If you listen to a National Public Radio affiliate, you may have heard the following underwriting announcements earlier this year:
“Vanderbilt University Medical Center. Using a patient’s genome to develop customized blood thinner therapy. At Vanderbilt Health.com.”
“Vanderbilt University Medical Center. Choosing a statin drug based on a patient’s DNA. At Vanderbilt Health.com.”
Personalized medicine is the theme that connects these announcements. One of the innovative Vanderbilt University Medical Center (VUMC) personalized medicine initiatives is PREDICT, which genotypes patients for genetic variants that predict an individual’s response to a variety of prescription drugs. What’s special about PREDICT is that this genotyping is done preemptively — that is, before a drug is prescribed. The pharmacogenomic information becomes part of a VUMC patient’s electronic medical record (EMR), which guides that patient’s physician in adjusting a drug dose or prescribing a different drug entirely.
Eventually, everyone’s EMR will include this kind of personalized pharmacogenomic information so therapies can be prescribed confidently and administered without delay. VUMC just happens to be the first academic medical center to launch such an initiative — it wants to get the word out early with announcements on NPR.
“We want people to know that this is one of the few places that’s doing preemptive genotyping,” says Dan M. Roden, MD, director of the Oates Institute for Experimental Therapeutics and assistant vice-chancellor for personalized medicine at Vanderbilt University School of Medicine, in Nashville. “It is about recognition for the kinds of science that we’re trying to do.”
Far more rational
PREDICT — an acronym for Pharmacogenomic Resource for Enhanced Decisions in Care and Treatment — involves more than just sending a tube of blood to the laboratory for analysis. VUMC aims to genotype 10,000 patients over two years.
The initiative will include consumer and physician education, informed consent documentation, and molecular diagnostic assaying. VUMC also has to make sure that the pharmacogenomic information is entered into the patient’s EMR and that the point-of-care decision support prompt pops up when a physician enters a drug that might conflict with the patient's genotype. Provider behavior in response to genotype information and patient outcomes are monitored.
“It’s an investment in the way healthcare will be delivered in the future,” says Roden.
For a VUMC patient who has been genotyped, that future could be years away or tomorrow. Launched in 2010, PREDICT was initially implemented for cardiac catheterization patients. Forty percent of these patients end up on clopidogrel (Plavix), one of many prescription drugs whose safety and efficacy depend on phenotype — in this case, whether an individual is a poor, intermediate, extensive, or ultrarapid metabolizer. For an antiplatelet drug like clopidogrel, the wrong dose can result in internal bleeding or in blood clots that may cause a heart attack or stroke. Fortunately, an individual’s CYP2C19*2 genotype is highly predictive of phenotype.
“CYP2C19*2 just happened to be the first actionable single nucleotide polymorphism we chose, but PREDICT is not about clopidogrel,” says Roden. “It’s about trying to create an infrastructure that will allow us to move genetic information into the chart as it becomes actionable.”
Clopidogrel is one of 57 drugs where the label includes genetic information and the FDA either recommends or requires genetic testing. As the number of such drugs increases, it becomes inefficient and costly to test for each pharmacogenomic variant individually before prescribing the associated drug.
“In an era of rapidly falling genotyping costs and rapidly advancing EMR sophistication and cost containment, we think it’s far more rational to genotype people for many variants preemptively and to include that information in the patient’s chart before they need the drug,” says Roden.
Consumer interest
Patients whose blood is drawn as part of a routine office visit at VUMC are engaged by their providers in a discussion about PREDICT. The initiative includes consumer and physician education, informed consent documentation, and molecular diagnostic assays. For those patients participating in PREDICT, VUMC proceeds with genotyping in its CLIA-certified molecular diagnostics laboratory using the Illumina Vera-Code ADME Core Panel, which interrogates 184 variants in 34 genes.
Patients who participate in PREDICT can access their genotyping results online. Roden declined to disclose the percentage of patients who opt out but points out that focus groups conducted by VUMC reveal a high degree of consumer interest in using genetic testing to improve treatment outcomes.
“People are a little uncertain about the role of genetics in their healthcare overall, but when it comes to the question, ‘What would you think of using genetic information to guide drug therapy?’ That’s something people are very enthusiastic about,” says Roden. Patients, he says, have “less concern that pharmacogenomic information might come back to bite them” than they do about the potential risks inherent in other kinds of genetic information.
A VUMC pharmacy and therapeutics subcommittee drives the scope of the genotyping effort through a careful and deliberate process. Working on a drug-by-drug basis, the subcommittee evaluates FDA labeling, published evidence, and professional society consensus before deciding whether to include a product in PREDICT. So far, clopidogrel, simvastatin, and warfarin have undergone that process.
“It’s an investment in the way healthcare will be delivered in the future,” says Dan M. Roden, MD, at Vanderbilt Medical Center, in talking about PREDICT.
“The big picture of PREDICT is what does it take to get genetic information embedded into the electronic medical record?” Roden points out. There are other questions too, Roden says. “What kind of review process, what kind of assays, what kind of electronic decision support do you have to have? How good are the genetic data in predicting what you wanted to predict? How good are they across various patient populations? How good are the assays? What do you tell practitioners once they have a patient identified? Every one of these genetic variant stories has some promise in healthcare, but every single one of them, if you were to implement them, has a set of questions around them.
“With PREDICT, we’re starting with what we think are so-called low-hanging fruit, and that is drug-gene interaction pairs, but the model should apply to anything that has an important enough influence on healthcare to be embedded in the medical record.”
Economic viability
VUMC is not alone in implementing preemptive genotyping. Roden mentions other initiatives at St. Jude Children’s Research Hospital, Mayo Clinic, and the Scripps Research Institute. VUMC is also one of seven U.S. medical research institutions in the Electronic Medical Records and Genomics (eMERGE) Network, funded by the National Institutes of Health, which combines DNA repositories with EMR systems for large-scale, high-throughput genetic research.
By far the most comprehensive initiative of its kind, PREDICT is being done solely on VUMC’s dime at no cost to patients’ insurers.
After the 10,000th patient is genotyped, VUMC plans to analyze this “very large experiment” (as Roden describes it) to ascertain how much it costs and what VUMC and the patients got out of it. VUMC will then make a pitch to payers.
“This experiment involves generating enough data to convince ourselves, our payers, our patients, and everybody else involved in this community that preemptive genotyping is the way personalized medicine should be done in the future,” says Roden.
Theoretically, there’s a lot at stake if Vanderbilt is right. Already, PREDICT has identified poor metabolizers of clopidogrel, potentially avoiding complications resulting from inappropriate prescribing. And that may be just the beginning. In a 2011 Vanderbilt Magazine article titled “The Promise of Personalized Medicine,” Bill Snyder and Dagny Stuart cite a study that estimates genotyping patients for their ability to metabolize warfarin could avoid 85,000 serious bleeding events, 17,000 strokes, and $1 billion in annual costs of care.
Snyder and Stuart speculate that it may not make financial sense for payers today to invest in genotyping because of member turnover. But with the proliferation of payment reforms, that sort of thinking may be short-lived. In that same article, Jeff Balser, MD, vice chancellor for health affairs and dean of the School of Medicine at Vanderbilt, says the inexorable shift to global reimbursement will vindicate personalized medicine and PREDICT.
“What we’re doing, even though it may not look economically attractive today, will be extraordinarily economically attractive down the road as healthcare payment systems become a lot more sensible.”