteristics of study population according to clinical outcome
The mean age of the entire study population was 61.5 ± 10.2 years; 76.1% were male patients. Table 1shows the patients’ characteristics according to clinical outcome.
CETPgenotypes and intermediate lipid phenotypes
The genotype distribution for CETP/C-629Asignificantly deviated from the Hardy-Weinberg expectations (p = 0.008) due to a deficit of homozygotes. Because the cohort was composed of patients with CAD, this result might be explained by a lower frequency than that expected of the AAgenotype, as already reported in patients with CAD (13). To check this hypothesis, we compared the genotype frequencies of this cohort with those of a control sample of 574 German individuals recruited at the same time as the cohort (19). The frequencies of the CC, CA, and AAgenotypes were 30.8%, 51.7%, and 17.4%, respectively, in the control sample (p = 0.20 for the Hardy-Weinberg equilibrium test), compared with 36.9%, 50.6%, and 12.5%, respectively, in the CAD cohort (p = 0.004 for the genotype difference between groups), confirming the lower frequency of the AAgenotype in this CAD cohort. The genotype distribution for CETP/I405Vwas compatible with the Hardy-Weinberg expectations (p = 0.21) and did not differ from that of the control group (p = 0.78). Both polymorphisms were in moderate linkage disequilibrium (D’ = +0.26, p < 0.001), with the −629Aand 405Valleles being preferentially associated.
The two polymorphisms influenced HDL cholesterol and CETPactivity levels, with the −629Aand the 405Valleles being associated with both an increase in HDL cholesterol and a decrease in CETP activity in a co-dominant fashion (Table 2).
CETPgenotypes and clinical outcome
A significant association was found between the CETP/C-629Apolymorphism and the combined cardiovascular end point (Table 3). Carriers of the −629Aallele had a reduced risk compared with CChomozygotes (p = 0.032). This association was mainly driven by cardiovascular mortality (p < 0.0001), with no effect observed for nonfatal MI or stroke (Table 3). There was also a borderline association of this polymorphism with noncardiovascular mortality (p = 0.045), but the number of deaths was small. The association between the CETP/I405Vpolymorphism and the combined cardiovascular end point did not reach statistical significance. However, as for the CETP/C-629Apolymorphism, there was a significant association with cardiovascular mortality (Table 3). Further survival analyses focused then on cardiovascular mortality. The relationship between genotypes and cardiovascular mortality was illustrated in the Cox survival plot (Fig. 1).
To explore whether CETPgenotypes predicted future fatal cardiovascular events independent of most significant clinical and therapeutic confounders, several models were fitted (Table 4). Despite the effect of the CETP/C-629Apolymorphism on HDL cholesterol and CETP activity, inclusion of these variables in different models hardly modified the association of genotype with clinical outcome. Similar results were observed with the CETP/I405Vgenotype, although the association was weaker and lost significance after adjustment for CETP activity (Table 4). In a model including both polymorphisms, only the CETP/C-629Apolymorphism remained significantly associated with clinical outcome. Assuming a co-dominant allele effect on risk, the HRR associated with the −629Aallele was 0.51 (95% confidence interval [CI] 0.35 to 0.75; p < 0.001), and the HRR associated with the 405Vallele was 0.71 (95% CI 0.48 to 1.06; p = 0.10) in a model including the two polymorphisms simultaneously. The protective effect of the −629Aallele on cardiovascular mortality was of a similar magnitude in stable patients (HRR 0.48, 95% CI 0.31 to 0.73; p < 0.001) and in patients with acute coronary syndrome (HRR 0.53, 95% CI 0.25 to 1.12; p < 0.10) (p = 0.85 for homogeneity of the effect).
Benefit of statin therapy according to CETP/C-629Agenotype
A significant interaction between the CETP/C-629Agenotype and statin treatment on cardiovascular mortality was observed. Actually, the benefit of statin therapy was restricted to patients homozygous for the −629Callele. In these patients, mortality was divided by half in those taking statin medication, whereas no difference was observed in patients carrying the −629Aallele (Table 5, Fig. 2). The same trend was observed with the CETP/I405Vpolymorphism, but the interaction did not reach statistical significance.
This prospective study, based on a large cohort of German patients with documented CAD, confirmed that the CETP/C-629Apolymorphism influenced CETP activity and HDL cholesterol levels and, more importantly, demonstrated, for the first time, that the −629Aallele was associated with a strong protective effect on future cardiovascular mortality. Despite heterogeneity in clinical presentation and pathophysiology, this association was present in all subgroups evaluated and remained highly significant even after controlling for most potential confounders, including the strong predictors of age and ejection fraction. Furthermore, the CETP/C-629Apolymorphism appeared to modulate the long-term clinical benefit of statin therapy.
CETP was originally identified as a factor promoting the transfer of cholesteryl esters from HDL to very-low-density lipoprotein (VLDL) and LDL in exchange for triglycerides (20,21). As a result of increasing VLDL/LDL cholesterol concentrations and decreasing HDL cholesterol concentrations, CETP has often been considered as being “pro-atherogenic.” On the other hand, CETP is postulated to be a direct (22)and indirect (5)mediator in RCT and should, therefore, also be considered as “anti-atherogenic.”
