Researchers Detail Challenges of Personalized Medicine in Colloquium at AAAS

Despite some successes in the increasingly intense search for very specific molecular signposts of disease, the era of medicine custom-tailored to a patient's own genetic profile is not yet at hand and faces significant scientific and policy hurdles.

That was the consensus at a recent two-day colloquium at AAAS on the future of personalized medicine. Researchers and clinicians still are grappling with how best to translate advances in genetics, molecular biology and imaging techniques into clear benefits for patients.

While doctors have always considered individual factors such as lifestyle and family history when recommending treatment, the decoding of the human genome in 2000 set the stage for sweeping changes in the practice of medicine. As scientists came to better understand the roughly 20,000 to 25,000 human genes and the proteins they code for, so the reasoning went, highly targeted diagnostic tests and drugs could be developed, replacing the usual trial-and-error approach. Recent advances in imaging are opening another door to a more individualized picture of how patients respond to disease and medical treatment.

But even some of the successful diagnostic tests have had mixed results, experts said at AAAS, and there is a complicated regulatory picture that means varying standards in the validation of such tests. There are other issues: educating physicians and patients on the potential benefits and limitations of the tests; convincing insurers to cover the costs of the new procedures; ensuring equitable access to tests that prove useful; protecting the privacy of those whose genetic tests may suggest an elevated risk of future disease.

The 1-2 June colloquium, co-organized by the AAAS Scientific Freedom, Responsibility and Law Program and the Food and Drug Law Institute, paid particular attention to the quest for biomarkers—molecular signposts for disease or a patient's response to a drug. The use of biomarkers is not new. For decades, clinicians have looked for the presence of certain heart muscle proteins circulating in the blood as reliable indicators that a person has suffered a heart attack, for example.

But the emphasis now is on using rapid screening methods and powerful new technologies to find similarly useful biomarkers for a wide range of conditions. The aim is to find diagnostic markers for disease that can then be used in the development of new drugs as well as the selection and appropriate dosing of existing drugs.


Kathy Hudson

The challenges have a lot to do with the science in developing the evidence across the spectrum from analytic validity to clinical utility, said Kathy Hudson, director of the Johns Hopkins University Genetics and Public Policy Center. Personalized medicine will only be as good as the tests used to inform drug selection and dose, Hudson said.

The Early Promise of Biomarkers

There have been some promising early developments. A woman with breast cancer now has the option for a test that can tell whether her tumor is overproducing a gene product called HER-2. If she tests positive for the biomarker, she is a good candidate for a companion drug called Herceptin that targets the excess HER-2 and cuts the risk of disease recurrence nearly in half.

Another biomarker test can tell whether a patient has a particular genetic combination of liver enzymes, known as cytochrome P450, that help metabolize more than half of all drugs. The test can be used to fine-tune drug dosage based on a person's metabolism rather than on cruder factors such as body weight. But Dr. Howard Levy, a geneticist and physician at Johns Hopkins University, said there remain questions on whether the cytochrome P450 system is better than trial-and-error when it comes to its clinical utility for patients.

Researchers also have found that two genes can affect how patients respond to warfarin, a widely used blood-thinning drug. While it likely that the biomarkers for the two genes have helped some doctors do a better job of determining warfarin dose, Levy said, there remain insufficient data on whether the new test has led to improved drug efficacy and fewer adverse reactions. And even with the test, he said, about 45% of the variation in response to the drug remains unexplained and is likely due to other factors such as diet, exercise and compliance in taking the medicine.

Another test can find the presence of a mutant gene in patients with chronic myelogenous leukemia. A companion drug called Gleevec binds to the faulty genes product and inhibits its cancer-causing action.

While there is no question that some of the new biomarker tests are worthwhile, Levy cautioned that we don't truly know the prospective performance of most biomarkers. Forward-looking studies that track the performance of the test in a well-controlled population simply have not been done in many cases, he said, adding that just because a test is offered on the market, its validity is often assumed. But a positive test does not always correlate with disease, Levy said. Physicians and patients alike need to understand the strengths and limitations of the tests now becoming available, according to Levy and others.

There is no lack of biomarker candidates, said Leigh Anderson, chief executive officer of the Washington-based Plasma Proteome Institute. A literature search in 2006 found 1261 candidate proteins related to cancer, he said. But which ones are useful as diagnostic indicators of cancer or the effectiveness of treatment?

Seeking Answers in a Vast Ocean of Data

Even as thousands of proteins are being identified and linked to specific genes, the verification and qualification of potential biomarkers continues to lag, Anderson said. The limiting step, he said, is looking at a large number of patient samples to determine whether a biomarker really is related to the disease in question. Despite the interest in biomarkers for personalized medicine, Anderson said, we are doing less well at finding them and getting them into clinical practice.

He estimates that more than half of the published protein biomarker studies have been compromised by factors such as sex and age mismatches among the groups providing samples and differences in handling and storage of samples. Instances of sampling bias erode confidence in the relevance of proteins as biomarkers, he said.


