February 11, 2008

Volume 86, Number 06

pp. 17-27

Personalized Medicine

Armed with the decoded human genome, drug researchers and clinicians are working to close the gap between patient and therapy

Rick Mullin

"IF THIS IS GOING to move forward, it is going to take a whole new approach to science," says Gordon B. Mills, chairman of the department of systems biology at M. D. Anderson Cancer Center in Houston, part of the University of Texas.

While sitting at his desk, on which four computer screens are partially covered by Post-it notes, Mills points to a whiteboard. Nestled between charts and memos hung on the wall, the board holds a long list of the center's research partners, both academic and industrial. "It is going to require team science," he says. "And the rate-limiting factor," he adds amid the swirl of digital and hardcopy scrawl, "is bioinformatics."

Mills is talking about personalized medicine, the practice of catering therapies to the specific genetic profiles of patients and their diseased cells. Fueled by the genetic data and diagnostic tools that emerged in the wake of the decoding of the human genome 10 years ago, personalized medicine is viewed by many in the health care community as a major shift away from the standard approach of developing drugs that work for the largest number of patients in a given therapeutic category.

Work is moving slowly, Mills admits. Although cracking the genetic code enabled drug researchers to make credible claims about the potential to treat diseases such as cancer and central nervous system disorders, significant breakthroughs have been few in the decade-long slog through new genetic data. Pharmaceutical companies, biotechnology firms, and institutional researchers are rallying around personalized medicine, but the timeline to new therapies is still being recalibrated.

There is no question that new information on the genetics of human disease is being put to breakthrough use in the development of therapies that target patients most likely to benefit. This, in turn, is leading to a big change in the design of clinical trials. And pioneer targeted therapies—such as Genentech's Herceptin for breast cancer and Novartis Pharmaceuticals' Gleevec for lung cancer—constitute a variety of new-age blockbuster drugs from which researchers are still learning.

Personalized medicine, however, represents only a thin slice of what is taking place across the health care sector, where blockbuster drugs are still defined as $1 billion-a-year sellers that target the largest number of patients with a disease. And the practice poses a challenge to traditional pharmaceutical companies by suggesting that new drugs will target smaller populations.

Change comes slowly in a regulated industry dependent on a complicated health insurance scheme. The Food & Drug Administration and insurance providers are seeking new methods for assessing targeted diagnostic tools and the effectiveness of new therapies. Meanwhile, chemists and biologists are turning to translational research—a closer linking of laboratory drug discovery and clinical development—and are hunting for biomarkers, molecular indicators of drug efficiency.

In the health care community, personalized medicine is seen variously as a fundamentally new medical practice or as a continuation of a nearly century-old tradition, but with better tools and more data. In Mills's view, cancer therapies have been personalized for a while now, through the long-standing practice of classifying tumors via their anatomical and visual characteristics.

The practice, however, has gained a fundamentally new dimension with the decoding of the genome. "What's changed is that we now have the ability to go beyond anatomic and structural differences," he says. "We go for underlying mechanisms."

In breast cancer and leukemia, he notes, cancers and patient biology have been classified in order to deliver targeted treatment. The use of hormonal manipulation to treat breast cancer has been highly effective for more than 20 years for patients with the estrogen receptor, Mills says.

More recently, work on the HER2 receptor led to the introduction of Herceptin. Gleevec, effective in the treatment of chronic myeloid leukemia, a cancer characterized by a chromosomal abnormality known as the Philadelphia chromosome, has since been used to treat other diseases with particular molecular mutations. That somewhat serendipitous finding significantly increased the market for the drug.

Work on these drugs has set the template for research currently getting under way. "We have taken a quantum leap forward over the last five years," Mills says. "We now have the beginning of a real understanding of what is coming out of the human genome project, and we are developing drugs and tools and approaches to test our hypotheses."

THE LEAP IS occurring, he notes, at a time when fewer than 10% of drugs in clinical trials make it to market. "With all our new knowledge, we are actually doing worse than we were 15 years ago," he laments, noting that 50% of failures now occur in Phase III trials, maximizing the cost of failure. Mills says genetic research may also provide the solution to this problem by prompting the redesign of clinical trials.

