This Article Abstract Full Text (PDF) All Versions of this Article: 111/3/278    most recent 01.CIR.0000153341.46271.40v1 Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Permissions Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Boekholdt, S.M. Articles by Kastelein, J.J.P. Search for Related Content PubMed PubMed Citation Articles by Boekholdt, S.M. Articles by Kastelein, J.J.P. Pubmed/NCBI databases Gene GEO Profiles HomoloGene UniGene Compound via MeSH Substance via MeSH Related Collections Genetics of cardiovascular disease Lipid and lipoprotein metabolism Related Article

(Circulation. 2005;111:278-287.)
© 2005 American Heart Association, Inc.

Coronary Heart Disease

Cholesteryl Ester Transfer Protein TaqIB Variant, High-Density Lipoprotein Cholesterol Levels, Cardiovascular Risk, and Efficacy of Pravastatin Treatment

Individual Patient Meta-Analysis of 13 677 Subjects

S.M. Boekholdt, MD; F.M. Sacks, MD; J.W. Jukema, MD, PhD; J. Shepherd, MD; D.J. Freeman, PhD; A.D. McMahon, PhD; F. Cambien, MD; V. Nicaud, MA; G.J. de Grooth, MD, PhD; P.J. Talmud, PhD, DSc, FRCPath; S.E. Humphries, PhD, FRCPath, MRCP; G.J. Miller, MD, FRCP; G. Eiriksdottir, MSc; V. Gudnason, MD, PhD; H. Kauma, MD; S. Kakko, MD, PhD; M.J. Savolainen, MD, PhD; M. Arca, MD; A. Montali, BSc; S. Liu, MD; H.J. Lanz, MD; A.H. Zwinderman, PhD; J.A. Kuivenhoven, PhD; J.J.P. Kastelein, MD, PhD

From the Departments of Cardiology (S.M.B.), Vascular Medicine (G.J.d.G., J.A.K., J.J.P.K.), and Clinical Epidemiology and Biostatistics (A.H.Z.), Academic Medical Center, Amsterdam, the Netherlands; Nutrition Department (F.M.S.), Harvard School of Public Health, Boston, Mass; Department of Cardiology (J.W.J.), Leiden University Medical Center, Leiden, the Netherlands; Department of Pathological Biochemistry (J.S.), Glasgow Royal Infirmary, Glasgow, Scotland; Division of Developmental Medicine (D.J.F.) and Robertson Centre for Biostatistics (A.D.M.), University of Glasgow, Glasgow, Scotland; Institut National de la Santé et de la Recherche Médicale Unité 525 (V.N., F.C.), Paris, France; Division of Cardiovascular Genetics (P.J.T., S.E.H.), Department of Medicine, Royal Free and University College Medical School, London, England; MRC Cardiovascular Research Group (G.J.M.), Wolfson Institute of Preventive Medicine, St. Bartholomew’s and Royal London School of Medicine, London, England; Hjartavernd Research Institute (G.E., V.G.), Icelandic Heart Association, Kopavogur, Iceland; Department of Internal Medicine (H.K., S.K., M.J.S.), University of Oulu, Oulu, Finland; Department of Clinical and Applied Medical Therapy (M.A., A.M.), University of Rome La Sapienza, Rome, Italy; Division of Preventive Medicine (S.L.), Harvard Medical School and Brigham and Women’s Hospital, Boston, Mass; and Bristol-Myers Squibb Co (H.J.L.), Princeton, NJ.

Correspondence to John J.P. Kastelein, MD, PhD, Department of Vascular Medicine, Academic Medical Center, Meibergdreef 9, 1105 AZ Amsterdam, The Netherlands. E-mail e.vandongen{at}amc.uva.nl

Received April 22, 2004; revision received September 13, 2004; accepted September 24, 2004.

	Abstract

Top Abstract Introduction Methods Results Discussion Conclusions References

 
Background— Several studies have reported that the cholesteryl ester transfer protein (CETP) TaqIB gene polymorphism is associated with HDL cholesterol (HDL-C) levels and the risk of coronary artery disease (CAD), but the results are inconsistent. In addition, an interaction has been implicated between this genetic variant and pravastatin treatment, but this has not been confirmed.

Methods and Results— A meta-analysis was performed on individual patient data from 7 large, population-based studies (each >500 individuals) and 3 randomized, placebo-controlled, pravastatin trials. Linear and logistic regression models were used to assess the relation between TaqIB genotype and HDL-C levels and CAD risk. After adjustment for study, age, sex, smoking, body mass index (BMI), diabetes, LDL-C, use of alcohol, and prevalence of CAD, TaqIB genotype exhibited a highly significant association with HDL-C levels, such that B2B2 individuals had 0.11 mmol/L (0.10 to 0.12, P<0.0001) higher HDL-C levels than did B1B1 individuals. Second, after adjustment for study, sex, age, smoking, BMI, diabetes, systolic blood pressure, LDL-C, and use of alcohol, TaqIB genotype was significantly associated with the risk of CAD (odds ratio=0.78 [0.66 to 0.93]) in B2B2 individuals compared with B1B1 individuals (P for linearity=0.008). Additional adjustment for HDL-C levels rendered a loss of statistical significance (P=0.4). Last, no pharmacogenetic interaction between TaqIB genotype and pravastatin treatment could be demonstrated.

Conclusions— The CETP TaqIB variant is firmly associated with HDL-C plasma levels and as a result, with the risk of CAD. Importantly, this CETP variant does not influence the response to pravastatin therapy.

Key Words: genetics • cholesterol • lipoproteins • coronary disease

	Introduction

Top Abstract Introduction Methods Results Discussion Conclusions References

 
A strong inverse relation exists between HDL cholesterol (HDL-C) plasma levels and the risk of coronary artery disease (CAD).^1,2 Cholesteryl ester transfer protein (CETP) plays a central role in HDL-C metabolism by shuttling cholesteryl esters from HDL particles to apolipoprotein B–containing particles in exchange for triglycerides.^3,4 A common polymorphism in intron 1 of the CETP gene, denoted TaqIB,⁵ was among the first genetic variations to be associated with HDL-C plasma levels.⁶ The less common B2 allele occurs at a frequency of 40% and is associated with lower CETP levels compared with the more common B1 allele. Since its first description, a relation between the TaqIB genotype and HDL-C levels and the risk of CAD has been investigated in numerous population-based studies, albeit without consistent results. In addition, a pharmacogenetic interaction has been implicated between TaqIB genotype and pravastatin treatment,⁷ but this observation has not been found in other studies.^8,9 Such inconsistencies are often encountered among genetic association studies and can be due factors such as genetic admixture or lack of statistical power.¹⁰ In this specific case, an alternative explanation may derive from proposed interactions with sex,^11,12 smoking,^8,13 body mass index (BMI),^13,14 and use of alcohol,¹⁵ all modifiers of CETP action, HDL-C levels, and CAD risk. In view of our recent report about an interaction between CETP plasma levels and triglyceride levels,¹⁶ we also explored a potential interaction between TaqIB genotype and triglyceride levels.

