The precision medicine initiative proposed by President Barack Obama last week would center on a huge new biobank containing medical records and genetic information for perhaps a million Americans. It would not be created from scratch by enrolling new volunteers, however, but would instead pull together existing studies into one giant database.
That’s according to several scientists familiar with the broad outlines of the project who spoke on background with ScienceInsider. The biobank would be used for studies ranging from finding new disease-gene associations to working out how to use genomic and other molecular information in routine medical care. On Friday, the White House is expected to reveal details of the initiative, which will reportedly cost hundreds of millions of dollars.
The term “precision medicine,” however, is relatively new. It comes from a 2011 report from the National Academies’ National Research Council (NRC) that called for combining medical records and genetic and other molecular data for large groups of people into a single “knowledge network” that would be used for understanding diseases and tailoring treatments.
Keith Yamamoto, a member of the NRC panel and vice chancellor for research at the University of California, San Francisco, insists that precision medicine is not just a new buzzword for “personalized medicine.” Instead, it is a much broader endeavor, because it would integrate a huge range of biological data, for example on model organisms. Both basic researchers and clinicians could draw on the network, he says. “It’s a giant integration mill from which stuff would fall out from the bottom as new knowledge,” says Yamamoto, who has conferred witht the White House Office of Science and Technology Policy on the topic.
As a pilot project, NRC suggested building a large research database with medical and genetic data on 1 million adults. That appears to be reflected in the precision medicine initiative. The plan is to link up existing NIH-sponsored cohort studies and large biobanks created by health care providers. They may range from the famed 67-year-old Framingham Heart Study in Massachusetts funded by the National Heart, Lung, and Blood Institute (NHLBI) to research databases being built by the Marshfield Clinic in Wisconsin and Kaiser Permanente in San Francisco that are linking genetic data with health records.
Some have questioned whether it is feasible to combine medical records from different sources for research—data are often missing or collected in different ways. “There’s a lot of fuzziness,” says cardiac disease researcher Dan Roden, who leads Vanderbilt University’s BioVU biobank in Nashville. But an NIH-funded project called eMERGE that combines medical records from Vanderbilt and eight other medical centers has shown that “it can be done,” Roden says.
One matter to be worked out for a megabiobank is which cohorts to include, says human geneticist David Goldstein of Columbia University, a member of the 2011 NRC panel. For example, “you absolutely must have recontactability,” or permission from patients to be called and asked to come into a clinic for further exams and tests. Some biobanks, such as Vanderbilt’s, do not have that consent from participants, Roden notes.
Assembling a cohort that represents the diversity of the U.S. population will also be important. For that reason, one of the largest planned U.S. biobanks—the Department of Veterans Affairs’ Million Veteran Program—would not be enough, because it’s mostly men.
Another question is whether to sequence the whole genomes of participants, or just the 1% that codes for proteins, which would be cheaper. Researchers will also need to work out ways to share genomic data securely, perhaps drawing on existing efforts to develop standards.
Beyond creation of the biobank will be questions about how useful it will be. Many feel the promise of genomic medicine has not yet been realized. A search for common mutations underlying major diseases found that most such variants raise risks only slightly and are of little help for predicting whether an individual will develop a disease. Pharmacogenomics, or using genetic variants to determine how a patient will respond to a particular drug, has been a mixed success—such testing has not improved dosing for the blood thinner warfarin, for example.
So far, the clearest success in genomic medicine is for finding genes underlying rare diseases and for treating cancer patients, whose tumors can sometimes be genetically tested to find faults that suggest a specific drug. For that reason, the United Kingdom chose to focus on cancer and rare diseases with a project that is sequencing 100,000 patient’s genomes for medical care.
But Goldstein thinks for certain diseases, such as Lou Gehrig’s disease and epilepsy, genetics will ultimately pay off. And Roden points to a new wave of studies that are finding very rare mutations that more dramatically raise or lower risk. For example, the discovery of rare variants in a gene called PCSK9 that regulates cholesterol levels has led to a new class of potential drugs for lowering cholesterol.
To study such extremely rare mutations requires “a very large denominator,” or huge numbers of subjects, Roden says.
Congress, which so far seems supportive of the precision medicine initiative, will need to approve the funding proposed in next week’s presidential budget proposal for the 2016 fiscal year, which begins in October. It’s not clear how much will be “new” money and how much is reprogrammed or shifted from existing programs. Nor is it known how much of the funding will go to specific institutes, such as the National Cancer Institute.
Still, NIH seems ready to move ahead: NHLBI and NHGRI are holding a meeting 2 weeks from now on how to build a large U.S. cohort study.