"The key right now is the concurrent development of biomarkers and drugs," he says. "We need, right from the beginning, to put as much effort as we can into identifying a target, making a drug, and identifying which patient you put it in."

M. D. Anderson is testing this approach in a year-old lung cancer clinical trial pilot program called Biomarker-Integrated Approaches of Targeted Therapy for Lung Cancer Elimination, or BATTLE. The trials, involving patients for whom the traditional lung cancer therapies have failed, entails rounds of biopsies directed at progressively moving each patient toward a targeted treatment. The purpose of the program is to establish a biomarker-based clinical trial protocol for lung cancer while studying the molecular mechanisms of response and resistance to targeted agents.

"We are at a kind of crossroads where we have developed a large number of new agents with pharmacologically sound activities," says Roy S. Herbst, chief of thoracic medical oncology at M. D. Anderson and codirector of the BATTLE trial. "But clearly, to hit the home run, you have to match the right drug to the right patient. It's easier said than done."

Development of diagnostic techniques for characterizing disease tissue has also become a key focus at some biopharmaceutical companies. Celera, a pioneer in decoding the genome in the late 1990s, has shifted its focus to developing clinical tools for personalized medicine.

John Sninsky, vice president of discovery research at Celera, agrees with Mills that such tools are emerging from a tradition in medicine. "Targeted practices have been going on for a long time," Sninsky says, "we just haven't been using the molecular information that has emerged in recent years with the sequencing of the human genome and transcription analyses."

What is new, Sninsky points out, is the shift in the approach to complex conditions such as cardiovascular disease, Alzheimer's disease, rheumatoid arthritis, and cancer. "These are catch-all diseases that look the same," he says, "but when you scratch below the surface, you begin to understand that the underlying physiology of similar phenotypes can be fundamentally different." Researchers are now able to go beyond visual characterization to segregate diseases within traditional categories. "Molecular information is now being added to other information," he says.

The volume of that information is changing everything, according to Sninsky. "People used to study one gene their whole life. Now there are 25,000 genes you can study in parallel," he says. "You can investigate their impact on the disease process rather than just investigate the gene itself." Tools are now available to survey gene expression in RNA or proteins, he says.

The volume of data that needs to be processed in drug research is pushing toward a multidisciplinary pharmaceutical industry, Sninsky says. Chemists and biologists will need to routinely work with computational biologists and cancer bioinformatics experts to develop therapies based on the biological characteristics of disease and then develop trials based on the biological characteristics of patients. Sninsky says he left Roche, where he headed research for Roche Molecular Systems, in 2001 to apply the new tools of genomics to the development of such targeted therapeutics.

Similarly, David R. Parkinson, chief executive officer of Nodality, left big pharma—he had been vice president of clinical research for oncology for Novartis, in addition to working for Amgen and Biogen Idec—to develop biologically targeted diagnostics. He sees a more fundamental change occurring.

"There are pessimists who see personalized medicine as nothing but words, or say that it isn't economically feasible," Parkinson says. "But people don't understand they are in the middle of a revolution until it's over." He sees an entirely new approach to research emerging from the decoding of the genome. "I think we are changing our thinking about what malignancies are and how they should be classified and treated," he says.

According to Parkinson, drug researchers have become increasingly good at developing agents against targets but have made little headway in predicting how that match of therapeutic with target will translate across the range of human malignancies of that type. "Part of the inefficiency is the lack of a close connection between biology as we understand it, and the clinical classification of tumors" under the current system of categorizing patients.

Personalized medicine, Parkinson says, starts with changing that relationship. "As a drug developer, I spent the last 20 years either as a clinician or a clinical investigator or running government drug development programs with fantastic people, with great agents able, in vitro, to interfere effectively with what we expected were good targets based on our understanding of cancer biology," he says. The net result in most cases, however, was a highly imprecise match between a biologically targeted therapeutic and the patient.

