Human Genetics

, Volume 121, Issue 3, pp 401–411

Cholesteryl ester transfer protein gene haplotypes, plasma high-density lipoprotein levels and the risk of coronary heart disease


    • Laboratory for Genetic Epidemiology, Western Australian Institute for Medical Research, UWA Centre for Medical ResearchUniversity of Western Australia
  • John P. Beilby
    • Clinical BiochemistryPathWest, Laboratory Medicine Western Australian
    • School of Surgery and PathologyUniversity of Western Australia
  • Caroline M. L. Chapman
    • Clinical BiochemistryPathWest, Laboratory Medicine Western Australian
  • Joseph Hung
    • Sir Charles Gairdner Hospital Campus of the Heart Research Institute of Western Australia
    • School of Medicine and PharmacologyUniversity of Western Australia
  • Brendan M. McQuillan
    • Sir Charles Gairdner Hospital Campus of the Heart Research Institute of Western Australia
    • School of Medicine and PharmacologyUniversity of Western Australia
  • Peter L. Thompson
    • School of Surgery and PathologyUniversity of Western Australia
    • Sir Charles Gairdner Hospital Campus of the Heart Research Institute of Western Australia
  • Lyle J. Palmer
    • Laboratory for Genetic Epidemiology, Western Australian Institute for Medical Research, UWA Centre for Medical ResearchUniversity of Western Australia
    • School of Medicine and PharmacologyUniversity of Western Australia
Original Investigation

DOI: 10.1007/s00439-007-0326-2

Cite this article as:
McCaskie, P.A., Beilby, J.P., Chapman, C.M.L. et al. Hum Genet (2007) 121: 401. doi:10.1007/s00439-007-0326-2


High-density lipoprotein cholesterol (HDL-C) is a known inverse predictor of coronary heart disease (CHD) and is thus a potential therapeutic target. Cholesteryl ester transfer protein (CETP) is a key protein in HDL-C metabolism such that elevated CETP activity is associated with lower HDL-C. Currently available HDL-C raising drugs are relatively ineffective and evidence suggesting the role of CETP in HDL-C levels has promoted the development of CETP inhibitors as potential therapeutic agents for CHD. We investigated three SNPs in the CETP gene in two cross-sectional community-based populations (n = 1,574 and 1,109) and a population of 556 CHD patients to determine if reduced CETP activity due to genetic variations in the CETP gene would increase HDL-C levels and reduce the risk of CHD. CETP genotypes and haplotypes were tested for association with lipid levels, CETP activity and risk of CHD. Multivariate analysis showed the common AAB2 haplotype defined by the G-2708A, C-629A and TaqIB polymorphisms, was consistently associated with reduced CETP activity and increased HDL-C levels. A mean increase in HDL-C levels of 0.16–0.24 mmol/l was observed in individuals with two copies of the AAB2 haplotype relative to non AAB2 carriers across all three populations (P < 0.001). A case-control study of males indicated no association between single SNPs or haplotypes and the risk of CHD. These results suggest that raising HDL-C via CETP inhibition may not alter risk of CHD. Randomized control trials are needed to determine whether CETP inhibition will in reality reduce risk of CHD by raising HDL-C.


The concentration of high-density lipoprotein cholesterol (HDL-C) has consistently been shown to be a powerful negative predictor of coronary heart disease (CHD) in men and women (Ansell et al. 2004; Barter et al. 2003). The antiatherogenic effect of HDL-C includes anti-inflammatory and antioxidant activity along with promoting reverse cholesterol transport, the pathway by which cholesterol is transferred from the peripheral tissues to the liver for either recycling or excretion (de Grooth et al. 2004). Variations in HDL-C levels are controlled by environmental and genetic factors, with at least 50% of the variation thought to be due to genetic factors (Lusis 2003).

