The purpose of our study was to replicate an association between the A387P polymorphism in the thrombospondin-4 (THBS4) gene and myocardial infarction (MI), as previously reported by our group while taking confounding into account, and to assess whether ascertainment by age of onset would affect this association.

We performed a case-control study of 474 white patients with MI (not selected on the basis of age of onset nor family history) and 472 white control subjects. We then applied our findings to our original population of 184 white patients with premature, familial MI and 406 white control subjects.

In the replication population, no significant association was found between THBS4 genotype and MI (P = .41) with univariate analysis. However, after adjusting for age, sex, first-degree family history, and waist-to-hip ratio, an association was apparent in the replication population (P = .032), and the original association became much stronger (P = .00008). Both studies showed a 2.5- to 3-fold increased odds of MI in individuals with the P allele. Furthermore, several variables appeared to modify the effect of THBS4 on MI, including waist-to-hip ratio, diabetes mellitus, and hypertension. When we stratified our cases by age of onset (≤45 years in men, ≤50 years in women), there were no significant differences in genotype frequencies when comparing premature cases with late-onset cases or premature cases with control subjects in either unadjusted or adjusted analyses (all P values >.25).

Our findings suggest that the A387P variant of the THBS4 gene may be an important determinant in the development of MI at any age. Careful assessment of clinical covariates helped to unmask a significant association and therefore may be an important reason for why studies do not replicate.