Dr. Rick Hockett

Our big problem is how to pinpoint the right biomarker to satisfy the condition that were interested in, agrees Dr. Rick Hockett, chief medical officer for Affymetrix, a biotechnology company based in Santa Clara, California. The challenge is to sift through what Hockett called this vast ocean of stuff—millions of small segments of DNA called single nucleotide polymorphisms (or SNPs) and huge numbers of proteins—for the handful of relevant markers for an actual medical condition of interest.


Dr. Klaus Lindpaintner

More than 200 molecular biomarkers are in various stages of development, according to one estimate. The real challenge now is to find the biomarkers that are relevant, said Dr. Klaus Lindpaintner, director of Roche Molecular Medicine Laboratories. That is going to be about as difficult, or perhaps even more difficult, than finding a new drug, he said.

Complex Medicine, Complex Regulation

Courtney C. Harper, head of the U.S. Food and Drug Administrations Office of In Vitro Diagnostic Device Evaluation and Safety, offered a perspective on the complex regulatory regime for molecular diagnostic tests. She noted that a long-standing FDA policy has resulted in a non-level playing field for the review of the tests.

Those that are distributed as kits from commercial manufacturers must undergo FDA review prior to marketing. But a biomarker test that is developed by a single laboratory for use only in that lab—even if intended for use in personalized medicine—does not need to undergo FDA pre-market review. Nor is the developer required to show clinical validity for the test. Rather, these tests are regulated under the Clinical Laboratory Improvement Amendments, administered by the Centers for Medicare and Medicaid Services.

Genentech, a major biotech firm, has petitioned the FDA to include these so-called laboratory developed tests (or LDTs) under its pre-market review umbrella, particularly if they are intended for use in personalized medicine. Harper said LDTs are considered medical devices and so FDA does have enforcement discretion over most of them if there are risks to patient safety. And if an LDT is meant to be used in conjunction with a drug as a companion diagnostic, FDA approval is required.

There are a lot of problems with this two-path system, and I'm hopeful the new administration will move rather quickly to level this playing field, said Hudson of Johns Hopkins.

Hudson also said FDA needs to be more consistent on the amount of time and level of evidence it requires before changing a drug label to reflect the fact that a companion biomarker test is available. She cited the cases of abacavir, an anti-AIDS drug, and carbamazepine, an anti-epilepsy drug. In both cases, researchers found gene-related markers that were predictive of adverse reactions to the drugs. In the case of abacavir, Hudson said, the FDA required a randomized, double-blind, prospective clinical trial and took about 78 months before requiring a black box warning on the drug label about the possibility of gene-linked hypersensitivity reactions. For carbamazepine, she said, the agency accepted retrospective case-control studies (considered a weaker standard of proof) and issued the black box warning about 50 months after the first evidence of gene-linked adverse events, particularly in Asian populations.

Regulators also are scrambling to keep up with direct marketing of personalized medicine procedures to consumers. Tests for determining your genetic profile are being offered for sale over the Internet. Last year, California asked 14 companies to stop online offers of genetic tests to its residents. Beatrice O Keefe, acting chief of laboratory field services for the California Department of Health, said the labs were subject to California jurisdiction and were not licensed. Five of the labs have since been licensed by the state, she said, and others have put disclaimers on the Web saying their tests are not available to California residents.

Public Outreach: Under-Promise and Over-Deliver

There is little doubt that many consumers have an interest in the effort to develop custom-tailored medical tests and procedures. People want to know about their genetic information, said Linda Avey, founder of 23andMe, a firm that provides genetic testing for more than 100 traits and diseases as well as DNA ancestry. Avey said her firm is creating an online platform that will allow patient advocacy groups to keep in contact and share information about advances in genetic medicine. One goal is to build a community of patients who are receiving warfarin to help determine the clinical value of the companion biomarker test and explore why it is not being used more widely.

Paula Kim, a patient advocate and chief executive officer of a healthcare consulting firm called Translating Research Across Communities, said she has great confidence that personalized medicine will become the norm rather than the exception. But she urged scientists to avoid hype and premature enthusiasm. You want to under-promise and over-deliver, Kim said.

Conference participants spoke of the need to integrate traditional patient care with the new technologies and approaches being pursued under the rubric of personalized medicine. There also was considerable emphasis on getting patients more involved in the process. Dr. William Dalton, director of the H. Lee Moffitt Cancer Center and Research Institute in Tampa, Florida, said the biggest challenge is how to involve the community in the discovery and delivery of personalized health care. His center has started an ambitious program, involving a consortium of institutions in Florida and elsewhere, aimed at building an improved infrastructure for the delivery of personalized cancer care.

The program will create a Total Cancer Care network of health care providers who contribute and share information from individual cancer patients in hopes of improving care for patients from Florida and beyond. The network will store tumor tissues for study using molecular techniques, gather data on the results of clinical trials, offer patients specialized diagnostic services, and suggest the best treatment options based on the clinical and biological character of a patients cancer. There are more than 34,000 cancer patients now enrolled, Dalton said, and he expects there will be about 100,000 in the network within 3 to 5 years. People in the community saw an opportunity to participate in the development of personalized cancer care, Dalton said.