We hypothesized that a meta-analysis could give better insight into the relations among CETP TaqIB genotype, HDL-C levels, and the risk of CAD. However, a conventional meta-analysis does not allow for the assessment of the interactions mentioned earlier. We therefore performed a meta-analysis of individual patient data. To circumvent the risk of publication bias, we included only those studies that were likely to be published irrespective of the results, ie, randomized, controlled trials and large, population-based studies. This approach is likely to yield a less biased result and a more modest effect estimate compared with a meta-analysis that also includes small studies.¹⁷ Using this criterion and taking into account the obligatory presence of CETP TaqIB genotyping data, we were able to analyze data from a total of 13 677 individuals.

	Methods

Top Abstract Introduction Methods Results Discussion Conclusions References

 
Literature Search
We identified all population-based studies published before September 2003 on the CETP TaqIB polymorphism and its association with HDL-C plasma levels and/or the risk of CAD. The literature was scanned by a formal search of the MEDLINE electronic database. The search terms were both MeSH terms and (part of) the text words "cholesteryl ester transfer protein" or "CETP," in combination with "polymorphism," "mutation," or "genetics." Reference lists of retrieved articles were scanned for additional potentially relevant publications. Finally, for each retrieved publication, an electronic "cited reference search" was performed (Web of Science version 4.1.1, Institute for Scientific Information 2000, available at http://www.isinet.com/isi/products/citation/wos/), identifying all articles citing the index publication.

Selection Criteria
The meta-analysis was limited to population-based studies that included >500 individuals. In addition, we included all randomized, double-blind, placebo-controlled trials of pravastatin treatment from which subanalyses on the role of the TaqIB polymorphism were performed. To avoid population admixture due to genetic heterogeneity between races, only white subjects were used in the analyses. All identified publications were independently evaluated by 2 investigators for compliance with these criteria, and the results were compared. Disagreements were very few and were resolved by discussion and rereading of the original manuscripts.

Data Collection
Principal investigators from all qualifying studies were contacted and asked to provide a database with individual patient data. Variables from the individual databases were made compatible with each other and entered into a pooled database. Again, this procedure was performed in duplicate by 2 independent investigators, the results were compared, and disagreements were resolved by consensus. An "HDL database" was composed of all studies that assessed the relation between TaqIB genotype and HDL-C plasma levels. A "CAD database" was composed of all studies that assessed the risk of prevalent or incident CAD and used (apparently) healthy individuals as a referent group. A "pravastatin database" was composed of all trials that assessed the efficacy of pravastatin therapy.

Data Analysis
A fixed-effects, linear regression model was used to assess the relation between patient characteristics and HDL-C plasma levels, adjusted for study, sex, and age.¹⁸ The variables that had a significant association with HDL-C plasma levels and all 2-way interactions between them were entered into a multivariate linear regression model, and backward stepwise selection was used to identify the variables that had an independent association with HDL-C plasma levels. A fixed-effects logistic regression model was used to assess the relation between patient characteristics and the risk of CAD.¹⁹ Odds ratios (ORs) with corresponding 95%confidence intervals (CIs) were calculated for the risk of CAD in B1B2 individuals and B2B2 individuals, with B1B1 individuals as the reference group.

Both for the outcome HDL-C and for the outcome CAD risk, we tested for interactions between TaqIB genotype and several environmental factors. To assess whether we had sufficient statistical power to detect interactions of the magnitude that have been reported in the literature, we calculated the magnitude of interactions and corresponding 95% CIs by using the pooled databases. We investigated interactions between TaqIB genotype (B1B1 individuals versus B2B2 individuals) and risk factors as dichotomous variables (men versus women; smokers versus nonsmokers; use of alcohol, yes versus no; BMI 27 kg/m² versus <27 kg/m²; triglycerides 1.7 mmol/L versus <1.7 mmol/L; pravastatin use, yes versus no). To adjust for heterogeneity between studies, all regression models were adjusted for study number as a fixed factor in the model. In addition, to adjust for residual heterogeneity, we used hierarchical random-effects models for the outcomes HDL-C and CAD risk, with study number as a random factor. Additional analyses were performed to assess the consistency between prospective and cross-sectional studies and to investigate whether an additive, recessive, or dominant genetic model was the best way to describe the observed data. Only individuals with a complete data set were entered in the linear and logistic regression models. A probability value <0.05 was considered to indicate statistical significance. Analyses were performed using SPSS version 12.0 (SPSS Inc) and SAS software (SAS Institute).

	Results

Top Abstract Introduction Methods Results Discussion Conclusions References

 
Our MEDLINE search identified 425 potentially relevant publications. A total of 9 population-based studies qualified for inclusion: 4 prospective studies: the Framingham Offspring Study (FOS),²⁰ the Physicians’ Health Study (PHS),²¹ the Northwick Park Heart Study (NPHS),²² and the Reykjavik study;²³ and 5 cross-sectional studies: the European Atherosclerosis Research Study (EARS),²⁴ Etude Cas-Témoins de l’Infarctus du Myocarde (ECTIM),^15,25,26 the Oulu Project Elucidating Risk of Atherosclerosis (OPERA),^11,27 a study performed by Arca et al,²⁸ and one performed by Corella et al.²⁹ Databases with individual patient data were obtained from all studies except for the FOS and the study performed by Corella et al^20,29; thus these 2 studies were not included. We identified 3 randomized, double-blind, placebo-controlled trials of pravastatin treatment wherein a subanalysis had been performed on the role of the TaqIB polymorphism: the Regression Growth Evaluation Statin Study (REGRESS),⁷ the Cholesterol And Recurrent Events trial (CARE),⁹ and the West of Scotland Coronary Prevention Study (WOSCOPS).⁸ The study databases with individual patient data were obtained from all 3. The characteristics of the selected studies are summarized in Table 1. Four studies matched a referent group to the cases: PHS, WOSCOPS (both age- and smoking-matched), ECTIM, and EARS (both age-matched). The patient characteristics of the individual studies are summarized in Table 2. In all individual studies, the genotype distributions were within the expected range according to Hardy-Weinberg equilibrium. Figure 1 shows the relation between TaqIB genotype and HDL-C levels for each of the included studies. Figure 2 shows the relation between TaqIB genotype and CAD risk for each of the included studies.