Cholesteryl ester transfer protein (CETP) is a key protein in HDL-C metabolism and generally considered proatherogenic as it promotes the transfer of cholesteryl esters for triglycerides from HDL-C to ApoB-containing lipoproteins (Agerholm-Larsen et al. 2000; Barter et al. 2003). Thus increased CETP activity is associated with lower HDL-C and higher low-density lipoprotein cholesterol (LDL-C) levels, that is, a more atherogenic lipid profile. However, it is also thought that CETP may be antiatherogenic by enhancing the rate of reverse cholesterol transport (de Grooth et al. 2004)-the mechanism by which cholesterol in peripheral tissues is transported to the liver for elimination. CETP inhibition is therefore currently being investigated as a possible management strategy for CHD. One CETP inhibitor-JTT-705-is currently in clinical testing while a second-torcetrapib-has recently been withdrawn from clinical trial. Early results suggest that these inhibitors can lower CETP activity and increase HDL-C levels in humans; however, there is no current definitive evidence that increasing HDL-C levels by this method will prevent CHD (Clark 2006).

Animal studies using mice models to study over expression of CETP, and rabbit models to study CETP inhibition have produced conflicting results (Forrester et al. 2005). Unlike mice, rabbits have a high level of CETP activity in plasma and are thus an ideal model to investigate the effects of CETP inhibition (Barter and Kastelein 2006). In one study, the CETP inhibitor JTT-705 increased plasma HDL-C and decreased aortic arch lesions by 70% (Okamoto et al. 2000). A subsequent study in rabbits, however, showed no significant difference in aortic cholesterol deposition between those administered the CETP inhibitor and the control group, despite elevated HDL-C levels in those receiving JTT-705 (Huang et al. 2002). Human subjects with genotypes which confer CETP deficiency have significantly elevated HDL-C levels; however, studies have not shown a consistent reduction in the risk of CHD, with both decreased and increased CHD risk reported (Hirano et al. 1997; Inazu et al. 1990a). Although reduced CETP activity and increased HDL-C levels have been observed in clinical trials’ participants receiving CETP inhibitors, the impact of these inhibitors on CHD risk is yet to be determined (Clark 2006).

The CETP gene located on chromosome 16q21 is highly polymorphic (Klerkx et al. 2003) and a number of single nucleotide polymorphisms (SNPs) have been studied in both the coding and non-coding regions of the gene (Bauerfeind et al. 2002; Lu et al. 2003). The hypothesis that CETP may be a potential target for reducing cardiovascular disease risk stems from a study of seemingly healthy, CETP-deficient Japanese subjects (Brown et al. 1989; Inazu et al. 1990b). These individuals who lacked a functional copy of the CETP gene exhibited no measurable CETP activity, large increases in HDL-C and lower LDL-C. A number of more recent studies have demonstrated that haplotypes of the CETP gene affect HDL-C levels by varying amounts, but there is currently no evidence that CETP haplotypes are associated with CHD risk (Bauerfeind et al. 2002; Klerkx et al. 2003; Knoblauch et al. 2004; Lu et al. 2003). In the current study, two SNPs from the promoter region-G-2708A and C-629A-and the TaqIB polymorphism from intron 1, which have previously been shown to be associated with HDL-C, were chosen to determine if individual SNPs and haplotypes were associated with CHD risk.

Ultimately, randomized controlled clinical trials of CETP inhibitors will determine if increased HDL-C levels will reduce CHD risk. In the meantime, observational studies may provide some insight into whether increasing HDL-C levels through CETP inhibition will improve clinical outcomes. Common polymorphisms in the CETP gene that cause a lifelong increase in HDL-C levels due to reduced CETP activity provide a useful model in which to study CETP inhibition. The aim of this study was to determine the inter-relationship between three CETP SNPs, CETP activity, lipid levels and the risk of CHD.

Patients and methods


Subjects were selected from three Western Australian populations: the Carotid Ultrasound Disease Assessment Study (CUDAS) (Chapman et al. 2001; McQuillan et al. 1999), the Busselton Population Health Survey (Olynyk et al. 1999) and the Carotid Ultrasound in Patients with Ischaemic Heart Disease (CUPID) (McCaskie et al. 2006). These three populations were predominantly European-Australian and were collected from the same region of Western Australia. The CUDAS group consisted of 1,109 subjects with an equal male to female ratio and equal numbers in each age decile between 20 and 70 years. This population was selected from a random electoral survey from the Perth metropolitan area and collected in 1995. The Busselton Population Health Survey cohort consisted of 1,574 subjects and was studied as part of a cross-sectional community study in 1994/95. The CUPID cohort was collected in 1995 and consisted of 556 subjects between 26 and 60 years of age. All CUPID subjects were patients of Sir Charles Gairdner Hospital, Perth who had presented at the Cardiac Catheter Laboratory for angiographic assessment. All patients were medically stable at the time of data collection but all had a past history of angina or myocardial infarction (MI) and angiographically demonstrated CHD with at least one coronary vessel with >50% stenosis.

