October 2008: A 62-year-old man, otherwise healthy, notices his heart is beating rapidly and irregularly. When he goes to see his doctor, a common heart-rhythm abnormality called atrial fibrillation is diagnosed. Many therapeutic options are discussed, and the patient elects to try drugs to keep his heart rhythm normal. Half a dozen drugs are available to choose from, each with a 50 to 75 percent chance of helping to maintain a normal heart rhythm, and the doctor prescribes one. Because atrial fibrillation is associated with a risk of blood-clot formation in the heart, the doctor also prescribes the anticoagulant warfarin at the usual starting dose of 5 milligrams a day.
A week later the man returns to his physician—still with symptoms of atrial fibrillation, and now with extensive bruising caused by excess warfarin.
Fast forward to October 2020: Different patient, same symptoms. While the physician and patient discuss atrial fibrillation and its potential treatment, the computerized medical information system interrogates the patient’s DNA sequence, which was obtained and archived on his personal medical Web site in 2012. The interrogation identifies variants that likely pre-disposed the patient to the abnormal rhythm. Based on the specific genes involved, a medicine with more than a 95 percent chance of suppressing his symptoms is prescribed. The interrogation reports variants in the genes responsible for degradation of warfarin, and a safer starting dose, 2 milligrams a day, is suggested.
It is axiomatic that not every patient responds to drugs in an identical fashion. Similarly, we vary in our susceptibility to most diseases like infections, Alzheimer’s or cancer. The idea that this variability lies in our genes has been widely accepted for decades—but which genes? And can knowing which genes help us better to care for patients?
Vanderbilt is positioned to be a leader in answering these questions and moving the 2020 scenario to reality. In 2004, Vanderbilt University Medical Center, recognizing the opportunities as well as the challenges in implementing a genome-based view of 21st-century medicine, committed to the creation of a large DNA databank—a project now termed “BioVU.”
The university now has, by far, the largest DNA biobank in the country. During the course of the next several years, it will grow into one of the largest worldwide. BioVU’s overriding aim is to serve as a very large clinical laboratory in which questions like those posed above can be addressed.
Building such a capability is a team effort, and VUMC is uniquely poised to bring that team together. Translational science, the idea of bringing advances at the laboratory bench to the bedside and vice versa, has been a traditional strength of ours. The Division of Clinical Pharmacology has had a decades-long interest in the mechanisms underlying variability in response to drug treatment in humans and is a world leader in identifying genetic causes for such variability. Vanderbilt’s Center for Human Genetic Research has become a leader in the application of new genomic technologies to understand disease susceptibility.
Nationally and internationally, Vanderbilt’s capabilities in the discipline of biomedical informatics are second to none. Vanderbilt’s Department of Biomedical Informatics is the largest in the country by far, reflecting a commitment to information technology and medicine made almost two decades ago.
Before we delve deeper into Vanderbilt’s plans, though, a quick refresher on genetics and DNA may prove useful.
A Primer in Modern Genetics
The classic pea-cultivation experiments of the Bohemian monk Gregor Mendel in the late 19th century established the basic ground rules for modern genetics and how diseases can be transmitted in families. Mendel had no idea how this process worked at the molecular level. Indeed, it was not until the 1950s that the mechanism for transmitting genetic information from parent to child and from cell to cell was identified and represented accurately in the now-familiar model of the DNA double helix structure. Cells use the 3-billion-letter-long biochemical code to manufacture proteins—the molecules that determine virtually all cellular functions.
The discovery of DNA as the carrier of the code led quickly to identification of very small changes in its sequence—termed “mutations”—that cause thousands of rare diseases like sickle cell anemia or cystic fibrosis. For the family affected, the consequences of a mutation can be devastating. A huge effort has gone into identifying mutations, understanding how they affect cellular function, and developing methods for early diagnosis and treatment.
But genetics determine common characteristics as well: You look like your grandfather; your uncle had high blood pressure, and so do you; people in your family react badly to some medicines.
As a medical student in the 1970s, I was taught that common diseases like cancer, atherosclerosis or Alzheimer’s disease include a “genetic component,” but that’s about as far as things got. The notion of actually naming those genes was far-fetched. In the 1980s, however, two apparently unrelated events spawned a revolution in modern genetics that is now identifying those genes and many others that contribute to the ways we differ in our disease susceptibility or drug responses.
The first event was development of the “polymerase chain reaction,” a simple method to rapidly generate large quantities of DNA that can then be analyzed in myriad ways in the laboratory. The second was the very rapid acceleration of information technology.
The merging of these two advances
allowed researchers to generate and share DNA sequence—from humans and other animals down to bacteria. So during the 1980s it became increasingly clear that each human being carries millions of DNA variants.
Only a few variants, however, cause diseases like sickle cell anemia. One reasonable idea is that those millions of others, which we term polymorphisms, contribute to variability in the way we respond to our environment: to viruses, drugs or toxins, for example.
This change in emphasis from rare disease-associated mutations to common polymorphisms was accompanied by a change in terminology. Studying one gene at a time is one way of thinking about “genetics,” while studying large collections of genes in a patient or a population is now termed “genomics.”
