The Possibilities of Personalized Medicine
What if your doctor could tell by reading your genetic code which drugs were most likely to work for you, and which you should avoid, even before you try them?
What if a “genetic biopsy” taken from your cancer could pinpoint the treatment most likely to kill the tumor, and with the fewest side effects?
What if your “genetic medical record” flagged previously undetected risk factors for illnesses you could take steps to prevent?
These are some of the tantalizing prospects in the burgeoning field known as personalized medicine that are being pioneered by scientists and health care providers at Vanderbilt University Medical Center.
Not too many years ago the idea was considered fanciful, that with genetics as their thread, doctors could fashion for each of their patients a medical care coat that fit them perfectly, restored them to health, kept them well and saved money to boot.
Today personalized medicine is rapidly inserting itself into the scientific literature, muscling into the doctor-patient relationship and sweeping through the popular culture. The question no longer is what if these things could happen, but when will they?
For the past several years, for example, Vanderbilt scientists have been testing the value of “pharmacogenomics.” Can variations in an individual’s genetic makeup, his or her genotype, when correlated with clinical data, reliably predict response to commonly prescribed drugs?
If so, which of these variations should be routinely embedded in the electronic medical record even before the patient gets those medicines? Could this information help doctors avoid adverse drug effects, a leading cause of death among hospitalized patients?
Vanderbilt researchers hope to answer these questions in their latest study, called Vanderbilt Electronic Systems for Pharmacogenomic Assessment (VESPA), supported by a two-year, $6.4 million stimulus grant from the federal government.
“I think it’s very likely that within the next year we could actually have a real-world, on-the-ground plan for personalized medicine, where we would recognize that it is best practice and cost-effective for us to do a standard set of limited genotypes on every single person,” predicted Dan Masys, M.D., chair of Biomedical Informatics.
Masys is co-principal investigator of the VESPA project with Dan Roden, M.D., assistant vice chancellor for Personalized Medicine.
Before we start screening everybody for everything, however, it’s worth pointing out a few challenges.
For one thing, this is not going to be easy.
Most common diseases, like diabetes or hypertension, result from the interactions of more than one gene and probably the environment as well. Multiple genetic variables also determine how medicines are metabolized, or broken down by the body, thereby complicating the search for predictors of drug response.
“The likelihood that we can solve most of these problems solely by genetics I think is optimistic at best,” cautioned clinical pharmacologist Alastair Wood, M.D., an emeritus professor at Vanderbilt and partner in a New York-based biopharmaceutical investment firm.
That’s not the half of it.“Personalizing medicine is about far, far more than people’s genes,” said Roden, who also directs the Oates Institute of Experimental Therapeutics. “It’s about who you are and what you eat, what you’re exposed to.”
For example, take two people who carry a gene for asthma susceptibility. Neither smokes, but one is a Phoenix resident with good health coverage while the other lives in a city with dirtier air and is uninsured. “Those are two very different kinds of people,” Roden said.
A third challenge is economic.
Despite the cost-saving promises of personalized medicine, the nation’s health care bill could actually balloon if genetic screening is allowed to flourish without the restraints of evidence-based medicine – doing only those things that are proven to work.
It’s well known that screening newborns for certain inherited disorders can yield significant savings by preventing the costly consequences of disease.
“But in general, the more you test, the more you do and the more it costs,” said Ellen Wright Clayton, M.D., J.D., director of the Vanderbilt Center for Biomedical Ethics and Society. “We end up treating a lot of abnormalities that probably don’t need to be treated.”
Yet the pressure is on.
As the number of tests for different genetic diseases has soared, from about 100 in 1993 to more than 1,700 in 2008 according to one survey, so has the number of gene-testing firms.
More than 170 U.S. companies now market directly to the consumer, Clayton said, even though the medical value for many of the tests they offer has not been established. “We have real reason to be concerned about whether people are even getting useful information,” she said.
The Right Dose
That doesn’t mean personalized medicine should be abandoned – on the contrary.
In the 1990s, Richard D’Aquila, M.D., helped pioneer HIV genotyping. The method identifies genetic variations in AIDS viruses that infect individual patients, and which allow the viruses to “escape” or become resistant to drug therapy.
