PENTACON - Personalized Therapy

Personalization of drug therapy

Traditionally, our approach to assessing the value of new therapeutic entities has been to balance evidence of efficacy versus risk based on the outcomes of large scale, randomized controlled trials (RCTs). However, this approach seeks large average signals and may miss both substantial benefit (1) and risk (2) in subgroups of the populations under study. Examples of molecularly defined subgroups susceptible to therapeutic benefit from imatinib (Gleevec) and trastuzumab (Herceptin) in cancer (1, 3) have been used to highlight the promise of a more personalized approach to medicine. However, prediction of risk or, indeed, of efficacy in the treatment of common chronic diseases has been elusive.

Physicians, of course, practice medicine on a personalized basis; ideally, they make therapeutic decisions based on their clinical assessment of the individual patient informed by the outcome of the relevant RCTs – evidence based medicine. However, this “evidence” is often inconclusive (4-5) and is overwhelmingly based on the detection of large average effects. Thus, in practice, it is but one factor, along with the, concurrent medications, conventional laboratory tests of organ function and the physician’s clinical evaluation and experience that goes into making a therapeutic decision.

The advent of comprehensive genomic screening at rapidly decreasing cost has catalyzed interest in exploring approaches to personalized medicine. However, the availability of even comprehensive information on an individual’s genome alone will be insufficient to predict most responses to therapeutics. Rather, it is how the changing array of environmental exposures – most of them unrecognized – interacts with an individual’s genome to condition drug exposure and response that will determine the balance between efficacy and risk.

Technologies pertinent to assessment of environmental exposure have also begun to mature. The field of epigenomics affords insight into how the environment interacts with genomic variation in conditioning drug response. Metabolomic signatures may discriminate unanticipated, disparate and potentially dose dependent actions amongst drugs that are traditionally classified based on their effect on a single target. Increasingly, we recognize how variation in the microbiome, in part reflective of our diets, may also influence drug response.

The use of cell based screens and model organisms in drug development have also begun to offer novel opportunities for a strategic approach to the personalization of medicine. Barcoding technologies in yeast may predict toxicity profiles (6) and inbreeding strategies using mice and fish can identify unsuspected genes that modify drug response (7-8). Sensitive and specific quantitative readouts of drug response, biased and unbiased by the presumed mechanism of action, can be combined with measurements of drug exposure and projected across the translational divide. Such translational studies can be used to model dosing regimens and to guide deep phenotyping studies in humans that elucidate factors which contribute to variability in drug response. Computational methods can be constructed from such translational data sets, iteratively enriched by actual human data to derive drug response networks that predict individual variation in response (9).

Despite all of these developments, the fundamental challenge is to harness the science in a way that can add value to the traditional approach of a physician to making therapeutic decisions. Such strategic integration of effort across previously segregated disciplines has yet to be applied to the personalization of medicine. Applying this approach rigorously presents a major challenge – first to elucidate potential signatures of drug response, then to understand the molecular networks evoking these signatures, and finally to test whether they actually add value to the prediction of efficacy and risk.

Personalization of drug therapy

Upcoming Meetings

News

Intranet