CLOSING THE GAP is a matter of achieving better biological characterization, Parkinson argues. "The concept is that, by doing more biological studies around the interface of a particular therapeutic with a particular patient with some sort of malignancy, we will gain a better understanding," he says.

Nodality has developed a "cell flow" technology that characterizes biochemistry or physiology at the level of the individual cell. Parkinson says it is based on work done at Stanford University by Gary Nolan, associate professor of microbiology and immunology, to study the signaling networks of lymphocytes and malignant cells.

The company applies a laser beam to cells with fluorescence-labeled antibodies against specific phosphoproteins important in intercellular signaling pathways. "If the cell has signaling-related antigens, the cell will light up," Parkinson says. The technology provides phosphoprotein levels in individual cells, which indicate the extent to which signaling pathways in cancer cells are activated.

"In some sense it is a systems biology way of looking for pathways," Parkinson says. "You use the technology to measure the quantity of these phosphoproteins in leukemia cells, for example, just as you take them from the person. What you also do is expose those leukemia cells to a stimulation panel of cytokines or growth factors that are relevant to that leukemic cell population, or to a breast cancer cell population."

Parkinson says Nodality's business plan is to develop routine tests that match patients to appropriate therapies. The work touches on many of the precepts of personalized medicine, including that of translational research. It also raises the question of how new tests will be regulated and who will pay for them. And it highlights the need for collaboration among industry and academic researchers to expedite the development of affordable new diagnostic tools.

Most major pharmaceutical manufacturers have begun to incorporate tenets of personalized medicine, primarily in the use of biomarkers. Some are restructuring research to accelerate the process. Still, industry watchers are skeptical as to whether the large drug companies, which have long focused on broadly applied therapies expected to achieve blockbuster sales, are really committed to developing targeted therapies.

THE BLOCKBUSTER approach does not bode well for big companies grappling with personalized medicine, according to Garth Powis, who chairs the department of experimental therapies at M. D. Anderson. "If you think about it, the blockbuster drug mentality that a lot of drug companies still have is unsustainable," he says. "If you have a blockbuster drug, you have to treat a lot of patients. If you're treating only 10% of lung cancer patients, you don't have a blockbuster."

As can be expected, the big drug companies object to this characterization, although they are used to hearing it from academic researchers, says Wayne Rosenkrantz, external affairs director for evidence-based medicine at AstraZeneca. "But I know what is in our pipeline, and I have a good idea of GlaxoSmithKline's and Genentech's pipelines," he says. Rosenkrantz predicts big pharma will deliver its first genetically targeted therapeutic in three to five years, "with more coming at an accelerated pace after that."

He notes that AstraZeneca has been expanding the use of biomarkers in drug discovery. "A number of years ago, we began requiring a biomarker strategy for development," he says. "But we've learned that strategic thinking needs to start earlier."

Rosenkrantz sees real change as biomarkers and other aspects of personalized medicine take hold in the lab. "I would say it is new," he says. "It requires additional thinking about how we discover and develop processes, as opposed to just taking a broad population approach to development and commercialization. It really comes into tight focus as you think of taking a diagnostics-led approach."

Genetics-based diagnostics, he says, focus researchers on biocomplexity and translational research techniques.

"As you understand the molecular basis of the pharmaceutical you're developing, you start with assumptions about the basic mechanism," Rosenkrantz says. "The more you understand about the human implications, the more things change. This information needs to be fed back to the development of diagnostics. We hadn't done that. We're learning."

Roche built similar thinking into its year-old decentralized research program, which established five separate research and business organizations for distinct therapeutic areas such as cancer or inflammatory disease. Along with leaders for discovery, development, and business planning, each research group has a leader for clinical research and exploratory development, or CRED.

"The point of CRED is to place greater emphasis up front on identifying biomarker strategies and to direct more effort toward understanding disease and disease pathways," explains Andreas Wallnoefer, corporate head of the CRED initiative at Roche. "We are trying to apply personalized health care throughout the portfolio as a fundamental principle."

