Precision medicine has a commitment problem. There’s no question that understanding the biology behind disease can lead to tailored treatments. Take the cancer drug crizotinib, for example. It can extend the life of some of the 7 percent of lung cancer patients who have an abnormality in a particular gene. But right now, there aren’t nearly enough targeteted drugs like it.
Just 5 percent of patients who had DNA from their tumor sequenced – an emerging but costly precision diagnostic – could be matched with targeted drugs, a recent study showed. And that’s just cancer. There are thousands of orphan diseases that scientists don’t yet understand the causes of, let alone have treatments for.
At this rate, there’s little hope that we’ll make it to the finish line. “At the current rate of drug development, it will take 2,000 years before every disease is treatable,” says Christopher Austin, M.D., Director of the National Center for Advancing Translational Sciences (NCATS).
“We live in this paradoxical, bittersweet time when we know more about ourselves and health and disease than ever. But is health care any different now than 30 years ago? The answer is pretty much ‘no.’”
The crux of the problem, he argues, is that the way scientists and doctors bring discoveries from bench to bedside is too haphazard. Procedures vary across labs and institutes, and the fact that there is relatively little conversation between them doesn’t help. There needs to be more consistency when testing drugs, organizing clinical trial networks, and sharing data. These are just a tiny fraction of the troubles that NCATs set out to solve when the organization was put into motion five years ago by the National Institutes of Health. Now, they are defining a brand new research field: the science of advancing translational science, in part by creating checkboxes and determining how best to fill them.
Rakesh Nagarajan, Ph.D., Founder and Chief Biomedical Informatics Officer at Pierian Dx, lays out a major headache in his world: getting all the databases to “talk” to each other. Nagarajan’s group is one of several across the country that uses computational methods to connect genome variations — changes in DNA — to precision treatments. Labs, hospitals and research facilities use different databases to interpret their data — from DNA sequencing information to doctor-written notes on clinical symptoms. Depending on which one they use, they may come up with different solutions.
A misstep, Nagarajan says, can have big implications, like misnaming the cause of a disease. The error can impact how a patient is treated, or confuse development of new therapies. Often such discrepancies aren’t outright mistakes but are caused by databases having different protocols for assigning names. “These databases are growing, but there are conflicts between the databases. Who has got the right answer?” he asks. “You’ve got to try to figure that out.” His group has built what’s called Biomedical Knowledge Base to standardize the petabytes of data that streams in from different sources.
In the not-too-distant future, this issue will become even more complex. “One can imagine the billions of data points to be collected,” says Willard Dere, M.D., Co-Director of the Utah CCTS and Executive Director of the Program in Personalized Health at the University of Utah, referring to additional types of information — including on diet, exercise and other environmental impacts on health — that scientists anticipate incorporating into precision medicine.
“Our scientific leaders hope that systematic evaluation and tools to evaluate large data sets will lead to important new information and discoveries which can effectively be utilized to enhance public health,” says Dere. The Utah CCTS, a member of the NCATS network, is heading regional efforts to apply the same concept to other aspects of translational research, from data analysis to clearly communicated informed patient consent. He says that implementing these steps will hasten research and produce more reliable results.
“How do we improve translational science, not just incrementally, but in a logarithmic way?” says Austin. “Only when translation is turned into a predictive science will clinical success rates improve and cost go down.”