Dan Roden, M.D., studies the mechanisms underlying variability in response to drug therapy and is principal investigator for BioVU, Vanderbilt’s biorepository. Roden is also William Stokes Professor of Experimental Therapeutics, professor of Medicine and Pharmacology, and director of the Oates Institute for Experimental Therapeutics.
What led to VUMC’s investment in personalized medicine?
There’s our longstanding commitment to clinical pharmacology, with its interest in variable drug response and pharmacogenetics. This is one of the first academic institutions to have recognized the value of information science, and informatics infrastructure, as it applies to health care and biomedical research. And VUMC has always valued translation, the idea that basic research informs clinical research and patient care, and vice-versa. With all that as prologue, the mid 2000s brought an effort to update the VUMC strategic plan, and that’s when personalized medicine emerged as a key discipline for VUMC.
What role does BioVU have in personalized medicine?
BioVU is the largest repository that links DNA samples to electronic medical records at a single academic institution in the world. When we started the biobank, my idea was that we would find genetic markers for disease susceptibility, and perhaps for variable drug responses. But there are all kinds of other questions that you could ask, and that we’re now developing ways of asking. A striking example would be a study that came out last summer from the Department of Biological Sciences, examining Neanderthal remnants in the human genome and associated present-day disease susceptibilities.
What might a more mature personalized medicine delivery system look like?
The notion of personalized medicine is that we will understand enough about an individual to be able to create tailored treatment plans. It could be predicting drug response. But it could also be understanding the socio-cultural environment, and what resources patients have.
Now that it’s becoming cheaper to get lots of genetic information, we have a science question in terms of what do these genetic variants do, and then a question about how to educate providers and patients to understand how best to deal with these data.
How do you go from population data to tailoring clinical practice?
With so many genetic variants and scenarios, it would be difficult to do a randomized clinical trial for every one. More generally, I think evidence will increasingly come from what I call orthogonal data sets: if you discover a signal by trolling through BioVU, the next step may be to ask whether the discovery holds up in a completely different biological system. Newer genome editing techniques make this a less onerous proposition. If your finding holds up in an animal model and in prospective clinical studies, you again may be able to change practice without a randomized controlled trial.