Skepticism in Insurance Industry


Pamela Sankar


Dr. Louis B. Jacques

As personalized medicine enters the mainstream, insurers will face decisions on whether to cover novel tests and procedures. Pamela Sankar, a medical ethicist at the University of Pennsylvania, noted that CIGNA decided late last year not to cover the Cytochrome P450 test for drug metabolism because the insurer considered the test to be experimental, investigational or unproven.

Dr. Louis B. Jacques, director of the division of items and devices at the Centers for Medicare and Medicaid Services, said that an advisory committee on Medicare coverage guidelines recommended recently that expectations for evidence gathered by genetic tests should be at least as high as for other diagnostic technologies. He said the agency will be looking for clear evidence of clinical benefit. In May, the agency proposed not to pay for the genetic tests to assess the best dose of warfarin, the blood thinner, except when the tests are offered in a clinical trial to determine whether they lead to improved patient health. After a public comment period, a final decision on the tests is expected in July.

Jacques also said his agency will not offer reimbursement for what he called recreational testing, such as tests aimed at providing a general genetic profile. And he noted that if a test is not ordered by a physician, we won't pay for it.

With an increasing number of tests related to personalized medicine on the horizon, experts said it is crucial that medical professionals and consumers alike become more knowledgeable about the field. Joseph McInerney, executive director of the National Coalition for Health Professional Education in Genetics, said many health care professionals, including physicians, are generally unprepared to deal with genetics-based medicine. He noted that an advisory panel to the U.S. Secretary of Health and Human Services concluded in April 2008 that practitioners cannot keep up with the pace of genetic tests (and) are not adequately prepared to use test information to treat patients appropriately. The panel also said that practice guidelines are insufficient to ensure appropriate care.

Part of the problem, McInerney said, is that medical students do not receive sufficient training in medical genetics. A 2007 survey of U.S. and Canadian course directors in medical genetics or curricular deans found that 77% of the 112 medical schools responding to the survey teach medical genetics in the first year during basic science classes. About two-thirds of the respondents devoted 20 to 40 hours to medical genetics instruction, largely focused on general concepts rather than practical applications, which accounted for only 11% of course time. Often, the students receive no further instruction or do not receive a solid grounding in the latest clinical developments in medical genetics, McInerney said.

There is a disconnect between the basic sciences and clinical experiences during training, he said. The result is that some physicians later may have difficulty interpreting genetic tests and explaining genetic risks to patients.

We're very, very early in the infancy of educating physicians about how to use this kind of data, said Hockett of Affymetrix. If multiple biomarkers of interest were pinpointed for particular conditions, he asked, how many physicians could handle 25 genetic tests dropped in their lap?

Sankar, the medical ethicist, said it should be no surprise that the study of how genetics can affect our response to drugs is turning out to be more complex and challenging than anticipated. Not only does one size not fit all, she said, the factors that affect size go beyond genetics to include environmental influences, disease states that can affect drug metabolism, interactions with other drugs, and patient compliance with dose schedules.

Sankar also cautioned against over-optimistic projections on personalized medicine. Some of the resources now being devoted to personalized medicine research might more properly be devoted to pressing national health issues, including infant mortality, AIDS and obesity, she argued.

Does this mean that personalized medicine is wrong or should be abandoned? No, I am not suggesting that at all, Sankar said. All I am suggesting is that it is not ready for moving into the clinic. It is not ready for prime time. She said personalized medicine is appealing as a silver bullet or quick fix, but allocation of medical resources remains an important ethical and economic issue that should not be ignored.

The Benefits of a Slow-and-Steady Approach


Dietrich Stephan

In the keynote address, Dietrich Stephan, the president of the Virginia-based Ignite Institute for Individualized Health, said it is important to recognize that the United States is on the verge of a health care crisis. He said there are projections that 40% of GDP could go to health costs within about two decades, up from 16.7% now, unless costs are brought under control.

Better prevention and management of chronic disease can have a serious impact on costs, Stephan said, and efforts to identify all heritable risks associated with diseases such as stroke, diabetes, Alzheimers disease, and cancer are worth the effort. While there has been what he called an exponential increase in molecular types of information related to disease, Stephan also acknowledged that there has been very little oversight as to whats real and whats not real in the realm of personalized medicine.

One way Stephan and his collaborators are tackling that problem is with a comprehensive Alzheimers program that has been identifying genes associated with both high and low risk of the disease. The researchers want to run a series of small, nationally based clinical trials during the next decade to develop preventive drugs for Alzheimers, which currently affects as many as 5.3 million Americans with direct and indirect costs of at least $148 billion a year, according to the Alzheimers Association.

The researchers hope to use FDA-approved biomarkers and brain imaging to assess the effectiveness of prevention therapies. The payoff could be both medical—improved quality of life for patients—and financial. Delaying onset of the disease by even five years across the population, when balanced against average life expectancies, could cut prevalence of the disease in half, Stephan said.

The colloquium on personalized medicine was the first of three on the topic by AAAS in partnership with the Food and Drug Law Institute. The second will be held at AAAS on October 26-27, 2009. The third in the series will be held in early 2010.