Males subject from CUPID (n = 485) and healthy males with no history of CHD from CUDAS (n = 502) were used to form a case-control study to test for association between CETP SNPs and haplotypes and risk of CHD.

All subjects gave written informed consent. The study protocols were approved by the Institutional Ethics Committee of the University of Western Australia, the Sir Charles Gairdner Hospital Research Institutional Ethics Committee and the Busselton Population Medical Research Foundation.

A self-administered questionnaire was used to report the prevalence of smoking and physician-diagnosed diabetes. Resting blood pressures and anthropometric measurements (height, weight, waist and hip circumferences) were taken from all subjects in an identical manner across the three populations. A fasting venous blood sample was collected for lipid measurements. No subfraction analysis was performed.

Laboratory measurements

Biochemical parameters were analysed as previously described using the same assays and instrumentation for all three populations (Chapman et al. 2001). The G-2708A (rs12149545) and C-629A (rs1800775) polymorphisms were genotyped by mutagenically separated PCR (Rust et al. 1993) assay based on published methods (Klerkx et al. 2003). The TaqIB (rs708272) polymorphism was genotyped as described elsewhere (Fumeron et al. 1995). CETP activity was determined from a subset of 57 CUDAS individuals chosen to be homozygous for CETP genotypes to provide certain phase information (n = 20 for GCB1, n = 21 for AAB2, n = 16 for GAB2) using a CETP activity assay from Roar Biomedical Inc., USA (Ordovas 2000). The coefficient of variation for the assay was 4.0% at a CETP activity of 313 nmol/l/h and 4.7% at a CETP activity of 155 nmol/l/h.

Statistical analysis

The primary outcome variables for the association analyses were the serum HDL-C, LDL-C, CETP activity, and the presence or absence of CHD. The principal explanatory variables were the three genotyped CETP polymorphisms. The SNP genotypes were coded into three classes and analyzed categorically with the most common homozygous genotype for each SNP as the reference category. Individual haplotypes were analyzed categorically with zero copies of the haplotype as the reference category to which the effects of one and two copies were compared. Goodness-of-fit tests using Akaike Information Criteria and r-squared values were used to determine if a dominant or recessive model would explain the data better than the standard co-dominant fit.

Hardy–Weinberg equilibrium was tested at each SNP locus on a contingency table of observed-versus-predicted genotypic frequencies using a modified Markov-chain random walk algorithm (Guo and Thompson 1992). Pairwise linkage disequilibrium (LD) between each pair of SNP loci was analyzed by a likelihood-ratio test, whose empirical distribution was obtained by a permutation procedure (Slatkin and Excoffier 1996). Lewontin’s disequilibrium coefficient D´ was estimated for each pair of SNPs using the JLIN software package (Carter et al. 2006).

Analysis of variance was used to compare CETP activity and HDL-C level, across the three levels of each genotype and χ2 tests on contingency tables were used when comparing genotype to categorical variables.

Genotypic and haplotypic associations with the outcomes of interest were investigated using the SimHap program ( This approach uses current estimation-maximization based methods for the estimation of haplotypes from unphased genotype data and incorporates simulation techniques to model haplotypic associations in population-based samples. Within SimHap, generalized linear models (linear and logistic regression) (GLMs) were used to model the effects of multiple covariates on the continuous and dichotomous outcomes, including an investigation of the need for interaction or polynomial terms. Stepwise variable selection procedures were employed to identify potential covariates in multivariate analyses. The current implementation of haplotype analysis in SimHap resolves all possible haplotype configurations for each individual with unknown phase, and the posterior probability of each configuration is calculated. Association analysis within the GLM framework uses simulation to incorporate these probabilities and correctly deal with the uncertainty around imputed haplotypes. P-values were derived by empirical simulation where possible. Statistical significance was defined at the standard 5% level.