BIOMARKERS ARE NOT a new concept at Roche, Wallnoefer emphasizes, but they have not been implemented evenly across the entire research enterprise. "We have always used pharmacodynamics and surrogate markers in drug development," he says. "But with our focus on bridging clinical input to discovery, and our active effort to understand the systems of diseases through systems biology, we felt we needed a more systematic approach."

The company has determined that every drug discovery program needs to have a biomarker strategy. "This will not only help with personalized health care," Wallnoefer says, "it will improve early development decision-making. It is a strategy that will help us in terms of improving our R&D productivity—deciding which programs to progress." At the same time, he says, Roche wants to deliver drugs with better efficacy and lower risks of side effects.

Among the furthest advanced personalized medicine programs at Roche is a partnership with Plexxikon to study an investigational cancer therapy that selectively inhibits a mutated form of the BRAF kinase gene, which is involved in numerous cancer pathways.

James (Andy) Williams, head of molecular medicine for anti-angiogenesis programs at Pfizer, sees such developments not so much as a revolution but as part of an evolution. "Is personalized medicine new? I don't think it is," he says.

Pfizer, like most drug companies, has a tradition of adopting new technologies to conduct exploratory research, Williams maintains. He notes, for example, that cholesterol is a long-standing biomarker for cardiovascular disease and that the predictive value of biomarkers for cancer have been studied for years.

"We have invested in personalized medicine for a long time," he says, "though we had different terminology for it."

Pfizer is integrating personalized medicine into drug research in part through partnerships. For example, the firm recently announced it is collaborating with Source MDx on the development and validation of RNA-based pharmacodynamic and predictive biomarkers within Pfizer's cancer and inflammation therapy development programs.

MANAGERS AT the big drug companies say the quest to better understanding disease at the molecular level is intriguing work that presents a range of scientific hurdles. But they also point to regulatory hurdles and the question of how the health care benefits infrastructure will evolve to implement and pay for personalized medicine.

"Frankly, FDA doesn't know how to regulate this," Rosenkrantz says. "We really are working blind on what FDA and payers will require in the way of evidence that new diagnostics and therapies have medical value. We have some questions about whether we should be pursuing this course given all the ambiguity, but ultimately, the answer is that we have to do it, or else there will be no new drugs."

According to Steven I. Gutman, director of the office of in vitro diagnostic device evaluation and safety at FDA's Center for Devices & Radiological Health, the onus of developing a framework for personalized medicine falls to a large extent on researchers whose first job is to establish that tests are analytically sound and have clinical value.

"Establishing analytical performance for these tests is sometimes challenging, but if you haven't locked in analytical performance you're playing in a sandbox with no sand," he says. "It's a very heterogeneous research community in which there are those who have the analytics nailed down and those who don't."

Gutman argues that oversight of new practices might evolve within the current regulatory infrastructure. "I'm not sure it is a well thought out proposition that new science calls for a new paradigm at FDA," he says. "The new technology is daunting in terms of clinically credentialed technology. But people who think there is a radical paradigm shift have not looked carefully at what we do."

FDA's general inquiry into the function and efficacy of new diagnostics is straightforward, he says. "We ask questions like, 'Is the test accurate and repeatable? What is the associated drug impact?' " Gutman explains. "The sponsors should be embarrassed not to be able to answer them."

Gutman notes that personalized medicine is also being monitored closely at the National Institutes of Health. A recent report by the Secretary's Advisory Committee on Genetics, Health & Society at NIH suggests there are gaps in oversight of genetic testing and calls for public/private partnerships to assess and fill them.

Regulatory oversight should advance as the science does, according to Gutman, who quips that for once it may be the science holding things up. "It's not the damned FDA, it's the damned science," he says. "It's my view that if sponsors have good science and they bother to write it down in a coherent package, it goes flying through FDA faster than a speeding rocket."

INSURANCE FOR personalized clinical treatment presents another hurdle. Experts point to the irony that, at the dawn of an age of personalized medicine, the doctor-patient relationship is at its least personal. Doctors are paid by the number of patients they see, not for the time they invest in a personalized health maintenance regime or for health outcomes.

