Selecting Predictive Markers for Pharmacogenetic Traits: Tagging vs. Data-Mining Approaches.

Objective: The tagging approach appears as a promising tool to test the association of genetic variants with complex traits such as disease susceptibility or drug response. However, since tag markers are selected only on the basis of inter-marker LD properties, regardless of any phenotypes, it remains unclear to what extent they can be useful to predict variable drug responses, once typed in clinical material. We undertook a study to provide further insights into the usefulness of the tagging approach for selecting phenotype-associated markers relevant to drug response. Methods: Several tagging methods were applied to the genotyping data of two drug-metabolizing enzymes, NAT2 and CYP2D6, and the ability of the selected tagging markers to predict the individual metabolizer status was empirically evaluated. We also assessed the impact of LD levels, tagging thresholds and allele frequencies on tagging efficiency. Results: We found that the functional variation was adequately represented by the selected tagging markers, these latter providing a classification accuracy for the individual metabolizer status close to the maximal 100% value observed with the entire set of polymorphisms. Conclusion: The tagging approach is an interesting approach to select candidate gene markers predictive of drug response in pharmacogenomic studies. Copyright © 2008 S. Karger AG, Basel [ABSTRACT FROM AUTHOR]