The Winner's Curse

Abstract

Abstract Winners in competitive bidding are losers in that they frequently pay too high a price. This phenomenon has recently been noted in genetic association studies of common diseases. The winner's curse in genetic association studies appears as upward bias in the estimated effect of a newly identified allele on disease risk when the study design lacks sufficient statistical power. The winner's curse manifests mostly in genome‐wide association (GWA) studies in which 300 000–1 000 000 single‐nucleotide polymorphisms are tested. The winner's curse also occurs in meta‐analysis of several GWA data sets. To counter this effect, construction of a large‐scale GWA study or a consortium‐based meta‐analysis of GWA studies that is sufficiently powered to account for the presence of between‐study heterogeneity is required. Key Concepts: ‘Winner's curse’ is named for the phenomenon whereby winners at competitive auctions are likely to pay in excess of the value of the item. Genetic association studies have been conducted to identify susceptibility genes underlying common diseases such as diabetes, schizophrenia and coronary artery disease. In genetic association studies, the winner's curse is the phenomenon whereby the disease risk of a newly identified genetic association is overestimated when the statistical power of original study is not sufficient. The winner's curse implies that the sample size required for confirmatory study will be underestimated, resulting in failure of replication study to corroborate the association. The winner's curse is common in genome‐wide association (GWA) studies because most single‐GWA studies are underpowered to detect small genetic effects at a stringent genome‐wide significance level. In consortium‐based meta‐analyses of several GWA studies having high between‐study heterogeneity, there is an increased probability of the winner's curse. In the discovery phase, construction of a larger scale GWA study or a consortium‐based meta‐analysis of GWA studies that evaluates between‐study heterogeneity is required to reduce probability of occurrence of the winner's curse. In the replication phase, methodologies for reducing bias in the estimates of genetic effect are helpful to calculate the sample sizes required to replicate the discovered associations.