Population characteristics

DNA from a total of 3,239 subjects was available for genotyping. The characteristics of the study populations are described in Table 1. As expected, the CHD risk factors of the CUPID population were significantly different from the other two populations with a greater proportion of males, increased prevalence of smoking and diabetes mellitus, lower HDL-C, higher BMI and triglycerides. They also had lower mean cholesterol, and LDL-C levels due to the increased use of cholesterol-lowering medication.
Table 1

Description of three study populations

Variable and mean (SD)

CUDAS (1,109)

Busselton (1,574)

CUPID (556)

Age (years)

53 (13)

49 (16)*

50 (5) #

BMI (kg/m2)

26.1 (4.1)

26.1 (4.2)

28.2 (4.1) #

WHR (cm)

0.84 (0.09)

0.85 (0.09)*

0.92 (0.07) #

Systolic BP (mmHg)

128 (19)

124 (18)*

126 (10) #

Diastolic BP (mmHg)

80 (10)

75 (10)*

81 (10) #

Total Cholesterol (mmol/l)

5.6 (1.0)

5.6 (1.1)

5.2 (1.0) #

LDL Cholesterol (mmol/l)

3.6 (0.9)

3.6 (1.0)

3.2 (1.0) #

Triglycerides (mmol/l)

1.3 (0.7)

1.3 (0.9)

2.0 (1.4) #

HDL-C Cholesterol (mmol/l)

1.3 (0.4)

1.4 (0.4)*

1.1 (0.3) #

Alcohol intake (g/day)

8.6 (11.0)


12.9 (13.7) #

Males, % (n)

50 (558)

45 (865)*

87 (487) #

Physician-diagnosed diabetes, % (n)

2.1% (23)

4.7% (73)*

15.8 (88) #

Past history of MI or stroke

5.7 (63)

4.9 (73)

61.3 (341) #

Cholesterol lowering medication


6.8 (75)

66.7 (371) #

Smoking (ever smoked), % (n)

49.5 (549)

46.5 (732)

72.8 (406) #

BMI indicates body mass index, WHR indicates waist hip ratio, BP blood pressure, NI no information

P < 0.001 for CUDAS versus Busselton

# P < 0.001 for CUDAS versus CUPID

Genotype and allele frequencies are reported in Table 2. The genotypic distributions of G-2708A, C-629A and TaqIB genotypes in each population were in Hardy–Weinberg equilibrium. The 3 SNPs were in strong pairwise linkage disequilibrium (D′ > 0.8) within each study population (Fig. 1).
Table 2

Genotypic and allelic frequencies for CETP SNPs


Major homozygote


Minor homozygote

Minor allele frequency (SE)



540 (49%; GG)

455 (41%; GA)

106 (10%; AA)

0.30 (0.013)


313 (29%; CC)

537 (49%; CA)

248 (22%; AA)

0.47 (0.015)


379 (34%; B1B1)

534 (48%; B1B2)

190 (17%; B2B2)

0.41 (0.015)



707 (46%; GG)

698 (45%; GA)

141 (9%; AA)

0.32 (0.001


390 (25%; AA)

795 (51%; CA)

375 (24%; CC)

0.49 (0.012)


481 (31%; B1B1)

794 (50%; B1B2)

295 (19%; B2B2)

0.44 (0.013)



285 (51; GG)

225 (41%; GA)

42 (8%; AA)

0.28 (0.019)


155 (28%; CC)

278 (50%; CA)

119 (22%; AA)



196 (36; B1B1)

262 (47; B1B2)

93 (17%; B2B2)


Data are number of samples assessed (%; genotype)
Fig. 1

Pairwise linkage disequilibrium for 3 CETP SNPs in the CUDAS, Busselton and CUPID populations

Lipid levels

Association with non-genetic covariates

Multivariate modelling determined important independent predictors of HDL-C in all three populations (footnote, Table 3). These covariates accounted for 48% of the variation in HDL-C levels in the CUDAS population, 24% in CUPID and 38% in Busselton. There was a strong inverse correlation between HDL-C and triglyceride levels in CUDAS (r = −0.51), CUPID (r = −0.27) and Busselton (r = −0.39); all P < 0.01. Since triglyceride levels were strongly correlated with HDL-C, triglycerides were excluded from the model for single SNP and haplotype analysis as it was thought this covariate may mask the effects of other potential predictors.
Table 3