View this table: [in this window] [in a new window]   TABLE 1. Characteristics of Population-Based Studies and Randomized, Controlled Trials Included in the Present Meta-Analysis

View this table: [in this window] [in a new window]   TABLE 2. Characteristics of Study Samples Included in the Meta-Analysis

View larger version (28K): [in this window] [in a new window]   Figure 1. HDL-C levels according to TaqIB genotype in individual studies. Graphical representation of absolute HDL-C difference between B2B2 vs B1B1 individuals. Mean±SD levels of HDL-C and number of observations in parentheses are shown for B1B1, B1B2, and B2B2 individuals. Data shown here were defined by inclusion criteria of present meta-analysis and were used in meta-analysis. Thus, data may differ from those in original publication, for instance, because present meta-analysis was restricted to whites only. Abbreviations are as defined in text.

View larger version (17K): [in this window] [in a new window]   Figure 2. ORs for CAD according to TaqIB genotype in individual studies. Graphical representation of OR for CAD among B2B2 compared with B1B1 individuals. For each study, percentages of individuals with CAD and numbers of cases and controls in parentheses are shown for B1B1, B1B2, and B2B2 individuals. Data shown here were defined by inclusion criteria of present meta-analysis and were used in meta-analysis. Thus, data may differ from those in original publication, for instance, because present meta-analysis was restricted to whites only Abbreviations are as defined in text.

Pooled Databases
The HDL database comprised 13 677 individuals from the following studies: REGRESS, CARE, WOSCOPS, PHS, ECTIM, OPERA, EARS, NPHS, Reykjavik, and Arca et al. The Reykjavik study (diabetes and use of alcohol) and the study by Arca et al (use of alcohol) did not record all risk factors used in the fully adjusted models. These 2 studies were therefore excluded from the fully adjusted models, which now included a total of 11 751 individuals. The CAD database comprised 8815 individuals from the following studies: PHS, ECTIM, OPERA, NPHS, WOSCOPS, Reykjavik, and Arca et al Again, the Reykjavik study and the one by Arca et al were excluded from the fully adjusted models, which now included a total of 6889 individuals. The pravastatin database comprised 5691 individuals randomized in the following placebo-controlled pravastatin trials: REGRESS, CARE, and WOSCOPS. The baseline characteristics of the pooled databases used for the analyses are summarized in Table 3.

View this table: [in this window] [in a new window]   TABLE 3. Characteristics of Pooled Study Databases Used for the Present Meta-Analysis

CETP TaqIB Genotype and HDL-C Plasma Levels
After adjusting for study, sex, and age, TaqIB genotype was significantly associated with HDL-C plasma levels (P<0.0001). Mean HDL-C levels were 1.05 (±0.32), 1.08 (±0.33), and 1.1 (±0.33) mmol/L in B1B1, B1B2, and B2B2 individuals, respectively. Among men, the respective values were 1.02 (±0.29), 1.05 (±0.30), and 1.12 (±0.3) mmol/L, and among women they were 1.26 (±0.39), 1.32 (±0.39), and 1.41 (±0.40) mmol/L. Other variables that had a significant association with HDL-C plasma levels (adjusted for study, sex, and age) were smoking, diabetes, prevalence of CAD, BMI, total cholesterol, LDL-C, (logarithmically transformed) triglycerides, and use of alcohol (P<0.0001 for each). TaqIB genotype (adjusted for study, sex, and age) was also significantly associated with total cholesterol (P=0.01) and triglyceride levels (logarithmically transformed, P=0.02), so we did not adjust for these variables in subsequent analyses in this section. Mean total cholesterol levels for B1B1, B1B2, and B2B2 individuals were 5.75 (±1.05), 5.77 (±1.07), and 5.82 (±1.06) mmol/L and mean triglyceride levels were 1.83 (±1.01), 1.80 (±1.02), 1.79 (±1.17) mmol/L, respectively.

After adjustment for study, sex, age, BMI, diabetes, smoking, LDL-C, and use of alcohol, TaqIB genotype still exhibited a highly significant association with HDL-C among cases and among people without CAD. No significant interaction was observed between CAD prevalence and TaqIB genotype for HDL-C levels, so people with and without CAD were combined. In the combined dataset, after adjusting for study, sex, age, BMI, diabetes, smoking, LDL-C, use of alcohol, and CAD prevalence, B1B2 individuals had 0.04 mmol/L (0.03 to 0.05, P<0.0001) higher HDL-C levels and B2B2 individuals had 0.11 mmol/L (0.10 to 0.12, P<0.0001) higher HDL-C levels compared with B1B1 individuals. There was no significant heterogeneity between the studies for the relation between TaqIB genotype and HDL-C levels (P=0.9). There was, however, significant heterogeneity between the studies for absolute HDL-C levels, but when we adjusted for this in a random-effects model, the results were similar to those in the fixed-effects model: HDL-C levels were 0.04 mmol/L (0.03 to 0.05, P<0.0001) higher among B1B2 individuals and 0.11 mmol/L (0.10 to 0.12, P<0.0001) higher among B2B2 individuals than in B1B1 individuals. The relation between the number of B2 alleles and HDL-C levels deviated significantly from linearity (P=0.002). Given this observation and the absolute HDL-C differences observed in B1B2 and B2B2 individuals compared with B1B1 individuals, we conclude that B2 homozygosity has a higher penetrance than heterozygosity.

