Kozak sequence polymorphism in the platelet GPIb𝛂 gene is not associated with risk of myocardial infarction

The study by Croft et al is a retrospective case-control study of patients that have survived myocardial infarction showing no differences in allele frequency for the recently described GPIbα Kozak polymorphism between a population of cardiac patients who had survived myocardial infarction and a control population. We have previously shown that the polymorphism correlates with glycoprotein Ib-IX-V complex density on the platelet surface and on the surfaces of transfected cells expressing the complex. The −5C allele,which more closely approximates the consensus sequence derived by Kozak for optimal messenger RNA translation, was associated with increased receptor levels in a genedosage-dependent manner. The differences in receptor density between individuals with the different genotypes was modest, with the C/C homozygotes on average expressing 50% more of the receptor on their platelets than the T/T homozygotes. Although one would expect that a higher receptor level would yield platelets that were more adhesive and hence more prone to initiate thrombosis, the higher receptor levels may also be associated with higher levels of glycocalicin (the soluble extracellular portion of GPIbα) in the plasma, which could modulate any increased adhesiveness of the platelets. Similarly, although gain-of-function mutations of GPIbα associated with platelet-type von Willebrand disease are associated with an increased affinity of the receptor for vWI, they paradoxically cause a bleeding disorder. Nevertheless, it is still possible, and perhaps even likely, that the −5C allele is associated with an increased propensity of those carrying the allele to develop thrombosis. Such an association may not be uncovered in a study such as the one of Croft et al, for several reasons. First, the study is retrospective, which precludes the analysis of an at-risk population for future events. Therefore, analysis has to rely on the identification of the proper case-controls, which are difficult to match randomly. A truly matched control population would be matched for all variables that confer risk for myocardial infarction, which these controls clearly are not. Note, for example, the differences in the percentages of smokers and diabetics in the 2 populations. Second, the study analyzed only survivors of myocardial infarction, which introduces a survival bias; ie, those with the unfavorable genotypes may have not survived the hospitalizations. Finally, those at the greatest risk for myocardial infarction are also likely to be those with the rarest genotype, the C/C homozygotes. In both the case and control populations analyzed by Croft et al the frequency of this genotype was below 2%, with a tendency for a higher frequency in the case population. Thus, it is more appropriate to analyze risk by genotype frequency rather than by allele frequency, with rare genotypes requiring very large populations for analysis.