Pharmacogenomics and "individualized drug therapy": high expectations and disappointing achievements.

Since 1965 there have been more than 800 pharmacogenetics/genomics reviews - most suggesting that we are on the verge of offering individualized drug therapy to everyone. However, there are numerous reasons why this approach will be extremely difficult to achieve in the foreseeable future. Drug treatment outcome represents a complex phenotype, encoded by dozens, if not hundreds, of genes, and affected by many environmental factors; therefore, we will almost always see a gradient of response. Phenotyping assays of blood enzyme activities (if feasible) are generally more successful than DNA genotyping for predicting unequivocal outcomes of drug therapy in each and every patient. Phenotyping with probe drugs has generally not succeeded, because of the overlapping substrate specificities not only of drug-metabolizing enzymes but also transporters, receptors, ion channels, transcription factors, and other drug targets; drug-drug interactions, enzyme induction and inhibition, and multiple (enzyme, transporter, second-messenger, signal transduction) pathways also present enormous problems. Genotyping to predict drug disposition, efficacy, toxicity, and clinical outcome has been proposed, but the success of genotyping in individualized drug therapy currently appears unlikely because of the many shortcomings (frequency of DNA variant sites, ethnic differences, admixture) and complexities (plasticity of the genome, multiple mechanisms for determining sizes and locations of haplotype blocks) of this approach. Genomics is an important tool in basic research; yet, it is unrealistic to include genotyping within the realm of tests available to the practicing clinician in the foreseeable future. The same can be said for transcriptomics and proteomics, which also rely on available sources (tumors, biopsies, excreta). The newly emerging fields of metabonomics and phenomics might offer solutions to anticipating and decreasing individual risk for adverse drug reactions in each individual patient; however, tests based on these approaches are not expected to become available to the practicing clinician for at least the next 5-10 years.