Impact of overlapping recruitment on linkage analysis of complex disorders: simulation studies

Evidence for significant linkage in complexly inherited disorders usually necessitates independent, replicative studies. This study investigates the implications of including in the replicative studies families already used to suggest linkage in initial linkage analysis. We generated 1,000 unlinked replicates of 100 nuclear families with a complexly inherited disease but with no linkage to the markers analyzed. We then used a standard nonparametric linkage method to analyze these data. From the original 1,000 replicates of the original data set, one set was chosen as it yielded suggestive, but falsely positive, linkage results (LOD score = 3.4). Variable numbers of randomly selected families from this positive replicate (n = 100 families) were used to replace families in replicates of the original (unlinked) data set, and linkage analysis repeated. Overlap of families from the "positive data set" did increase the LOD scores for "unlinked data sets." While a small amount of overlap (replacement) between a positive linkage result and the replication sample is unlikely in practice to alter results, our study suggests that steps should be taken to ensure that overlap is minimized. The implications of this overlapping recruitment on replicative linkage studies are discussed.