The true nature of pharmacogenomic associations?

Abstract

The intricacies associated with single nucleotide polymorphism (SNP)-based pharmacogenomic studies are just beginning to come to the fore. With global attention turned towards building SNP databases and developing cost-effective genotyping strategies, little regard is given to the matters of complexity, functionality, utility and ethnic diversity of genetic associations revealed in patient populations. A recent study illustrates to what lengths we might need to go to uncover associations between genetic variants and drug response.In their article, Drysdale et al. 1xComplex promoter and coding region β2-adrenergic receptor haplotypes alter receptor expression and predict in vivo responsiveness. Drysdale, C.M. et al. Proc. Natl. Acad. Sci. U.S.A. 2000; 97: 10483–10488Crossref | PubMedSee all References1 describe haplotypes, or combinations of SNPs, in the gene encoding the β2-adrenoceptor (ADRB2), which is associated with the bronchodilating response to the β-adrenoceptor agonist albuterol in asthmatics. Mean responses varied by more than twofold in patients with different pairs of haplotypes. The first point to appreciate is the complex nature of the association. None of the 13 individual SNPs examined was predictive of the response to albuterol. The association was revealed only when haplotypes of the 13 SNPs were examined. Indeed, to make matters more complex, haplotypes by themselves were not as predictive as were specific pairs of haplotypes. What makes this story convincing, in spite of the complexity, is that the authors ascribed a functional consequence to the haplotypes. The pairs of haplotypes in question were shown to correlate with in vitro transcript and protein expression levels of the β2-adrenoceptor, thus lending biological plausibility. Demonstrating an in vitro consequence of associated polymorphisms is arguably more convincing than a statistical association and might in fact be a necessary step in weeding out the false positives from the true associations.These intriguing results beg the question, ‘how could this information be used in a clinical setting?’ Although the difference in efficacy between the most and least responsive pairs of haplotypes is significant (P=0.008), they together comprise only 20% of the Caucasian population. How do we deal with the remaining 80% of patients whose response falls on a gradient between poor and good? If the patients who carry the least responsive haplotype pair do not take the drug, the average response in the remaining population is virtually unchanged. Although this study represents a good first step, more work needs to be done if we are to improve drug efficacy in clinical trials or medical practice. The question of utility is complicated even further by ethnic diversity. The >20-fold differences in haplotype frequencies between major US ethnic groups described in this study is a recurring theme in genetics. The association of ADRB2 haplotypes with response to albuterol was found in Caucasian Americans. One question left unanswered is whether identical haplotypes from different ethnic groups exhibit the same in vitro and in vivo response. If not, other as yet unknown factors that interact with ADRB2 haplotypes might be implicated. In any event, by virtue of their divergent frequencies, the haplotypes might have greater utility for predicting drug response in one ethnic group compared with another. The mantra of pharmacogenomics has been ‘one drug may not fit all’. It might in fact be the case that one pharmacogenomic test might not fit all.