Coronary artery disease genetics: bigger is better.

The complexity of coronary artery disease genetics arises from the diversity of clinical phenotypes and from the many biological pathways that contribute to atherosclerotic plaque biology, including lipid metabolism, inflammation, endothelial function, oxidative stress and thrombosis. In addition to environmental factors, interindividual variation in disease susceptibility or outcome may arise from common polymorphic variation in the genes encoding the proteins that participate in these and other biological pathways. Twin studies suggest that the total genetic contribution to coronary artery disease risk is substantial, but dissecting out the nature of these effects in a complex polygenic trait with multiple subphenotypes remains a major challenge. In this issue, Murase et al.1 have investigated possible associations between the presence of coronary artery spasm and 35 single nucleotide polymorphisms (SNPs) in 29 candidate genes. In comparing these SNPs in approximately 600 patients with coronary artery spasm with 1600 controls, they observed a significant association between coronary artery spasm in men and an SNP in the NAD(P)H oxidase p22phox gene, and in women with polymorphisms in the stromolysin-1 and interleukin-6 genes. These findings add weight to smaller case-control studies that have found associations between these and other biologically plausible candidate genes and phenotypes such as angiographic coronary artery disease, myocardial infarction or coronary artery spasm.2–4 A longstanding problem with many of these smaller casecontrol studies has been inadequate statistical power (due to small sample sizes) often compounded by selection bias, multiple hypothesis testing and the possibility of imperfect genetic matching between case and control groups. Publication bias towards ‘positive’ results in candidate gene studies with apparent strong biological rationale has fuelled numerous possible associations with disease phenotype that have not been borne out in large-scale studies or meta-analyses, as exemplified by the ACE5 or paraoxonase PON16 genes. The study by Murase et al. is another contribution to a number of recent larger-scale studies that have sought to identify the importance of common genetic variants in coronary artery disease. In an earlier study, these investigators compared 112 individual polymorphisms in 71 candidate genes in 2819 cases with MI and 2242 controls, a total of almost 180,000 genotypes, using a two-step approach to narrow down promising associations. This substantial undertaking implicated SNPs, in the genes encoding connexin 37 in men, and both plasminogenactivator inhibitor-1 and stromelysin-1 in women, as significantly associated with myocardial infarction (MI). One conclusion drawn from both studies is that different susceptibility variants operate in men and women. Although this finding can plausibly be explained biologically (as can just about any result arising from tests of candidate genes) it perhaps more likely reflects the borderline power of studies even of this size. Given the weak genetic signals involved, individual studies have to be serendipitous to detect the association. Individual studies will only identify a subset of the variants involved, and replication in similar sized cohorts (in this case in the other gender group) will not be the norm. In all genetic studies, greater power is achieved not only by greater sample sizes but also by stringent selection of severe phenotypes to enrich for genetic load – studies in ‘average’ patients are much less informative. Additionally, the findings of candidate gene casecontrol studies, even large-scale ones, are potentially limited by phenotyping errors and by selection bias. For example, defining genetic associations with coronary spasm relies upon the validity of detecting and defining coronary spasm in cases and excluding coronary spasm in controls, whilst ensuring that both groups are derived from a similar population. Furthermore, the findings will necessarily be limited to genes that have been chosen, a priori, as known or plausible biological candidates. Accordingly, such studies are usually justified on the basis European Heart Journal (2004) 25, 900–901