A cautionary note on the appropriateness of using a linkage resource for an association study

Utilizing a linkage resource for association analysis requires consideration both of the marker data used and correlations among relatives in pedigrees. We previously developed a method for association testing in pedigrees. We applied our method to 50 replicates of microsatellite data surrounding five genes involved in high-density lipoprotein (HDL) in the Genetic Analysis Workshop 13 (GAW13) simulated data and examined association with HDL as well as linkage disequilibrium (LD) between markers.

Although no association was intentionally simulated, we found significant evidence of weak LD between microsatellite markers (flanking/~5 cM from the genes), in some but not all replicates. This level of LD compared well to that observed in the real GAW13 Framingham data. Only one region had sufficient replicates to assess power, and this was low (12.5-20.8%). More power was attained using all individuals and accounting for relationships, compared with one independent individual/pedigree, although this was not significant due to small sample sizes. Not accounting for relatedness inflated statistical significance (p < 0.0001).

A correction for dependence is necessary in association studies to avoid an inflation of significance probabilities. Our results further illustrate that use of microsatellite marker data is not an effective approach for association testing.