Subsequently, we tested for gene-environment interactions between TaqIB genotype and several risk factors that have been previously reported to show an interaction with TaqIB genotype: sex, BMI, smoking, and use of alcohol. In addition, we tested for an interaction with triglyceride levels. Significant interactions with TaqIB genotype (B2B2 versus B1B1 individuals) were identified for both sex and smoking (P=0.02 for each), such that the relation was attenuated among men and among smokers (Table 4). The adjusted HDL-C difference in B2B2 individuals relative to B1B1 individuals was 0.16 mmol/L (0.10 to 0.22) among women and 0.10 mmol/L (0.09 to 0.12) among men. The corresponding values were 0.11 mmol/L (0.09 to 0.13) and 0.09 mmol/L (0.06 to 0.12) among nonsmokers and smokers, respectively. These interactions were no longer statistically significant when we adjusted for the testing of multiple hypotheses. We did not observe a significant interaction between TaqIB genotype and any other variable.

View this table: [in this window] [in a new window]   TABLE 4. Interaction Between TaqIB Genotype and Several Risk Factors for the Outcomes HDL-C and CAD Risk

CETP TaqIB Genotype and Risk of CAD
When we adjusted for study number, sex, and age, the TaqIB genotype was significantly associated with the risk of CAD (P for linearity=0.001). The OR for CAD was 0.93 (0.84 to 1.04, P=0.2) in B1B2 individuals and 0.77 (0.66 to 0.89, P=0.001) in B2B2 individuals compared with B1B1 individuals. Other significant predictors of the risk of CAD were smoking, diabetes, BMI, systolic and diastolic blood pressure, total cholesterol, LDL-C, HDL-C, triglyceride levels (logarithmically transformed), and the use of alcohol. Because TaqIB genotype was associated with HDL-C and triglyceride levels (logarithmically transformed), we did not adjust for these variables in the subsequent analysis in this section.

After adjustment for study, sex, age, smoking, diabetes, BMI, systolic blood pressure, LDL-C, and use of alcohol, TaqIB genotype was still significantly associated with a lower risk of CAD: OR=0.95 (0.83 to 1.07) in B1B2 individuals and OR=0.78 (0.66 to 0.93) in B2B2 individuals compared with B1B1 individuals (P for linearity=0.008). On additional adjustment for HDL-C, this association was considerably attenuated and became statistically nonsignificant: OR=1.04 (0.90 to 1.19) in B1B2 individuals and OR=0.92 (0.77 to 1.11) in B2B2 individuals compared with B1B1 individuals (P for linearity=0.4). There was no significant heterogeneity between the studies for the relation between TaqIB genotype and CAD risk (P=0.8). There was significant heterogeneity between the studies for absolute CAD risk, but when we adjusted for this in the random-effects model, the point estimates were similar to those in the fixed-effects model. However, the 95% CIs were importantly narrower, causing the risk estimate for B1B2 individuals to become significant: OR=0.96 (0.92 to 0.99, P=0.03) in B1B2 individuals and OR=0.82 (0.75 to 0.91, P<0.0001) in B2B2 individuals compared with B1B1 individuals.

The relation between the number of B2 alleles and CAD risk did not deviate significantly from linearity (P=0.3). This implies that an additive model and not a dominant or recessive one is the best genetic model to describe the data. To assess the consistency between cross-sectional and prospective studies, we excluded the cross-sectional studies and retained only the prospective studies. The point estimates did not change substantially (OR=0.97, 0.79 to 1.21 in B1B2 individuals and OR=0.79, 0.59 to 1.05 in B2B2 individuals), but due to the loss of statistical power, the CIs widened and now included unity (P=0.1).

As described earlier for the outcome HDL-C, we tested whether gene-environment interactions existed betweenTaqIB genotype and traditional risk factors for the outcome CAD risk. No significant interaction terms were observed. As for HDL-C levels, smoking again appeared to attenuate the relation between TaqIB genotype and CAD risk, but this interaction was not statistically significant (P=0.2).

Relation Between CETP TaqIB Genotype and Efficacy of Pravastatin Therapy
In the pravastatin database, pravastatin significantly reduced the risk of incident CAD: OR=0.68 (0.58 to 0.78), P<0.0001. In this database, TaqIB genotype was not significantly associated with CAD risk: OR=1.06 (0.90 to 1.25) in B1B2 individuals and OR=0.93 (0.74 to 1.16) in B2B2 individuals relative to B1B1 individuals (P=0.5). No significant interaction was observed between TaqIB genotype and pravastatin therapy (P=0.7).

	Discussion

Top Abstract Introduction Methods Results Discussion Conclusions References

 
In this meta-analysis, we observed that after adjustment for study, age, sex, smoking, BMI, diabetes, LDL-C, use of alcohol, and prevalence of CAD, the TaqIB variant in the CETP gene still exhibited a highly significant association with HDL-C levels (P<0.0001). In quantitative terms, B2B2 individuals had 0.11 mmol/L (0.10 to 0.12, P<0.0001) higher HDL-C values than did B1B1 individuals. Second and more important, we observed that after adjustment for study, sex, age, smoking, BMI, diabetes, systolic blood pressure, LDL-C, and use of alcohol, the CETP genotype was significantly associated with the risk of CAD: OR=0.78 (0.66 to 0.93) in B2B2 individuals compared with B1B1 individuals (P for linearity=0.008). Finally, we could eliminate the notion of a significant interaction between TaqIB genotype and pravastatin in reducing the risk of cardiovascular events.

Since the discovery of CETP as a modulator of HDL-C levels, there has been much speculation about its role in human atherogenesis.^4,30 Because of its high frequency among whites and based on some reports of a strong association with HDL-C, the TaqIB variant in the CETP gene has played a prominent role in genetic association studies investigating the relation between CETP activity, lipids, and CAD risk. This variant, located in intron 1, has been suggested to act as a marker for a functional CA polymorphism in the promoter region of the CETP gene, located 629 bp upstream from the transcription start site.³¹ This –629CA variant has been shown to directly affect CETP promoter activity.³² However, not all studies are consistent with this.²³ Because almost complete linkage disequilibrium exists between the TaqIB and the –629CA gene variants³¹ and because of the extent of data available on TaqIB and not on –629CA, the first is at present the best candidate for a meta-analysis.

TaqIB Genotype, HDL-C Levels, and Risk of CAD
The results of the present meta-analysis support a strong relation among the TaqIB variant, HDL-C levels, and CAD risk. Of note, 2 studies that qualified for inclusion in our meta-analysis but from which the data could not be obtained both reported a significant relation with CAD that is consistent with the present results.^20,29 In other words, it is unlikely that additional inclusion of these 2 studies would have affected the evident relation between TaqIB genotype and CAD risk.