Multivariate CETP haplotype analysis of serum HDL-C (mmol/l) in the CUDAS, Busselton and CUPID populations


GCB1 estimated marginal mean (SE)


AAB2 estimated marginal mean (SE)


GAB2 estimated marginal mean (SE)



Frequency = 0.52


Frequency = 0.29


Frequency = 0.12


 0 copies

1.41 (0.14)

1.29 (0.13)

1.34 (0.14)

 1 copy

1.33 (0.02)

1.34 (0.02)

1.31 (0.02)

 2 copies

1.28 (0.06)

1.53 (0.04)

1.23 (0.08)


Frequency = 0.50


Frequency = 0.31


Frequency = 0.13


 0 copies

1.45 (0.15)

1.36 (0.15)

1.42 (0.15)

 1 copy

1.42 (0.02)

1.44 (0.02)

1.39 (0.02)

 2 copies

1.34 (0.02)

1.52 (0.03)

1.31 (0.06)


Frequency = 0.54


Frequency = 0.26


Frequency = 0.12


 0 copies

1.20 (0.23)

1.10 (0.23)

1.12 (0.23)

 1 copy

1.13 (0.04)

1.15 (0.03)

1.16 (0.03)

 2 copies

1.06 (0.04)

1.31 (0.06)

1.03 (0.17)

aHDL-C in CUDAS adjusted for age, sex, SBP, BMI, WHR, presence of diabetes mellitus, pack- years of smoking, alcohol and consumption

bHDL-C in Busselton adjusted for age, sex, SBP, BMI, WHR, presence of diabetes mellitus, smoking and alcohol consumption

cHDL-C in CUPID adjusted for age, sex, SBP, BMI, WHR, presence of diabetes mellitus, pack-years of smoking and alcohol consumption

Analysis of association between single SNPs and lipid values

After adjustment for non-genetic predictors, multivariate analysis indicated that the G-2708A (P < 0.001), C-629A (P < 0.001) and TaqIB (P < 0.001) polymorphisms were significantly associated with HDL-C levels in the CUDAS population (Fig. 2). Similarly, these SNPs were significantly associated with HDL-C levels in both the Busselton (TaqIB: P < 0.001; G-2708A: P < 0.001; C-629A: P < 0.001) and CUPID (TaqIB: P < 0.001; G-2708A: P = 0.002; C-629A: P < 0.001) populations (Fig. 2). The G-2708A, C-629A and TaqIB polymorphisms accounted for 1.6, 3.2 and 2.1% of the variation in HDL-C levels in CUDAS, 0.7, 1.7 and 0.6% in Busselton and 1.0, 0.7 and 1.2% in CUPID, respectively. When all three SNPs were included in the linear model, cumulatively they accounted for 3.6% of the total variation in HDL-C levels in CUDAS, 1.7% in Busselton and 0.9% in CUPID. The need for stratification by sex was examined via the inclusion of a genotype-sex interaction term in the various models. This term was not statistically significant therefore stratification by sex was not performed. However, sex was included as a covariate where it was a significant predictor of the outcome.
Fig. 2

Adjusted genotype-specific means in serum HDL-C (mmol/l) levels for three CETP polymorphisms in the CUDAS, Busselton and CUPID populations. HDL-C values are adjusted for age, sex, SBP, body-mass index, waist-hip ratio, alcohol consumption, pack years of smoking and presence of diabetes mellitus

The mean difference in HDL-C levels between genotypes with the lowest and the highest HDL-C values, across the three populations was 0.19 mmol/l for G-2708A, 0.13 mmol/l for C-629A and 0.14 mmol/l for TaqIB (Fig. 2). None of the SNPs were significantly associated with LDL-C levels in the three populations (data not shown). Goodness-of-fit tests determined that a co-dominant genetic model was the most appropriate for all SNPs.

