One Size Fits All?

Toni Miles, M.D., Ph.D., was still in medical school when she learned a very personal lesson about the dangers of adverse drug reactions.

Miles had been exposed to tuberculosis while caring for patients at a local hospital, and a routine test revealed that her immune system was producing antibodies in response -- a common occupational hazard. Although she hadn't yet developed tuberculosis, the standard protocol in such cases is to administer a drug called isoniazid for nine months to keep the disease in check.

But Miles didn't know -- until it was almost too late -- that her body possessed a genetic variation which makes it metabolize isoniazid more slowly than the average patient.

As a result, the drug built up to toxic levels, causing a form of chemically induced hepatitis that nearly destroyed her liver.

"About a third of the population will have the result that I did," says Miles, UofL's Ole A., Mabel Wise and Wilma Wise Nelson Chair in Clinical Geriatrics Research. "But you won't know that until you actually take the drug. A patient's only recourse at that point is to stop taking isoniazid and hope the liver heals. In my case, it did. But in debilitated patients, the adverse drug reaction can be fatal."

Miles' case illustrates an all-too-common problem in health care: Adverse drug reactions kill 100,000 Americans annually, in part because standard treatment protocols dictate dosing guidelines that are developed with "average" patients in mind.

The problem with standard protocols, however, is that there really is no such thing as a standard patient, Miles says. There is enough genetic variation across the population -- or even within distinct racial and ethnic groups -- to make a "one-size-fits-all" approach to dosing little more than educated guesswork.

Some patients metabolize drugs so rapidly, for example, that the pharmacological agents don't remain in the body long enough to achieve the desired effect unless dosage levels are increased. Other patients, like Miles, may metabolize the same drug so slowly it accumulates to toxic levels and causes symptoms of overdose unless dosage levels are decreased. Then there are people who might not respond to a certain class of drugs at all, or they may suffer adverse -- and potentially fatal -- reactions no matter what dose is prescribed.

"Right now, doctors have to go through a sequence of trial and error to match a patient with the most effective medication and dosage," Miles explains.

"That is, we give the person one drug and dose from the class of drugs needed, and if that doesn't work well or makes the patient sick, we try something else."

David Hein, Ph.D., likens the situation to an experiment.

"With the state of things today, a physician basically says, 'We'll try the drug and dose recommended for most patients, but we'll need follow-up to make sure this drug and dose works for you and isn't toxic.' "

The process wastes time and resources, he says, and it puts patients' health at risk.

But Hein, chairman of the Department of Pharmacology and Toxicology, thinks there is a better way -- examine the genetic makeup of patients in advance to determine which drugs and doses will be most effective based on their unique metabolisms.

The process is known as pharmacogenetic testing.

"If we knew in advance which drugs would work, which drugs would not be toxic and what dose to give, that would be a major advancement in medicine," says Hein, who has been teaching and conducting research in this area for more than 25 years.

"Rather than treating the average patient, we could treat specific patients with specific drugs and dosages using a targeted approach. Patients would get better sooner and we could avoid all those complications from adverse drug reactions, saving millions of dollars in hospitalization costs and lost productivity."

Personalized medicine

That's one of the ideas behind a new trend in health care called personalized medicine, an area in which UofL researchers and clinicians are taking a leading role.

One of those clinicians, Kerri Remmel, M.D., Ph.D., is preparing to launch a study that will examine the usefulness of genetic testing in patients who receive coumadin, an anticoagulant that is crucial for preventing some kinds of strokes.

More than 20 million people take coumadin worldwide, but the drug is notoriously difficult to prescribe because it has a narrow "therapeutic index" -- too little can fail to thin the blood adequately, while too much causes potentially fatal internal bleeding.

A patient's genetic variability complicates matters, since some people metabolize the drug faster or slower than "standard" patients, making dosage a high-stakes trial-and-error process.

"This is a huge dilemma for primary care doctors," says Remmel, interim chair of the neurology department at UofL and director of the Stroke Center at University Hospital.

The problem is so acute, many doctors simply avoid prescribing coumadin for fear of causing serious bleeding in their patients. One study found that 65 percent of patients who should be taking coumadin received none at all; of those who did receive the drug, just 15 percent got an appropriate dose.

Working with PGXL Diagnostic Laboratories, a private Louisville company that specializes in pharmacogenetics, Remmel will profile the genetic makeup of patients who need coumadin. The patients will then be treated using standard protocols to achieve an appropriate therapeutic index. Once that's attained, researchers will compare the final dose with the recommended dose predicted by genetic analysis.

"We want to see how close the numbers match in a controlled, clinical setting," Remmel says. "Based on anecdotal evidence, we believe that the new method will be very useful and accurate.

"If further testing proves that to be the case, the new approach will mean increased confidence for the physician, because the patient will receive the right medicine at an appropriate dose; and it will mean well-being for the patient because we're preventing some really frightening complications.

"I think we're looking at a new age of medicine."

Mark Linder, Ph.D., an associate professor of pathology and laboratory medicine at UofL, hopes to take the work one step further with a computer model he is developing to ease the task of drug dosing.

In the traditional approach to coumadin management, most patients are started with a standard dose of 5 milligrams per day for three days. Blood tests then are taken every few days to determine the level of anticoagulation, and doctors adjust the dose up or down based on how each patient responds to previous doses. But the process can sometimes take several weeks to stabilize, Linder says, because physicians inadvertently "oversteer" -- they increase or decrease the dose too much or too soon while trying to achieve the correct therapeutic level -- and because genetic variation means each patient will respond differently.

Linder's computer model takes into account several factors that affect a patient's ability to respond to coumadin, including multiple combinations of genetic variations. It then computes the optimum final "maintenance dose" for each genetic variation and tells doctors how to get there as quickly and safely as possible by displaying daily dose adjustments on a graph.