We observed that the significant association between TaqIB genotype and CAD risk was substantially attenuated on adjustment for HDL-C levels, suggesting that the association between TaqIB genotype and CAD risk is largely mediated through HDL-C levels. However, the risk estimate did not completely return to unity (OR=0.92, 0.77 to 1.11), so this observation does not confirm an association between CETP genotype and CAD risk independent of HDL-C levels, nor does it rule out a small residual association. Such a relation between CETP genotype and CAD risk independent of HDL-C levels would be consistent with several previous studies.^7,27,33 The inconsistency with our results may derive from the fact that those studies assessed different CETP variants,^27,33 used surrogate markers of CAD,^7,27 or studied individuals with prevalent CAD.^7,33

Effect Modification by Environmental Factors
We investigated whether our pooled dataset provided any evidence for the published interactions between TaqIB genotype and sex,^11,12 smoking,^8,13 BMI,^13,14 and use of alcohol.¹⁵ The present meta-analysis provides some evidence that interactions exist for both sex and smoking (P=0.02 for each), such that the effect of TaqIB genotype on HDL-C levels was attenuated among men and among smokers. However, these observations must be interpreted with caution because of the multitude of interaction terms tested in this analysis. In addition, assessment of the TaqIBxsmoking interaction may have been influenced by the fact that 2 large studies included controls that were matched to CAD cases based on smoking habit.^8,21 Effects of smoking on HDL-C plasma levels are well established,³⁴ but studies that investigated whether these effects are related to CETP activity have not yielded consistent results; smoking has been reported to reduce CETP activity,³⁵ to increase CETP activity in diabetics,³⁶ and to have no effect on CETP activity.³⁷ These inconsistencies may derive from differences between study samples or from differences between the complex CETP activity assays that were used. The significant TaqIBxsex interaction observed in this meta-analysis is consistent with a publication from the OPERA study (which was included in this meta-analysis) suggesting that the relation between TaqIB genotype and HDL-C levels is stronger in women.¹¹ The regression coefficient of the significant TaqIBxsex interaction in OPERA was 0.148, which is substantially larger than the one calculated in our pooled database, but the 95% CIs overlap, so the results are compatible with each other. It is also consistent with the observation that the relation between 2 other CETP variants in almost complete linkage disequilibrium with TaqIB (A373P and R451Q) and CAD risk is stronger among women not receiving hormone replacement therapy than in men.³⁸ However, in that study, the interaction term was not statistically significant. Our observation is not consistent with a study that suggested that the TaqIB–HDL-C relation is stronger in men,¹² but that study was performed among Japanese, whereas the present meta-analysis was restricted to whites.

In addition to smoking and male sex, obesity, visceral fat accumulation, and hyperinsulinemia have also been reported to blunt the effect of TaqIB genotype on HDL-C levels.^13,14 In the present analysis, we were only able to study an interaction with BMI, which we did not detect. Given the fact that the interactions described in the literature were of greater magnitude than the detectable effect size shown in Table 4, we conclude that an interaction between TaqIB genotype and BMI is unlikely. In the ECTIM study, an interaction was observed between the use of alcohol and TaqIB genotype for both HDL-C levels and CAD risk.¹⁵ We observed a strong trend toward an interaction between TaqIB genotype and use of alcohol for HDL-C levels (P=0.07). However, it must be kept in mind that the interaction observed in the ECTIM study was dose dependent and strongest in heavy drinkers. Because most studies included in this meta-analysis did not record alcohol consumption in a detailed manner, we were unable to differentiate between moderate and heavy drinkers and were therefore forced to study alcohol consumption as a dichotomous variable, which may have attenuated our power to detect such an interaction.

In addition to the environmental factors mentioned earlier that have been reported to interact with TaqIB genotype, we also studied a potential interaction with triglyceride levels. Triglycerides, which are substrates for CETP, are known to affect CETP metabolism.³⁹ In fact, we have recently reported an interaction between CETP plasma levels and triglycerides, such that the risk of future CAD increased with increasing CETP levels but only among individuals with high triglyceride levels.¹⁶ However, an interaction between TaqIB genotype and triglyceride levels has never been reported, and we did not observe one in the present analysis either.

Pharmacogenetic Interaction Between TaqIB Genotype and Efficacy of Pravastatin Therapy
The original description of a pharmacogenetic interaction between TaqIB genotype and pravastatin efficacy was based on angiographic progression of coronary atherosclerosis as outcome⁷ and is therefore not comparable to our analyses, which used HDL-C levels and CAD events as outcome. The only study ever to observe a pharmacogenetic interaction for clinical CAD events was published recently (and after the closure of our database), but the effect of TaqIB genotype on statin efficacy was opposite from that expected based on the angiographic trial.⁴⁰ The interaction term between TaqIB genotype (B2B2 versus B1B1) and statin use described in that study was 0.285 (–0.160 to 0.730), which is substantially larger than the one observed in our meta-analysis, although CIs overlap (0.090 [–0.355 to 0.535]). We conclude that a clinically relevant interaction for cardiovascular events does not exist between pravastatin and TaqIB genotype. It warrants mentioning that 2 of the 3 studies were secondary prevention trials,^7,9 which may have had limited statistical power to detect an effect on cardiovascular events, because both the CAD group and the control group consisted of CAD patients; the difference was defined by the incidence of a recurrent cardiovascular event during follow-up. However, an effect of TaqIB genotype on pravastatin efficacy was also not observed in WOSCOPS, which only enrolled patients who were free of clinical CAD.

Considerations
Any meta-analysis carries the risk of publication bias caused by the fact that small studies with positive results are more likely to be published than those with negative results. Therefore, a meta-analysis of large studies will generally yield more conservative results than a meta-analysis of all published studies.¹⁷ In the present meta-analysis, we therefore only included studies that were likely to be published irrespective of the results, ie, subanalyses of randomized, controlled trials and large, population-based studies.

Another drawback of any meta-analysis is the heterogeneity between the included studies in terms of the genetic structure of the studied samples, distribution of other cardiovascular risk factors, and CAD outcome definitions. First, to reduce the potential for population admixture, we restricted our analysis to whites. Indeed, the allele frequencies of the studied samples were very similar (Table 2). Second, to reduce the possibility that differential risk factor distributions would affect our results, we performed our analysis on individual patient data, which gave us the possibility to adjust for potential interacting or confounding variables. Finally, we adopted the outcome definitions used in each study, which included myocardial infarction by various definitions and CAD by various definitions, including both fatal and nonfatal events. This heterogeneity of outcome definitions is likely to have introduced random error, which may only lead to an underestimation of the real effect.