Analysis of association between haplotypes and lipid values

The combined effect of the CETP polymorphisms on plasma HDL-C levels are summarized in Table 3 and Fig. 3a. The three polymorphisms defined three main haplotypes-GCB1, AAB2 and GAB2-accounting for >90% of the observed haplotypes for G-2708A, C-629A and TaqIB. The univariate associations of the haplotypes with HDL-C levels among individuals homozygous at each SNP (so phase could be unambiguously assigned) are shown in Fig. 3a. HDL-C levels were observed to be the highest in individuals carrying two copies of the AAB2 haplotype. These results were consistent across all three populations.
Fig. 3

Haplotype-specific means of a measured serum HDL-C (mmol/l) levels in the CUDAS, Busselton and CUPID populations. b CETP activity for the three CETP polymorphisms. The number of individuals measured for each comparison is shown on the graphs. The error bars indicate one standard deviation from the mean

The multivariate modelling of serum HDL-C levels in all genotyped individuals from the CUDAS, Busselton and CUPID populations for the different CETP haplotypes, adjusting for known independent predictors of HDL-C, is shown in Table 3. HDL-C levels were on average 0.24 mmol/l higher for individuals carrying two copies of the AAB2 haplotype than in individuals not carrying this haplotype (P < 0.0001) among the CUDAS population. Similar increases of 0.16 mmol/l (P < 0.0001) and 0.21 mmol/l (P = 0.001) were estimated among individuals from the Busselton and CUPID populations respectively. This haplotype explained an estimated 2.7% of total variation in HDL-C levels in CUDAS, 1.9% in Busselton and 2.2% in CUPID. The most common CGB1 haplotype was associated with lower HDL-C levels when compared to all other haplotypes (Table 3). This haplotype explained an estimated 1.5% of total variation in HDL-C levels in CUDAS, 0.9% in Busselton and 2.2% in CUPID. Results of functional analysis in the CUDAS population (Fig. 3b) showed that CETP activity was highest for the most common haplotype GCB1, lowest for AAB2 and intermediate for GAB1 (P < 0.0001).

Association between SNPs, haplotypes and coronary heart disease

Using a Mendelian randomization framework (Davey Smith and Ebrahim 2003), the independent observed associations between HDL-C and CHD risk in the current case-control study (CHD OR = 0.89 for a 0.1 mmol/l increase in HDL-C), and CETP haplotypes and HDL-C (Table 3), predicted that the AAB2 CETP haplotype would confer a reduced risk of CHD (estimated OR = 0.80). Univariate analysis of male cases with CHD (= 487) versus healthy male controls from the CUDAS population (= 502) suggested that none of the CETP G-2708A, C-629A and TaqIB polymorphisms were significantly associated with the presence of CHD (OR = 0.91, 95%CI = 0.75–1.1, P = 0.31; OR = 0.97, 95%CI = 0.81–1.15, P = 0.61; OR = 0.96, 95%CI = 0.81–1.15, P = 0.90, respectively). Multivariate modelling confirmed that there was no significant association between these three SNPs and the likelihood of CHD. No associations with CETP haplotypes and risk of CHD were observed in either univariate (Table 4) or multivariate analyses adjusting for conventional risk factors of disease. Although both triglyceride level and HDL-C were significant predictors of CHD, neither of these were included as covariates as we were testing the hypothesis that CETP polymorphisms affect risk of disease via their association with lipids. In further analyses, the population was split by the median triglyceride level of 1.5 mmol/l and analysed for an association between the risk of CHD and the AAB2 haplotype under a co-dominant, dominant or recessive model. No association was found for either the high or low triglyceride groups.
Table 4

Univariate results of CETP haplotype case-control analysis


GCB1 odds ratio (95% CI)


AAB2 odds ratio (95% CI)


GAB2 odds ratio (95% CI)


0 copies







1 copy

1.14 (0.84,1.55)

1.01 (0.78,1.31)

1.23 (0.91,1.66)

2 copies

1.07 (0.76,1.52)

0.78 (0.48,1.27)

0.54 (0.20,1.44)

