Databanks of the Future

Eyeing "personalized medicine," leading research centers are creating vast storehouses of images and genetic information.

In the daily practice of medicine, electronic medical records are thought of as largely utilitarian tools that enable the rapid delivery of lab results or easy retrieval of historic records. For a handful of researchers, however, the EMR is a tool with the potential to redefine how medicine is delivered—right down to the individual patient.

Proponents of "personalized medicine" are in the early phases of research that they say will lead to highly tailored drugs and treatment plans, ones based on a patient's genetic makeup. The research efforts will involve vast clinical data repositories, advanced imaging techniques, and highly sophisticated computational models. The projects are complicated by privacy concerns. Yet researchers point to early successes that suggest personalized medicine is not as far-fetched as it may sound. Here are snapshots of two of the industry's noteworthy efforts.

Project: DNA Databank
Host: Vanderbilt University Medical Center

For the past year, researchers at the Nashville-based academic medical center have been gathering DNA from discarded blood samples, compiling a DNA database that spans some 40,000 patients. The databank reflects the marriage of two formerly independent efforts, says Jeffrey Balser, MD, associate vice chancellor for the four-hospital organization. "We've had an emphasis on pharmacogenomics and wanted to understand how variations in DNA predict reactions to drug therapy," he says. "We've also made a major investment in our own healthcare IT and have more than 60 faculty members in biomedical informatics. The DNA databank fuses the two."

Building the DNA databank has required careful attention to privacy issues, Balser says. Vanderbilt provides detailed accounts of the project—and its restrictions—in its efforts to convince patients to donate their blood samples. Individual DNA is tied to Vanderbilt's homegrown EMR and its record of lab results, diagnoses, and outcomes. However, this record has been stripped of any identifying patient information, such as name, date of birth, or even dates of service. "The bioinformatics team has created the ‘mirror image' of our EMR for research purposes," he says. "Dozens of elements have been changed or modified" to safeguard patient identity. The challenge with DNA-linked research, Balser says, is that for many tests, vast numbers of individuals must be considered to draw scientifically valid conclusions. "Proving that DNA variation actually predicts a side effect or therapeutic benefit is extraordinarily difficult," he says. "You need hundreds of thousands, if not millions, of patients."

Vanderbilt's own databank is growing steadily. About 900 new DNA samples are added each week, Balser says. Participants are recruited during office visits. The databank is large enough to enable validation of some limited studies, Balser says, such as side effects of certain drugs. Vanderbilt has already used its EMR-supported data to validate some published research on side effects of drugs given for rheumatoid arthritis. "We want to do more difficult things," he says. For example, the database could support research that would determine which patients would respond most favorably to antidepressant medication.

Balser is quick to point out that the DNA database, by itself, does not prove anything. "It tells us what prospective clinical trial to do with a specific group of patients. It is a hypothesis-generation tool." Even so, he notes that without the EMR and sophisticated analytical tools used to probe its data, determining any DNA links to medical outcomes would take years instead of minutes.