An extensive association analysis of a candidate gene for coronary heart disease, Cholesteryl Ester Transfer Protein (CETP) gene, was performed. Ten polymorphisms, out of which three were newly identified in regulatory regions, were investigated for association with myocardial infarction (MI) and 2 MI endophenotypes (CETP mass and HDL-cholesterol level) in 568 MI patients and 668 controls. The polymorphisms affecting codon 405 (Ile(405)Val) and the nucleotide 524 downstream from the stop codon (G(+524)T) were almost completely concordant and associated with plasma CETP mass (P < 0.001). The polymorphisms -629 (located in promoter), intron1 (Taq1B) and intron7 were almost completely concordant and associated with plasma CETP mass (P < 0.0001) and HDL-cholesterol levels (P < 0.0001). This latter association was not found in teetotalers and increased with the quantity of alcohol consumed. Heavy drinkers (>75g/day) homozygous for the (-628)A allele had a reduced risk of MI (OR = 0. 33, P < 0.02). Subjects both homozygous for (451)Arg and heterozygous for (373)Pro had decreased plasma HDL-cholesterol levels and this effect increased with alcohol consumption. The results illustrate the complexity of polymorphism-phenotype associations. They suggest that the CETP gene may carry several functional polymorphisms. Observed interactions between alcohol consumption and polymorphisms associated with HDL-cholesterol level constitute concrete examples of gene-environment interactions. Furthermore, the pattern of association between HDL-cholesterol levels and the polymorphisms at codons 373 and 451 illustrated how two polymorphisms may be confounders (in the usual epidemiological sense) one for the other: their marginal effects are neutralized because of linkage disequilibrium and thus are not detectable by standard univariate association analysis.
Copyright 2000 Wiley-Liss, Inc.