CUDAS males versus CUPID males


In this study, we have presented results on 3,239 subjects from three separate populations consisting of two large cross-sectional community based samples of adults (= 1,109 and 1,574) and the third with CHD (= 556). Strong associations between higher HDL-C levels and CETP-2708A, -629A and TaqIB-B2 alleles and the haplotype comprised of these alleles was replicated across all three populations. These associations were independent of other predictors of HDL-C including age, sex, SBP, BMI, diabetes and smoking. Haplotype analysis also showed the AAB2 haplotype was associated with the lowest CETP activity. However, results from the case-control study demonstrated no association between SNPs or haplotypes of CETP and the risk of CHD. These data may suggest that higher levels of HDL-C associated with reduced CETP activity are not necessarily also associated with decreased risk of CHD.

The TaqIB polymorphism in intron 1 of the CETP gene has been studied extensively due to associations with HDL-C and CETP activity (Gudnason et al. 1999; Klerkx et al. 2003; Lu et al. 2003; Ukkola et al. 1994), but so far the functional effects of this polymorphism are poorly understood. C-629A, G-2708A and TaqIB were in strong linkage disequilibrium in the current study, and it was not possible to determine which of these, if any, was the functional site. Another study of the promoter region SNPs including TaqIB, indicated that only the C-629A variant was independently associated with HDL-C levels (Klerkx et al. 2003), however, this has not been consistently found (Eiriksdottir et al. 2001).

A meta-analysis of CETP polymorphisms and HDL-C levels in over 10,000 individuals has demonstrated that the TaqIB-B2B2 genotype was associated with lower CETP activity and higher HDL-C levels (0.12 mmol/l, 95% CI = 0.11–0.13, P 0.0001) compared to the B1B1 genotype (Boekholdt and Thompson 2003). In the current study, individuals carrying the B2B2 genotype also had higher mean HDL-C levels of 0.14, 0.11 and 0.17 mmol/l for the CUDAS, Busselton and CUPID populations respectively, compared with the B1B1 genotypes, consistent with the results of the meta-analysis. However, like our study, the meta-analysis was unable to conclusively determine that higher HDL-C levels were associated with a reduced risk of cardiovascular disease (CVD) (Boekholdt and Thompson 2003; de Grooth et al. 2004; Liu et al. 2002). Another meta-analysis of 13,677 subjects has shown that the B2B2 genotype of the TaqIB variant was associated with 0.11 mmol/l (95% CI = 0.10–0.12, P < 0.0001) higher HDL-C levels compared with the B1B1 genotype, and that the B2B2 genotype was associated with a reduced risk of CHD (OR = 0.78, 95% CI = 0.66–0.93) compared with B1B1 individuals (Boekholdt et al. 2005). However, results from meta-analysis studies may be influenced by publication bias or the specific ascertainment and metabolic architecture of the study populations used (de Grooth et al. 2004) and so replication studies are needed to support this finding.

A number of studies have shown that the C-629A variant is a functional polymorphism associated with lower CETP activity and higher HDL-C levels (Dachet et al. 2000; Klerkx et al. 2003; Le Goff et al. 2003). A case-control study of 547 MI cases and 505 controls showed that the CETP C-629A variant was associated with a lower odds ratio of MI (OR = 0.82, 95% CI = 0.68–0.97, P = 0.025) when adjusted for age and sex (Tobin et al. 2004). In 1,211 patients with CHD, the CETP-629AA genotype was associated with a strong protective effect on future mortality from cardiovascular causes, independent of its association with HDL-C cholesterol and CETP activity (Blankenberg et al. 2003). Contrary to these findings, a recent population based cohort study of 8,141 individuals observed an increase in risk of coronary disease among carriers of the -629A allele compared to CC homozygotes (Borggreve et al. 2006). Data from the current study support the association between the -629AA genotype and higher HDL-C levels but not with the risk of CHD as either an atherogenic or antiatherogenic mechanism.

The CETP promoter polymorphism, G-2708A, has also been demonstrated to be an independent predictor of CETP activity and HDL-C levels (Klerkx et al. 2003). Data from the current study confirmed this report. The -2708AA genotype had 0.23, 0.15 and 0.20 mmol/l higher HDL-C levels than the -2708GG genotype for the CUDAS, Busselton and CUPID populations, respectively.

