Implications of multilocus inheritance for gene-disease association studies

Abstract

The impact of multilocus inheritance on the power of candidate gene association studies and the parameters derived therefrom is considered. Both case-control and family-based control designs are included. When the background heritability H (i.e., residual correlation) among sibs for the disease in question is high, the power of multiplex affected sibships versus singletons is diminished compared with the situation when background heritability is low. Thus, multiplex families for association studies are most advantageous for low heritability disorders. Estimates of genotype relative risk (GRR) are also distorted in multiplex sibships compared with singletons based on background heritability. In a case-control study, GRR is inflated in multiplex families, with the degree of inflation decreasing with background heritability. By contrast, in a TDT (parental control) study, GRR is deflated in multiplex families, with the degree of deflation increasing with background heritability (and no deflation when H is 0). Considering identical twins, the difference in genotype frequencies between concordant versus discordant MZ twin pairs for the candidate gene is large when the heritability is low, but modest when the heritability is high.