Amazing new technologies that can generate precise sequences of very long stretches of DNA have enabled genomic experiments. The first full sequence of a human genome (the “Human Genome Project”) was completed in 2000, took three years, and cost about $300 million. A next step, the creation of a catalog of several million common polymorphisms—ones shared by 1 to 50 percent of all humans—has been largely accomplished. Since 2005 this catalog has allowed researchers to identify regions of DNA, and sometimes specific variants, that increase risk for dozens of common diseases by 50 to 100 percent.
But that resource pertains only to the common variants; each of us also harbors millions of rare (less than 1 percent frequency) variants. The technology to find these is also on the horizon. Within five years (some say as soon as two years), we will be able to generate a full DNA sequence from any individual in less than 15 minutes. The cost to do so will be $1,000 or less, and because one’s DNA doesn’t change, it will be a once-in-a-lifetime expense.
Data Isn’t Information, and Information Isn’t Knowledge
The idea that we all harbor both rare and common DNA-sequence variants that make us like our grandparents, predispose us to disease, or cause us to have unusual drug reactions is at the heart of modern medicine. We are now at the threshold of actually knowing those variants and applying them to routine patient care.
Huge obstacles must be overcome, however, before that vision can even approach reality. How can we translate billions of pieces of raw data into useable information? Which of these DNA-sequence variants makes a difference? Does a large group of patients fare better if a disease is treated with drugs prescribed based on individual DNA sequences, or if everyone is given the same dose of the same drug? How much better? Is it cheaper? How much cheaper? How is that best measured? Is there a likelihood of discrimination (for jobs or insurance, for example) based on genetic sequences? Does a genetic sequence guarantee a disease or abnormal drug response, or does it merely predispose? How do multiple genetic variants interact with each other and with environmental stressors to which we all are subject every day? Can lives be saved by avoiding life-threatening drug reactions that are genetically determined?
During the next several years, it should be possible to begin answering these questions—to begin to use the raw data we can generate right now to create the knowledge that will impact patient care.
The BioVU project entered a three-year planning phase in 2004. Focus groups were conducted, sample handling and storage mechanisms were developed, and a plan for sample accrual was put in place.
Our electronic medical-record systems support all patient care at VUMC. They can be “mined” for information, such as outcomes of drug treatment, and can provide a platform for delivering patient-specific information—such as drug interactions (now) or genetic information (soon)—to prescribers.
Patients receiving treatment at VUMC now sign a new “consent to treat” form that features a prominently displayed box that allows a patient to opt out of participation in the BioVU project. Only samples that are left over after being obtained for routine clinical testing, and that also include a signed consent form without a mark in the opt-out box, are accrued into the databank.
Because of this unique design, the project is reviewed and has continuing oversight at many levels, including by Medical Center and external ethics boards, the Institutional Review Board that reviews all human-subjects research at Vanderbilt, the Medical Center’s legal department, the federal office responsible for human-subjects oversight, and a community advisory board made up of patients and lay representatives.
The opt-out approach has advantages and disadvantages.
The only information available about the diseases of a particular individual or drugs used by that person is the data contained in the electronic medical record, so new computer methods are in development to mine this information effectively. If a researcher needs information that is not in the electronic medical record (details of food intake or extensive family history, for example), BioVU may not be the most suitable platform for their work. Privacy and data security are continuing focuses of the project; indeed, propelled largely by BioVU, Vanderbilt is developing into a national center of excellence for work in this area.
One major advantage is that the resource is “real world,” with records containing many different diagnoses and drug therapies. Another advantage of the approach is scale: The project began collecting samples in spring 2007 and, by June 2008, had surpassed 40,000 samples—making it easily the largest DNA biobank in the country.
What Could BioVU Accomplish?
Just getting the DNA databank project off the ground has been an enormous undertaking. It already has generated national and international recognition for Vanderbilt Medical Center because of its vision and commitment to the project. BioVU and the Department of Biomedical Informatics are key partners in a network launched by the National Human Genome Research Institute to explore the utility of DNA biobanks linked to electronic medical-record systems. Most recently, the federal government awarded VUMC $1 million to purchase a robot for sample handling.
BioVU will be a crucial national resource in understanding how to get new genome science to the bedside to improve patient care. The databank will be invaluable to studies of the role played by genetics in the development of complications like kidney disease or amputation in diabetes, drugs used to treat common diseases like HIV or depression, survival of cancer, and many other diverse scenarios.
BioVU also is serving as a platform for the development of new science in the study of data privacy and security, and will be an integral partner in electronic surveillance to identify side effects of both new and old drugs.
We are only beginning to realize the promise of the BioVU DNA Databank Project. The ability to generate very large sets of records with defined diseases and controls offers us opportunities to understand not only how genetic variation affects outcomes in VUMC patients, but how that information can be coupled to advanced information technology to actually deliver improved health care. The July 2020 vision will become reality, and VUMC will have played a huge role in creating that reality.
Dr. Dan Roden is a professor of medicine and pharmacology, director of the Oates Institute for Experimental Therapeutics, and assistant vice chancellor for personalized medicine at the
Vanderbilt University School of Medicine. He is the principal investigator for the BioVU Project.