SUMMARY

Better science is the way to personalized medicine. It seems like the right time for clinicians to bring a "biological" perspective to the table and to become more engaged in the controversies, debates, and discussions around personalized medicine. This is their domain, albeit a shared domain, with biostatisticians, researchers, and others who are experts in measurement/estimation methodologies and disease diagnosis/progression, respectively. Personalized medicine is truly a multidisciplinary effort. In particular, there are numerous opportunities to utilize the core principles of clinical medicine (e.g., dose–exposure–response) to address the concerns and uncertainties surrounding the associations between genes, genetic variants (e.g., biomarkers), and clinical observations, and to provide a biological, mechanistic, quantitative, and model-based framework to deal with future decision-making in pharmacogenetics. Among other issues, there is a need to thoughtfully reexamine the quality and merits of observational gene association studies. It is impractical to think that all of the needed conclusive evidence of pharmacogenetic associations for drug dosing decisions (i.e., empirical certainty) will come from RCT because of the cost and time involved. It is also naïve to think that observational association studies and mathematical models (i.e., causal certainty) alone will convince those in clinical practice to utilize pharmacogenetics in patient care. Clinicians can better address the perceived unreliability of association studies and believability of models by bringing quantitation to the table in the form of drug–disease models that can (1) assess the strength of evidence of replication from genetic association studies, and (2) address uncertainties around genetic association studies by providing operational probabilities for clinical decisions such as dose selection and expected clinical outcomes.