Now standard practice, genotyping patients’ viruses helps doctors avoid drugs that won’t work, and it improves the success of treatment, said D’Aquila, who currently directs the Vanderbilt-Meharry Center for AIDS Research.
Vanderbilt researchers led by David Haas, M.D., also have helped identify several genetic markers in patients that predict how they will respond to AIDS drugs and whether they may experience life-threatening side effects.
“There is great enthusiasm about the potential to use human genetic information to inform prescribing of HIV medications,” said Haas, who directs the Vanderbilt AIDS Clinical Trials Program. “This may be particularly relevant to resource-limited countries that are most affected by the AIDS pandemic, and where the ability to closely monitor patients for treatment efficacy and toxicity is limited.”
Vanderbilt scientists also are helping to push forward the frontiers of anti-coagulation therapy.
Every year in the United States, more than 2 million people take warfarin, brand name Coumadin, primarily to prevent blood clots after a heart attack, stroke or major surgery. Because patients’ responses to the drug vary widely, however, it’s difficult to find the dose that blocks clots without causing serious internal bleeding.
It’s been known for several years that variations in two genes can affect how warfarin is metabolized and therefore how fast it is eliminated from the body. In 2007, the U.S. Food and Drug Administration notified doctors that testing for these variants could help them estimate the proper dose to give patients.
Last year, a report by an international research consortium including Vanderbilt helped put the FDA policy into practice. The researchers found that they could increase the accuracy of the dose estimate when genetic information was combined with clinical data.
The next step, to be pursued in the VESPA study, will be to show that such an approach actually improves outcomes. VESPA also will take on another best-selling blood-thinner, Plavix, which has been routinely prescribed to reduce blood clot formation following procedures to open blocked arteries.
Many patients aren’t at risk of clotting, and some patients have a reduced response to Plavix because they lack the enzyme necessary to convert it into its active form. But most get the drug anyway because their doctors can’t tell who really needs it.
“The most common adverse drug effect is … it just doesn’t work,” Masys said. Americans spend billions of dollars each year on prescription medication. Cutting out useless and harmful medication “could pay for a big chunk of health reform,” he said.
Work Station Medicine
The engine powering VESPA is Vanderbilt’s massive DNA databank, BioVU.
One of the largest of its kind in the country, the databank contains more than 75,000 DNA samples and their matching medical records that have been de-identified, meaning that all personal information has been stripped away to guarantee patients’ anonymity.
A young assistant professor, Bradley Malin, Ph.D., is the privacy expert in the Department of Biomedical Informatics, while two colleagues, Joshua Denny, M.D., and Hua Xu, Ph.D., have applied “natural language” processing technologies to extract relevant information from the electronic medical record.
Natural language is a way of teaching computers to “read” and evaluate provider notes and other relevant clinical information that have been entered into the medical record. It provides a more accurate picture of the patient’s diagnosis than do the billing codes used for reimbursement, Masys said.
But herein is the rub.
“There’s just no way you can insert this information into health care operations without what we call computerized, patient-specific clinical decision support, that is, work station-based medicine,” Masys said. “There are too many things to remember, too many facts, too many complexities.” Yet only about 5 percent of U.S. hospitals have this kind of decision-support infrastructure.
Roden agreed. “The barriers to implementing what we think we already know are pretty daunting, but they’re straightforward,” he said. “The data have to be in the record and accessible with a mouse click.”
And there are other questions that must be answered for personalized medicine to move into the mainstream.
Should research subjects be contacted if their DNA is found to contain a severe chromosomal abnormality they could pass on to their children?
If a patient is found to have a high-risk disease gene, should family members be notified so they also can be screened, even against the patient’s wishes?
Should people with a family history of a disease be required to undergo genetic testing?
Roden is optimistic that these challenges will be resolved, particularly as the cost of gene sequencing continues to fall and as the benefits of genotyping become more apparent.
Flight of fancy? Maybe. But 30 years ago, so was the Internet or the idea of a wireless phone you could put in your pocket.
Perhaps 30 years hence, Roden mused, “I wake up in the morning and put my finger in a hole in the wall, and a sensor tells me how much drug is circulating in my body … whether I’m going to get the flu (and) that my diabetes is in good control.
“It sounds nutsy, but nutsy in a doable way, with nanosensors and advanced mass spectrometry,” he said. “Maybe that’s the way it’s going to work.”