	Conclusions

Top Abstract Introduction Methods Results Discussion Conclusions References

 
In summary, the CETP TaqIB genotype was strongly associated with HDL-C levels and with the risk of CAD. B2B2 individuals had 0.11 mmol/L higher HDL-C plasma levels and consistently a 20% lower risk of CAD than did B1B1 individuals. The relation between TaqIB genotype and CAD risk was substantially attenuated on adjustment for HDL-C levels, suggesting that the association between TaqIB genotype and CAD risk is largely mediated through HDL-C plasma levels. Our observation that individuals with genetically determined lower CETP levels have a substantially lower risk of CAD supports further exploration of the potential benefits of pharmacological inhibition of CETP activity.

	Acknowledgments

 
P.J.T. and S.E.H. acknowledge support from the British Heart Foundation (RG2000/015). G.E. and G.V. were supported by the Icelandic Science Fund. H.K., S.K., and M.J.S. acknowledge support from the Academy of Finland, the Finnish Foundation for Cardiovascular Research, and the Sigrid Juselius Foundation. The WOSCOPS researchers acknowledge generous support from the British Hyperlipidemia Association Research Award and the British Heart Foundation (97/160). The EARS II study was supported by EC Concerted Action BMHI CT92–0206. The ECTIM study was supported by grants from the Squibb Laboratory, the Sandoz Laboratory, the British Heart Foundation, INSERM, the Institut Pasteur-Lille, and the Groupement de Recherches et d’Etudes sur les Génomes (GREG). M.A. was supported by a grant from the Italian Ministry of Health (ICS-120.4/RF99.11). The NPHS II Study was supported by the UK Medical Research Council, National Institutes of Health (HL-33014 and HL-54502), and Du Pont Pharma, Wilmington, Del. J.J.P.K. is an established investigator of the Netherlands Heart Foundation (2000 D039), and J.W.J. is an established clinical investigator of the Netherlands Heart Foundation (2001 D032). The REGRESS, CARE, and WOSCOPS studies were sponsored by Bristol-Myers Squibb Co (Princeton, NJ).

	References

Top Abstract Introduction Methods Results Discussion Conclusions References

 

1. Miller NE, Thelle DS, Forde OH, Mjos OD. The Tromso heart-study: high-density lipoprotein and coronary heart-disease: a prospective case-control study. Lancet. 1977; 1: 965–968.[Medline] [Order article via Infotrieve]
   
2. Wilson PW, D’Agostino RB, Levy D, Belanger AM, Silbershatz H, Kannel W. Prediction of coronary heart disease using risk factor categories. Circulation. 1998; 97: 1837–1847.[Abstract/Free Full Text]
   
3. Tall A. Plasma lipid transfer proteins. Annu Rev Biochem. 1995; 64: 235–257.[CrossRef][Medline] [Order article via Infotrieve]
   
4. Barter PJ, Brewer HB Jr, Chapman MJ, Hennekens CH, Rader DJ, Tall AR. Cholesteryl ester transfer protein: a novel target for raising HDL and inhibiting atherosclerosis. Arterioscler Thromb Vasc Biol. 2003; 23: 160–167.[Abstract/Free Full Text]
   
5. Drayna D, Lawn R. Multiple RFLPs at the human cholesteryl ester transfer protein (CETP) locus. Nucleic Acids Res. 1987; 15: 4698.[Free Full Text]
   
6. Kondo I, Berg K, Drayna D, Lawn R. DNA polymorphism at the locus for human cholesteryl ester transfer protein (CETP) is associated with high density lipoprotein cholesterol and apolipoprotein levels. Clin Genet. 1989; 35: 49–56.[Medline] [Order article via Infotrieve]
   
7. Kuivenhoven JA, Jukema JW, Zwinderman AH, de Knijff P, McPherson R, Bruschke AV, Lie KI, Kastelein JJ. The role of a common variant of the cholesteryl ester transfer protein gene in the progression of coronary atherosclerosis. N Engl J Med. 1998; 338: 86–93.[Abstract/Free Full Text]
   
8. Freeman DJ, Samani NJ, Wilson V, McMahon AD, Braund PS, Cheng S, Caslake MJ, Packard CJ, Gaffney D. A polymorphism of the cholesteryl ester transfer protein gene predicts cardiovascular events in non-smokers in the West of Scotland Coronary Prevention Study. Eur Heart J. 2003; 24: 1833–1842.[Abstract/Free Full Text]
   
9. De Grooth GJ, Zerba KE, Huang SP, Tsuchihashi Z, Kirchgessner T, Belder R, Vishnupad P, Hu B, Klerkx AH, Zwinderman AH, Jukema JW, Sacks FM, Kastelein JJ, Kuivenhoven JA. The cholesteryl ester transfer protein (CETP) TaqIB polymorphism in the cholesterol and recurrent events study: no interaction with the response to pravastatin therapy and no effects on cardiovascular outcome. J Am Coll Cardiol. 2004; 43: 854–857.[Abstract/Free Full Text]
   
10. Ioannidis JPA, Ntzani EE, Trikalinos TA, Contopoulos-Ioannidis DG. Replication validity of genetic association studies. Nat Genet. 2001; 29: 306–309.[CrossRef][Medline] [Order article via Infotrieve]
    
11. Kauma H, Savolainen M, Heikkila R, Rantala AO, Lilja M, Reunanen A, Kesaniemi YA. Sex difference in the regulation of plasma high density lipoprotein cholesterol by genetic and environmental factors. Hum Genet. 1996; 97: 156–162.[CrossRef][Medline] [Order article via Infotrieve]
    
12. Chen J, Yokoyama T, Saito K, Yoshiike N, Date C, Tanaka H. Association of human cholesteryl ester transfer protein-TaqI polymorphisms with serum HDL cholesterol levels in a normolipidemic Japanese rural population. J Epidemiol. 2002; 12: 77–84.[Medline] [Order article via Infotrieve]
    
13. Freeman DJ, Griffin BA, Holmes AP, Lindsay GM, Gaffney D, Packard CJ, Shepherd J. Regulation of plasma HDL cholesterol and subfraction distribution by genetic and environmental factors: associations between the TaqI B RFLP in the CETP gene and smoking and obesity. Arterioscler Thromb. 1994; 14: 336–344.[Abstract/Free Full Text]
    
