Signals Blog

The concept of personalized medicine is an intuitive one: knowing what treatment to provide a patient based on their own individual case of a disease. Molecular techniques and various flavours of “-omics” provide high precision in determining the status and types of many diseases, as well as our susceptibility to them. One key application of personalized medicine is to predict individual responses to chemotherapy prior to initiation – thereby reducing the number of patients on ineffective treatments while also contributing to the overall efficiency of health systems.

A dramatic example is provided by Avastin (bevacizumab), an anti-VEGF monoclonal antibody which costs up to $100,000 per year of treatment but where fewer than half of patients show a significant response. In Canada, most provincial health plans cover the cost of treatment, but in British Columbia, Avastin is currently funded on a case by case basis by the BC Cancer Agency. Though decisions depend on individual patient characteristics, they also depend on the availability of funding for the drug. This latter requirement generated enough contention that it was recently the focus in a Canadian television investigative report aired earlier this year. Better predictors of patient response to treatment regimes, particularly expensive ones, would reduce the dependence of treatment decisions on resources and potentially clarify motives behind health plan funding policies. In the case of Avastin, BG Medicine began investigating for early biomarkers to test prospective patients for positive response in 2007, but as of 2010 a straightforward test remains undiscovered. Presumably there are plenty of similar opportunities for development of future tests for drug response, particularly when medications are expensive to deploy.

Paradoxically, the concept that predictive tests can actually make significant contributions during pharmaceutical development is being realized.

In a recent Science Translational Medicine commentary, Dr. Joseph Nevins from the Duke Institute for Genome Sciences and Policy uses Herceptin (trastuzumab) as an example to concisely describe the problem created by the lack of predictive tests:

“Trastuzumab’s target, the HER2 growth factor receptor, is overexpressed in only about 20% of breast cancers, and of these, only about 30% respond to the drug. Thus, if trastuzumab had been developed in an unselected population of patients with breast cancer, its effectiveness would probably not have been detectable.”

The article explains that the difficulty in selecting populations responding well above thresholds of statistical significance results in larger clinical trials being organized at Phase II and III stages, with all their associated expenses and logistical challenges, delaying progression through the drug development cycle.

Most significantly, the use of unenriched populations can cause clinical results to under-represent a particular compounds’ effectiveness if an approach to personalize treatment based on molecular subgroups could be pursued. Survival improvements can be identified in subgroups retrospectively: for example, using the presence of EGFR mutations to identify enhanced response to gefitinib in patients with lung cancer, and KRAS wild type status to predict response to cetuximab in patients with colorectal cancer. Nevertheless, in extreme cases the lack of personalized medicine approaches during initial stages of development can mask the true activity of drugs and contribute to the abandonment of many promising treatments that would have otherwise performed well.

The following two tabs change content below.

Paul Krzyzanowski

Paul is a computational biologist and writer living in Toronto. He's been a contributor to Signals for three years, writing articles for the general public about how biotechnology and biomedical research can be used to solve pressing medical problems. Alongside Paul's experience in computational biology,
 bioinformatics, and molecular genetics, he's interested in how academic research develops into real world, commercial technology, and what's needed for the Canadian biotech industry needs to grow. Paul is currently a Post-doctoral Fellow at the Ontario Institute of Cancer Research. Prior to joining the OICR, he worked at the Ottawa Hospital Research 
Institute and earned a Ph.D. from the University of Ottawa, specializing in computational biology. And finally, Paul earned an H.B.Sc. from the University of Toronto a long time ago. Paul's blog can be read at