"The physician can visually see exactly what a patient's concentration is going to be for any given dosage he or she selects on any given day," says Linder, who also is vice president of operations for PGXL Diagnostic Laboratories. "So instead of experimenting with a patient over five or 10 weeks, the doctor can do it at a computer in a minute and clearly see when the coumadin reaches steady state. That's a great advantage to managing the potential for oversteer in dose adjustments."

Shaped by her medical school experience with isoniazid, Miles also is conducting research to gage the effectiveness of pharmacogenetic testing in patient care.

For the past two years, she's been gathering information on the genetic variations of patients who have been randomly selected from two primary-care practices in Louisville. One of the questions she hopes to answer is whether every patient in a given practice needs to be screened for genetic variations in order for physicians to make effective adjustments in drug dosing and selection.

"These genetic tests aren't cheap," Miles notes, ranging from $100 to $600 per gene. "And different genes are responsible for metabolizing different drugs. So do we want to test every gene for everybody? Probably not. There just aren't enough resources for that.

"But one of the more interesting findings we've had is that each practice is like its own little universe. There is surprisingly little genetic variation among patients in a specific practice. So maybe the data we get from testing a randomly selected sample in each practice is as effective as the data we would get from testing everybody."

Another facet of personalized medicine involves the identification of people who are genetically susceptible to developing certain diseases or who have already contracted a disease that's difficult to diagnose until much later.

In the former case, patients can modify their lifestyles or begin taking prophylactic drugs to reduce risks, Hein says. And in the latter case, therapies can be administered much more effectively once an appropriate diagnosis has been made.

Schizophrenia is one example of disease that could benefit from these approaches, says UofL researcher Mark Brennan, Ph.D.

Brennan has developed a test to enable early detection of schizophrenia by looking for two genetic variations seen in some patients with the disease. Siblings with the first variation are about three times more likely to develop the disease, and those with the second variation face an even higher risk, he says.

The test, which will be marketed by Louisville bio-tech startup SureGene, also could be used to tailor treatment.

"One of the main uses of the test probably will be figuring out which people will benefit most from certain therapies," Brennan says. "Everybody doesn't respond to treatment in the same way, and this will help find the right drug for the right person."

Breaking new ground

Hein believes personalized medicine will revolutionize health care, going so far as to predict that health insurance companies will one day require pharmacogenetic testing before doctors can prescribe certain drugs to their patients.

"I'm convinced that at some point in the future, no health insurance payer will consider it acceptable for physicians to disregard pharmacogenetics in the administration of particular drugs. They'll want to know which drug will work and why in order to minimize costs and improve care."

In the meantime, personalized medicine is helping physicians dust off effective medicines that were shelved years ago because they were too difficult to dose or were too toxic in certain segments of the population, even though they worked well in others.

For example, Hein currently is working with ChemGenex Pharmaceuticals to evaluate amonafide for use in the treatment of solid tumors.

"It is an effective drug, but it was not considered a first-line drug for many years because it was too toxic in too many people," Hein said. "But if you understand ahead of time how to prescribe doses that minimize toxicity, it can be very beneficial."

While the medical applications of personalized medicine are potentially groundbreaking, the movement also is breaking new ground in the legal and ethical arenas.

Larry Palmer, UofL's Urban Health Policy Chair, cites the rising potential for malpractice lawsuits based on pharmacogentic testing.

"One of the primary concerns physicians have to worry about is whether their knowledge base is going to be up to date," he says. "If new knowledge is not translated fast enough for them to know that they need to perform a genetic test before prescribing an ordinary drug, some physician is going to get caught in the sea change.

"There will probably be a case one day where a physician is going to be sued for failing to do a genetic test before giving an ordinary drug."

Mark Rothstein, who heads UofL's Institute for Bioethics, Health Policy and Law, is equally concerned about privacy issues, especially given the fact that the federal government is backing a proposal to create electronic Personal Health Records, or PHRs, for all Americans within seven years.

"A major problem in privacy is that there are many entities that have economic leverage over you and can in effect compel you to sign an authorization 'voluntarily' releasing all of your medical records," he says.

For example, your mortgage company could require release of health records as a condition of granting you a loan.

Rothstein and one of his colleagues have prepared a study, slated to appear in the American Journal of Bioethics this spring, which found that Americans were compelled to sign such authorizations about 25 million times each year.

"You want to file a workers' compensation claim? Are you applying for long-term-care insurance or life insurance? Do you have to take an employment entrance medical examination? All of these are instances in which authorizations may be required.

"To make matters worse, we don't have an effective way of limiting what information is disclosed. So when you sign this authorization, your complete file is sent. Even information that's totally irrelevant, that may be 20 years old and may be highly sensitive.

"This is an area of privacy that's been overlooked, and it will only get worse when we move to all-electronic records," says Rothstein, who chairs a panel that advises the secretary of Health and Human Services on health information policy.

The panel has recommended that any PHR proposal include the ability to limit data disclosures -- what Rothstein calls contextual access criteria -- but current plans have yet to take this into account.

Rothstein, who edited a book exploring the social, ethical and clinical implications of pharmacogenomics -- a field closely related to pharmacogenetics -- also is worried about the future economics of drug development.

"For so long, the pharmaceutical companies tried to discover the next blockbuster drug," he says.

"They focused on common chronic conditions for which they could treat huge segments of the population on a continual basis. Now the industry is going to have to develop a whole new business model based on more targeted therapies and sales to smaller segments of the population.

"It's going to take a lot more research to do that, and nobody knows how it's going to shake out."

Despite these concerns, Rothstein remains cautiously optimistic.

"I think this is very promising field of study, and I'm in awe of the scientists who are making it happen," Rothstein says. "I just think we need to tread carefully as medicine moves forward."