14. Vohl MC, Lamarche B, Pascot A, Leroux G, Prud’Homme D, Bouchard C, Nadeau A, Despres JP. Contribution of the cholesteryl ester transfer protein gene TaqIB polymorphism to the reduced plasma HDL-cholesterol levels found in abdominal obese men with the features of the insulin resistance syndrome. Int J Obes Relat Metab Disord. 1999; 23: 918–925.[CrossRef][Medline] [Order article via Infotrieve]
    
15. Fumeron F, Betoulle D, Luc G, Behague I, Ricard S, Poirier O, Jemaa R, Evans A, Arveiller D, Marques-Vidal P, Bard JM, Fruchart JF, Ducimetiere P, Apfelbaum M, Cambien F. Alcohol intake modulates the effect of a polymorphism of the cholesteryl ester transfer protein gene on plasma high density lipoprotein and the risk of myocardial infarction. J Clin Invest. 1995; 96: 1664–1671.[Medline] [Order article via Infotrieve]
    
16. Boekholdt SM, Kuivenhoven JA, Wareham NJ, Peters RJ, Jukema JW, Luben R, Bingham SA, Day NE, Kastelein JJ, Khaw KT. Plasma levels of cholesteryl ester transfer protein and the risk of future coronary artery disease in apparently healthy men and women; the prospective EPIC-Norfolk population study. Circulation. 2004; 110: 1418–1423.[Abstract/Free Full Text]
    
17. Ioannidis JP, Trikalinos TA, Ntzani EN, Contopoulos-Ioannidis DG. Genetic associations in large versus small studies: an empirical assessment. Lancet. 2003; 361: 567–571.[CrossRef][Medline] [Order article via Infotrieve]
    
18. Montgomery DC, Peck EA, Vining GG. Introduction to Linear Regression Analysis. 3rd ed. New York, NY: Wiley; 2001.
    
19. Hosmer DW, Lemeshow S. Applied Logistic Regression. New York, NY: Wiley & Sons; 1989.
    
20. Ordovas JM, Cupples A, Corella D, Otvos JD, Osgood D, Martinez A, Lahoz C, Coltell O, Wilson PW, Schaefer EJ. Association of cholesteryl ester transfer protein-TaqIB polymorphism with variations in lipoprotein subclasses and coronary heart disease risk: the Framingham study. Arterioscler Thromb Vasc Biol. 2000; 20: 1323–1329.[Abstract/Free Full Text]
    
21. Liu S, Schmitz C, Stampfer MJ, Sacks FM, Hennekens CH, Lindpaintner K, Ridker PM. A prospective study of TaqIB polymorphism in the gene coding for cholesteryl ester transfer protein and risk of myocardial infarction in middle-aged men. Atherosclerosis. 2002; 161: 469–474.[CrossRef][Medline] [Order article via Infotrieve]
    
22. Talmud PJ, Hawe E, Robertson K, Miller GJ, Miller NE, Humphries SE. Genetic and environmental determinants of plasma high density lipoprotein cholesterol and apolipoprotein AI concentrations in healthy middle-aged men. Ann Hum Genet. 2002; 66: 111–124.[CrossRef][Medline] [Order article via Infotrieve]
    
23. Eiriksdottir G, Bolla MK, Thorsson B, Sigurdsson G, Humphries SE, Gudnason V. The 629A polymorphism in the CETP gene does not explain the association of the Taq1B polymorphism with risk and age of myocardial infarction in Icelandic men. Atherosclerosis. 2001; 159: 187–192.[CrossRef][Medline] [Order article via Infotrieve]
    
24. Gudnason V, Kakko S, Nicaud V, Savolainen MJ, Kesaniemi YA, Tahvanainen E, Humphries S. Cholesteryl ester transfer protein gene effect on CETP activity and plasma high-density lipoprotein in European populations. Eur J Clin Invest. 1999; 29: 116–128.[CrossRef][Medline] [Order article via Infotrieve]
    
25. Corbex M, Poirier O, Fumeron F, Betoulle D, Evans A, Ruidavets JB, Arveiler D, Luc G, Tiret L, Cambien F. Extensive association analysis between the CETP gene and coronary heart disease phenotypes reveals several putative functional polymorphisms and gene-environment interaction. Genet Epidemiol. 2000; 19: 64–80.[CrossRef][Medline] [Order article via Infotrieve]
    
26. Le Goff W, Guerin M, Nicaud V, Dachet C, Luc G, Arveiler D, Ruidavets JB, Evans A, Kee F, Morrison C, Chapman MJ, Thillet J. A novel cholesteryl ester transfer protein promoter polymorphism (–971G/A) associated with plasma high-density lipoprotein cholesterol levels: interaction with the TaqIB and –629C/A polymorphisms. Atherosclerosis. 2002; 161: 269–279.[CrossRef][Medline] [Order article via Infotrieve]
    
27. Kakko S, Tamminen M, Paivansalo M, Kauma H, Rantala AO, Lilja M, Reunanen A, Kesaniemi YA, Savolainen MJ. Variation at the cholesteryl ester transfer protein gene in relation to plasma high density lipoproteins cholesterol levels and carotid intima-media thickness. Eur J Clin Invest. 2001; 31: 593–602.[CrossRef][Medline] [Order article via Infotrieve]
    
28. Arca M, Montali A, Ombres D, Battiloro E, Campagna F, Ricci G, Verna R. Lack of association of the common TaqIB polymorphism in the cholesteryl ester transfer protein gene with angiographically assessed coronary atherosclerosis. Clin Genet. 2001; 60: 374–380.[CrossRef][Medline] [Order article via Infotrieve]
    
29. Corella D, Saiz C, Guillen M, Portoles O, Mulet F, Gonzalez JI, Ordovas JM. Association of TaqIB polymorphism in the cholesteryl ester transfer protein gene with plasma lipid levels in a healthy Spanish population. Atherosclerosis. 2000; 152: 367–376.[CrossRef][Medline] [Order article via Infotrieve]
    
30. Barter PJ, Rye K. Cholesteryl ester transfer protein, high density lipoprotein and arterial disease. Curr Opin Lipidol. 2001; 12: 377–382.[CrossRef][Medline] [Order article via Infotrieve]
    