An interesting window to elucidate the role of CETP in atherosclerosis is provided by genetic studies. A common TaqIB polymorphism located in intron 1 of the CETPgene has been shown in several studies to be associated with variations in CETP mass or activity and HDL cholesterol. A protective effect of the B2allele of this polymorphism on the risk of CAD has been reported (10,23). Furthermore, in the REgression GRowth Evaluation Statin Study (REGRESS) clinical trial on pravastatin, the B2allele was reported to slow down the progression of coronary atherosclerosis in patients receiving a placebo, whereas no effect was observed in statin-treated patients (16). However, the TaqIB polymorphism has no functional role by itself, and other polymorphisms in linkage disequilibrium with it should be responsible for the observed effects. A polymorphism at position −629 in the promoter of the CETPgene was in nearly complete association with the TaqIB polymorphism. It was further shown that the −629Aallele, which is almost completely concordant with the B2allele, displayed a lower promoter activity than the −629Callele, supporting the notion that this polymorphism might be responsible, at least partly, for the effects initially observed with the TaqIB polymorphism (14). In the present study, the TaqIB polymorphism was also genotyped, but we did not report the results because of the redundancy due to strong linkage disequilibrium with the C-629Apolymorphism. The effects of the CETP/I405Vmutation on CETP activity and HDL cholesterol were in accordance with earlier studies (9,24,25). However, these effects appeared to be mainly explained by its linkage disequilibrium with the C-629Apolymorphism. The two polymorphisms included in the present study were selected on the basis of their potential functionality, their relatively high frequency, and their strong effects on HDL and CETP concentrations (13). However, other polymorphisms might also be of interest, in particular, two less frequent coding polymorphisms: A373P and R451Q. In the future, haplotype analyses should be developed to evaluate the impact of all CETPgene polymorphisms on prospective outcome.
In the present study, the mechanism relating CETPgenotype to cardiovascular risk appeared rather independent of HDL cholesterol and CETP concentrations. However, it should be stressed that the association of CETPgenotype with HDL cholesterol in this cohort of patients with CAD was weaker than that generally reported in apparently healthy subjects (9,10,24), and unlike previous studies, we did not find any association with apolipoprotein A-I levels. One explanation might be that lipid/lipoprotein levels in this cohort of patients are not relevant intermediate phenotypes because of modifications induced by treatment or disease. It might also be speculated that the impact of genotype on risk is rather mediated by a local effect of CETP within the vessel wall.
One additional finding of our study was the interaction between the CETP/C-629Apolymorphism and statin therapy on clinical outcome, with the beneficial effect of statin therapy regarding future cardiovascular mortality being restricted to CChomozygotes. Which putative mechanisms might explain the relationship between the CETP/C-629Apolymorphism and long-term response to statin treatment? The CChomozygotes have a stronger promoter activity and exhibit the highest CETPexpression level (14). Atorvastatin, which was the most frequently prescribed statin in our study, has been shown to decrease cholesteryl ester transport from the protective HDL to the highly pro-atherogenic VLDL subfraction and to decrease CETP activity (26). This reduced cholesteryl ester transport from HDL to VLDL, which appears as an important mechanism to lower LDL cholesterol, might be enhanced in CChomozygotes, explaining why these subjects benefit most from (atorva)statin therapy.
Some limitations of the present study must be stressed. The AtheroGenestudy is not a prospectively randomized statin-placebo trial. Moreover, we did not have the ability to evaluate lipid parameters before and after statin therapy. The results on the genotype-statin interaction should then be considered as hypothesis-generating. Because we tested several primary and secondary hypotheses that were not independent, it was not possible to correct for multiple testing; hence, the p values provided are nominal and require further confirmation. Another limitation of the study is the relatively low number of events. Therefore, it is important to replicate these results in other prospective studies. Finally, the issue of whether or not the genotype effect is mediated by modifications of circulating lipid concentrations requires further evaluation.
This study demonstrated, for the first time, the strong protective effect of the CETP/−629Aallele on the risk of future mortality from cardiovascular causes in patients with CAD. Our data further suggested that this genetic marker may help to identify those patients who would most likely achieve a clinical benefit from statin therapy.
☆ This work was supported by grant AZ 15202-386261/545 from the “Stiftung Rheinland-Pfalz für Innovation,” Ministry for Science and Education, Mainz, Germany, and by the MAIFOR grant 2001 from the Johannes Gutenberg-University Mainz. Dr. Blankenberg is currently supported by a post-doctoral grant from INSERM.
- coronary artery disease
- cholesteryl ester transfer protein
- confidence interval
- high-density lipoprotein
- hazard risk ratio
- low-density lipoprotein
- myocardial infarction
- reverse cholesterol transport
- very-low-density lipoprotein
- Received August 27, 2002.
- Revision received December 3, 2002.
- Accepted December 26, 2002.
- American College of Cardiology Foundation
- Gotto A.M. Jr.
- Tall A.R.,
- Jiang X.,
- Luo Y.,
- Silver D.
- Ordovas J.M.,
- Cupples L.A.,
- Corella D.,
- et al.
- Dachet C.,
- Poirier O.,
- Cambien F.,
- Chapman J.,
- Rouis M.
- Ballantyne C.M.,
- Herd J.A.,
- Ferlic L.L.,
- et al.
- Blankenberg S.,
- Rupprecht H.J.,
- Bickel C.,
- et al.
- Barbaux S.C.,
- Blankenberg S.,
- Rupprecht H.J.,
- et al.
- Bruce C.,
- Davidson W.S.,
- Kussie P.,
- et al.
- Collet X.,
- Tall A.R.,
- Serajuddin H.,
- et al.
- Fumeron F.,
- Jemaa R.,
- Poirier O.,
- Cambien F.,
- Tiret L.
- Agerholm-Larsen B.,
- Nordestgaard B.G.,
- Steffensen R.,
- Jensen G.,
- Tybjaerg-Hansen A.
- Bruce C.,
- Sharp D.S.,
- Tall A.R.
- Guerin M.,
- Lassel T.S.,
- Le Goff W.,
- Farnier M.,
- Chapman M.J.