Genetic association studies.

To the Editor: I am delighted to see that Arteriosclerosis, Thrombosis, and Vascular Biology has taken the lead in initiating an open discussion about genetic association studies in such a prominent manner. As correctly pointed out in the editorial by Dr Hegele,1 the number of publications has exponentially increased in the last decade. Despite the intrinsic problems with this approach, this exponential growth also demonstrates the popularity of this type of study. Anyone who follows the literature in genetic association studies will recognize, as Dr Hegele does, that none of the polymorphisms reported so far has been consistently associated with any phenotypic trait, either a single quantitative trait such as triglyceride level or a complex trait such as atherosclerosis. A typical example is the polymorphisms in endothelial NO synthase,2 which has been associated with various vascular diseases in some studies but not in others. Among many reasons for such inconsistent association studies, the importance of controlling for environmental influence should be specifically emphasized because it can modify the phenotypic effect of a given genotype and potentially alter the direction of the association.3,4 However, we should also not deny the fact that these association studies have stimulated many mechanistic investigations, often from a biochemically unconventional angle, which have the potential for novel findings. While I agree with most criteria set out by Dr Hegele1 and Drs Almasy and MacCluer,5 particularly multi-locus approaches and haplotype analyses, one must be aware that not many research groups have the resources to satisfy them all. This is particularly true in some countries with small research budgets, where investigators often have access to patients with unique phenotypes and ethic backgrounds. Denying public access to these unique data, although they may be statistically powerless, would minimize our chances for some novel discoveries. I applaud the suggestion of a shorter or electronic format for publication and support the criteria set out by Dr Hegele. I suggest that we should take one step further in a spirit similar to that of GenBank. Authors could be asked, when they submit a manuscript to ATVB, that they also deposit the original data in Excel (Microsoft) format to a specific ATVB database with some standardized entries such as age, sex, ethnicity. This database could be established under a heading for each specific phenotype, such as hypercholesterolemia. When the database has accumulated to a level that statistical power will be adequate, ATVB could commission meta-analyses. This would not only generate some definitive data that would satisfy criteria set down by statisticians, but it would also acknowledge the contributions of individual investigators to the genetic associations.