31. Klerkx AH, Tanck MW, Kastelein JJ, Molhuizen HO, Jukema JW, Zwinderman AH, Kuivenhoven JA. Haplotype analysis of the CETP gene: not TaqIB, but the closely linked –629CA polymorphism and a novel promoter variant are independently associated with CETP concentration. Hum Mol Genet. 2003; 12: 111–123.[Abstract/Free Full Text]
    
32. Dachet C, Poirier O, Cambien F, Chapman J, Rouis M. New functional promoter polymorphism, CETP/–629, in cholesteryl ester transfer protein (CETP) gene related to CETP mass and high density lipoprotein cholesterol levels: role of Sp1/Sp3 in transcriptional regulation. Arterioscler Thromb Vasc Biol. 2000; 20: 507–515.[Abstract/Free Full Text]
    
33. Blankenberg S, Rupprecht HJ, Bickel C, Jiang XC, Poirier O, Lackner KJ, Meyer J, Cambien F, Tiret L, AtheroGene Investigators. Common genetic variation of the cholesteryl ester transfer protein gene strongly predicts future cardiovascular death in patients with coronary artery disease. J Am Coll Cardiol. 2003; 41: 1983–1989.[Abstract/Free Full Text]
    
34. Garrison RJ, Kannel WB, Feinleib M, Castelli WP, McNamara PM, Padgett SJ. Cigarette smoking and HDL cholesterol: the Framingham offspring study. Atherosclerosis. 1978; 30: 17–25.[CrossRef][Medline] [Order article via Infotrieve]
    
35. Freeman DJ, Griffin BA, Murray E, Lindsay GM, Gaffney D, Packard CJ, Shepherd J. Smoking and plasma lipoproteins in man: effects on low density lipoprotein cholesterol levels and high density lipoprotein subfraction distribution. Eur J Clin Invest. 1993; 23: 630–640.[Medline] [Order article via Infotrieve]
    
36. Dullaart RP, Groener JE, Dikkeschei BD, Erkelens DW, Doorenbos H. Elevated cholesteryl ester transfer protein activity in IDDM men who smoke: possible factor for unfavorable lipoprotein profile. Diabetes Care. 1991; 14: 338–341.[Abstract]
    
37. Freeman DJ, Caslake MJ, Griffin BA, Hinnie J, Tan CE, Watson TD, Packard CJ, Shepherd J. The effect of smoking on post-heparin lipoprotein and hepatic lipase, cholesteryl ester transfer protein and lecithin:cholesterol acyl transferase activities in human plasma. Eur J Clin Invest. 1998; 28: 584–591.[CrossRef][Medline] [Order article via Infotrieve]
    
38. Agerholm-Larsen B, Tybjaerg-Hansen A, Schnohr P, Steffensen R, Nordestgaard BG. Common cholesteryl ester transfer protein mutations, decreased HDL cholesterol, and possible decreased risk of ischemic heart disease: the Copenhagen City Heart Study. Circulation. 2000; 102: 2197–2203.[Abstract/Free Full Text]
    
39. Guerin M, Le Goff W, Lassel TS, van Tol A, Steiner G, Chapman MJ. Atherogenic role of elevated CE transfer from HDL to VLDL₁ and dense LDL in type 2 diabetes: impact of the degree of triglyceridemia. Arterioscler Thromb Vasc Biol. 2001; 21: 282–288.[Abstract/Free Full Text]
    
40. Carlquist JF, Muhlestein JB, Horne BD, Hart NI, Bair TL, Molhuizen HO, Anderson JL. The cholesteryl ester transfer protein Taq1B gene polymorphism predicts clinical benefit of statin therapy in patients with significant coronary artery disease. Am Heart J. 2003; 146: 1007–1014.[CrossRef][Medline] [Order article via Infotrieve]
    

Related Article:

Issue Highlights
Circulation 2005 111: 245. [Full Text]

This article has been cited by other articles:

				A. Thompson, E. Di Angelantonio, N. Sarwar, S. Erqou, D. Saleheen, R. P. F. Dullaart, B. Keavney, Z. Ye, and J. Danesh Association of Cholesteryl Ester Transfer Protein Genotypes With CETP Mass and Activity, Lipid Levels, and Coronary Risk JAMA, June 18, 2008; 299(23): 2777 - 2788. [Abstract] [Full Text] [PDF]

				P. W. F. Wilson CETP Genes, Metabolic Effects, and Coronary Disease Risk JAMA, June 18, 2008; 299(23): 2795 - 2796. [Full Text] [PDF]

				M. P. LaValley Logistic Regression Circulation, May 6, 2008; 117(18): 2395 - 2399. [Full Text] [PDF]

				S. Kathiresan, O. Melander, D. Anevski, C. Guiducci, N. P. Burtt, C. Roos, J. N. Hirschhorn, G. Berglund, B. Hedblad, L. Groop, et al. Polymorphisms Associated with Cholesterol and Risk of Cardiovascular Events N. Engl. J. Med., March 20, 2008; 358(12): 1240 - 1249. [Abstract] [Full Text] [PDF]

				M. K. Jensen, K. J. Mukamal, K. Overvad, and E. B. Rimm Alcohol consumption, TaqIB polymorphism of cholesteryl ester transfer protein, high-density lipoprotein cholesterol, and risk of coronary heart disease in men and women Eur. Heart J., January 1, 2008; 29(1): 104 - 112. [Abstract] [Full Text] [PDF]

				M. Zeller, D. Masson, M. Farnier, L. Lorgis, V. Deckert, J.-P. Pais de Barros, C. Desrumaux, P. Sicard, J. Grober, D. Blache, et al. High Serum Cholesteryl Ester Transfer Rates and Small High-Density Lipoproteins Are Associated With Young Age in Patients With Acute Myocardial Infarction J. Am. Coll. Cardiol., November 13, 2007; 50(20): 1948 - 1955. [Abstract] [Full Text] [PDF]

				T. Y Li, C. Zhang, F. W Asselbergs, L. Qi, E. Rimm, D. J Hunter, and F. B Hu Interaction between dietary fat intake and the cholesterol ester transfer protein TaqIB polymorphism in relation to HDL-cholesterol concentrations among US diabetic men Am. J. Clinical Nutrition, November 1, 2007; 86(5): 1524 - 1529. [Abstract] [Full Text] [PDF]

I. Porch