SNPs within the same gene are known to be inherited randomly but as stable haplotype blocks of DNA that may provide a powerful method to determine if functional genetic variants are located within a region of DNA. A number of studies of CETP haplotypes have reported associations with increased HDL-C levels and reduced CETP activity that is similar to single SNP analysis (Bauerfeind et al. 2002; Blankenberg et al. 2003; Gudnason et al. 1999; Kakko et al. 2001; Klerkx et al. 2003; Knoblauch et al. 2004; Lu et al. 2003; Tobin et al. 2004). All these studies included a number of promoter region SNPs, TaqIB and several studies incorporated Ile405Val in the haplotypes. Two studies indicated that CETP haplotypes produced a greater change in the HDL-C levels than the single SNPs (Knoblauch et al. 2002; Winkelmann et al. 2003). This was also observed in the current study, consistent with previously suggested benefits of haplotype analysis for examining SNPs in strong LD (Akey et al. 2001).

To the best of our knowledge, only one other study has investigated the association between CETP haplotypes and the risk of CHD (Tobin et al. 2004). This study concluded that while the -629AA genotype produced a significant reduction in the risk of MI, when considered as part of a haplotype (together with the C-631A and Ile405Val polymorphisms), estimated haplotypic effect sizes were suggestive of a reduced risk of CHD, although this association did not reach formal statistical significance. This result was consistent with the results of our tests for haplotypic association.

Many studies have demonstrated associations with reduced CETP activity, increased HDL-C levels and either single SNPs or haplotypes in the promoter region of the CETP gene. However, the increased levels in HDL-C have not consistently translated into greater protection from CHD across a number of studies, and some studies have shown a paradoxical increase in CHD risk. These higher HDL-C levels should theoretically be associated with a reduction in CVD across large study populations, since it has been reported that a 0.025 mmol/l increase in HDL-C reduces the risk of CVD by 2–5% (Gordon et al. 1989).

The reason this has not been observed is unclear, but may be due to several factors. Firstly, CETP promotes reverse cholesterol transport and therefore CETP inhibition is potentially proatherogenic despite increasing HDL-C. The metabolic setting of the subjects studied may be important. Animal studies have shown that CETP can have antiatherogenic effects under hypertriglyceridemic conditions (de Grooth et al. 2004). A study of the effect of the C-629A SNP on HDL-C levels was modified by the serum TG levels (Borggreve et al. 2005). In the current study, there was no interaction between triglyceride levels and CETP SNPs or haplotypes and their effect on HDL-C levels and risk of CHD.

Another possible reason for lack of reported associations between CETP SNPs or haplotypes and CHD may be the effect sizes of these variants. The sample size in the present study provided sufficient power to detect an odds ratio of ≥1.5 or ≤0.6 for cases relative to controls. Our results suggest that although strongly associated with HDL-C, CETP SNPs and haplotypes are responsible for less than 4% each of the total variation in plasma HDL-C concentrations in all three populations. If the CETP polymorphisms are attributable to a difference in CHD risk between cases and controls, this difference is likely to be modest (OR = 0.8 as predicted via Mendelian randomization). The present study, as like many previous studies, does not have the statistical power to observe such a modest effect. To observe a small effect, much larger population-based samples are required.

The results of the current study together with the inconsistent ability of previous studies to observe associations between CETP SNPs and risk of CHD may be of clinical importance as they demonstrate that a genetically determined reduction in CETP activity is associated with an increase in HDL-C levels but that this is not translated into protection against CHD. If CETP does not implicitly increase risk of CHD, but is simply predictive of it because of reverse causality or confounding, the suitability of CETP as a potential therapeutic target for the prevention of CHD becomes doubtful. Only with the use of larger sample sizes better powered to detect modest genetic effects, and with the use of large-scale randomized controlled clinical trials evaluating the impact of CETP inhibitors on CHD development, can we determine the effectiveness of such inhibitors on reducing risk of disease.


This study was supported by grants-in-aid from the HeartSearch WA (to C.C and J.B) and National Heart Foundation grant-in-aid G97P 5002. The authors thank the Busselton Population Medical Research Foundation for access to the survey.

Copyright information

© Springer-Verlag 2007