IBM is among the health care insurance purchasers trying to change the system. "We are really unhappy with the current structure, in which we can buy episodic care based around procedures or processes, but we can't purchase a comprehensive care structure with meaningful doctor-patient relationships," says Paul Grundy, director of health care, technology, and strategic initiatives at IBM. "We can buy an amputation for a diabetic, but we can't buy the kind of doctor-patient relationship that will prevent that amputation from needing to be done."

Grundy also chairs the Patient-Centered Primary Care Collaborative, a 107-member coalition of private and public buyers of health care benefits, public health care agencies, and major insurance companies investigating new health care payment structures that will employ diagnostic-based preventive medicine and other precepts of personalized medicine. The collaborative is coordinating comprehensive health care pilots in at least six locations around the country.

Coming to an agreement on a payment structure poses a problem, according to Kathryn A. Phillips, professor of health care and economics at the University of California, San Francisco. "Where is the evidence in personalized medicine? What do payers want in terms of evidence? How do they view and use it? This is the key to coverage and reimbursement," she told attendees at a conference hosted by Burrill & Co. in San Francisco last November. "The payers are reluctant to be proactive unless they see the value of these products."

The problem, she says, stems from a lack of clinical data outcomes and economic evaluations, as well as a lack of industry incentives to provide adequate information for coverage decisions. Technical assessment guidelines need to be developed and risk needs to be shared. Her assessment echoes a call by the Secretary's Advisory Committee on Genetics, Health & Society at NIH for public/private partnerships to close the gaps in regulatory oversight.

The public is not waiting. Awareness of personalized medicine is increasing with the rise in news coverage of technologies for early cancer detection and determining genetic propensities for disease. Do-it-yourself genetic testing kits are now available, and private companies offer routine screening. For example, Navigenics, a Redwood Shores, Calif.-based genomics testing service, charges $2,500 for an initial genetic scan and counseling session, and $250 per year for its customers, which the firm calls members, for follow-up scans.

Navigenics CEO Mari Baker says public interest in taking advantage of genomics technology will ramp up, with the effect of lowering costs and making the technology more accessible. The ability to use genetic data to attain early treatment for disease will help encourage acceptance of personalized medicine, outweighing concerns about abuse of data by employers or others, Baker says. President George W. Bush, she notes, has said he will sign a bill currently before Congress that prevents the use of genetic testing by employers or insurers.

PERHAPS THE BIGGEST selling point for personalized medicine is the promise of cost savings. AstraZeneca's Rosenkrantz says the potential efficiency gain is a major incentive for insurers to work with the research and medical communities to build personalized and preventive medical treatment into the health care system.

Meanwhile, innovators are finding their way into the mainstream. Genomic Health, for example, has developed a diagnostic tool for breast cancer that CEO Randal W. Scott says currently holds 30% of the market for testing node-negative, estrogen receptor-positive patients, who account for half of all breast cancer patients.

"It does take a significant effort to get reimbursed," Scott acknowledges. "But our test meets American Society of Clinical Oncology and National Comprehensive Cancer Network clinical guidelines. All the major payers are onboard, including Medicare and United Healthcare."

Genomic Health's Oncotype DX uses gene expression technology similar to RT-PCR (reverse transcriptase-polymerase chain reaction) scanning, a mainstay in HIV viral load testing, to test tumor biopsies. The company is adapting the technology to scan other cancers, including colon, lung, and prostate. Scott says a colon cancer product may be validated next year and on the market by 2009.

From Scott's perspective, medicine is in the midst of a "revolution driven by tools." He compares the growth of acceptance of the technology to Moore's Law (the axiom that the number of transistors on a computer chip would double every two years) in the electronics industry. "The potential power of new technology is accelerating rapidly and will significantly impact drug use," he says. FDA's interest is clear enough to Scott that he sees the potential for increased regulation in the clinic and the laboratory.

"It is important to get the balance right," Scott says, adding that effective translational research will be crucial to success. "The key to it all is good clinical medicine."

Chemical & Engineering News

ISSN 0009-2347

Copyright © 2009 American Chemical Society

[Top of Page]