PHG Foundation | www.phgfoundation.org

The field of genetic epidemiology has been plagued by the difficulty of achieving independent replication of proposed associations between genetic variants and common, multifactorial diseases. A new study published in JAMA highlights this problem by reporting failure to replicate significant associations for any of 85 different variants in 70 genes that have been previously reported as susceptibility factors for atherosclerosis or for acute coronary syndromes including unstable angina and myocardial infarction [Morgan TM et al. (2007) JAMA 297, 1551-61].

The variants were tested in 811 white patients of European ancestry who presented with these conditions at two US university-affiliated hospitals, and in 650 age- and sex-matched controls. Only one variant, in the control region of the beta-fibrinogen gene, was found to be significantly associated with acute coronary syndromes at the P<0.05 level, but the significance level was marginal (P=0.03). A supplementary analysis to look specifically for weak associations found that only about half of the 85 variants were even marginally more frequent in cases than in controls.

Comment: The negative results from this study do not necessarily mean that positive results from previous studies were wrong. As the authors mention, a recent meta-analysis of 14 of the variants included in their study found odds ratios of 1.10 to 1.73 for risk of myocardial infarction, suggesting that any elevated risk, if present, is likely to be small and may not show up clearly in any but the very largest studies [Cascas JP et al. (2006) Ann Hum Genet 70, 145-69]. It is also possible that discrepancies between the new study and previous reports may be explained by differences between the study populations. Although the negative results from the work of Morgan et al should not necessarily be regarded as definitive, they do at least suggest that caution is needed and that any clinical use of reported associations to assess risk is premature.