Saturday, October 24, 2009

Personalized medicine research forum

Yesterday afternoon I attended a Personalized Medicine research forum at ASU's Biodesign Institute, sponsored by ASU's Office of the Vice President for Research and Economic Affairs (OVPREA) and hosted by Dr. Joshua LaBaer of ASU's Virginia G. Piper Center for Personalized Diagnostics.

The forum's speakers covered both the promise and problems and issues raised by the developing field of personalized medicine, which involves the use of molecular and genetic information in medical diagnosis and treatment. A few highlights:

Introduction (Dr. LaBaer)
Dr. LaBaer pointed out that these new diagnostics cost a great deal of money to develop, but they have the potential for cost savings, for instance, if they can be used to identify forms of disease that will not benefit from very expensive treatments. He gave the example of Genomic Health, which has developed a test for early stage breast cancer to determine if women will or won't benefit from adjuvant therapy (chemotherapy to prevent recurrence). A test that costs even a few thousand dollars to perform is something insurers will be willing to pay for if it has the potential of saving tens of thousands of dollars of expense on chemotherapy that will not provide any benefits. On the other hand, the mere promise of early detection of susceptibility for disease has the potential for overtreatment and an increase in healthcare expenses. This problem was discussed by a number of speakers, with particular bad potential consequences in the legal realm.

Personalized Diagnostics (Dr. LaBaer)
Dr. LaBaer talked briefly about his own lab's work in biomarker discovery and cell-based studies. In biomarker discovery, his lab is working in functional proteomics, using cloned copies of genes to produce proteins and building tests that allow examination of thousands of proteins at a time. His lab, formerly at Harvard and now at ASU, has 10,000 copies of human genes and 50,000 copies of genes from other animals, which are made available to other researchers. (There's more information at the DNASU website.)

The goal of biomarker discovery is to greatly improve the ability to find markers of human health using the human immune system, by identifying antigens that are markers for disease. The immune system generates antibodies not just in response to infectious disease, but against other proteins when we have cancer. Tumor antigens get into the bloodstream, though they may only appear in 10-15% of those who have the disease. Rather than testing one protein at a time, as is done with ELISA assays, LaBaer's lab is building protein microarrays with thousands of proteins, tested at once with blood serum. Unlike old array technology that purifies proteins and puts them into spots on arrays, where the proteins may degrade and lose function, their method involves printing the DNA that encodes the gene on the arrays, then capturing proteins in situ on the array at the time the experimental test is performed.

LaBaer's lab's cell-based work involves tryng to identify how proteins behave in cells when they are altered, in order to find out which pathways contribute to consequences such as drug resistance in women with breast cancer, as occurs with Tamoxifen. If you can find the genes that make cancer cells resistant, you can then knock them out and cause those cells to die. They tested 500 human kinases (5/7 of the total) and found 30 enzymes that consistently make the cancer cells resistant. Women with a high level of those enzymes who take Tamoxifen have quicker relapses of cancer.

Complex Adaptive Systems Initiative (George Poste)
George Poste, former director of ASU's Biodesign Institute and former Chief Scientist and Technology Officer at SmithKline Beecham, talked about the need to replace thinking about costs in the healthcare debate with thinking about value. The value proposition of personalized medicine is early detection, rational therapeutics where treatment is made based on the right subtype of disease being treated, and integrative care management where there's better monitoring of the efficacy of treatments. He said that the first benefits will come from targeted therapy and this will then overlap with individualized therapy, as we learn how our genome affects such things as drug interactions. He was critical of companies like 23andme, which he called "celebrity spit" companies, which do little more than give people a needless sense of anxiety about predispositions to disease that they currently can do nothing about except eat right and exercise.

Poste also had criticisms for physicians, pointing out that it takes 15-20 years for new innovations to become routinely adopted, and many physicians don't use treatment algorithms at all. Oncologists, he said, make money from distributing treatments empirically (that is, figuring out whether it's effective by using the treatment on the entire population with the disease) rather than screening first, even where tests exist to determine who the treatment is likely to work on. He said that $604 million/year in health care costs could be saved by the use of a single colon cancer screening test, and not proceeding with treatment where it isn't going to work. Today, where 12-40% of people are aided by treatments that cost tens of thousands of dollars, 60-88% of that spending is being wasted. With the aging population, he said that Humana will in the next several years see all profits disappear, spent on expensive treatments of people who don't respond to them.

Pharmaceutical companies are beginning to do diagnostic test development alongside drug development now, and insurers will push for these tests to be done. Poste suggested that we will see the emergence of "no cure, no pay" systems, and noted that Johnson & Johnson has a drug that has been introduced for use in the UK under the condition that the company will reimburse the national health care system for every case in which it is used but doesn't work. Merck's Januvia drug for type II diabetes similarly offers some kind of discount based on performance.

Poste pointed out another area for potential cost savings, related to drug safety. With some 3.1 billion prescriptions made per year, there are 1.5-3 million people hospitalized from drug interactions, 100,000 deaths, and $30 billion in healthcare costs, though he noted this latter figure includes caregiver error and patient noncompliance.

He bemoaned the "delusion of zero risk propagated by lawyers, legislatures, and the media," and pointed out that the FDA is in a no-win situation. (This is a topic that's been recently covered in two of my classes, my core program seminar and my law, science, and technology class with Prof. Gary Marchant. If the FDA allows unsafe drugs to be sold, then it comes under fire for not requiring sufficient evidence of safety. If, on the other hand, it delays the sale of effective drugs, it comes under fire for causing preventable deaths. The latter occurred during the 1980s with AIDS activists protesting against being denied treatments, described in books such as Randy Shilts' And the Band Played On and Steven Epstein's Impure Science. This led to PDUFA, the Prescription Drug User Fee Act of 1992, under which drug companies started funding FDA drug reviewer positions through application fees to help speed approval. That has been blamed for cases of the former, with the weight-loss drugs Pondimin and Redux being approved despite evidence that they caused heart problems. That story is told in the PBS Frontline episode "Dangerous Prescription" from November 2003.)

Poste pointed out that there have been 450,000 papers published which have claimed to find disease biomarkers, of which the FDA has approved only five. But he didn't blame the FDA for delay in this case, because this consists of a mass of bad studies which he characterized as "wasteful small studies" with insufficient statistical power. In the Q&A session, he argued that NIH needs to start dictating clear and strong standards for disease research, and that it has abrogated its role in doing good science. He said that "not a single national cancer study with sufficient statistical power" has been done in the last 20 years; instead research is fragmented across academic silos. He called for "go[ing] beyond R01 grant mentality" and building the large, expensive studies with 2,500 cases and 2,500 controls that need to be done.

He also raised challenges about the "very complex statistical analysis required" in order to do "multiplex tests" of the sort Dr. LaBaer is trying to develop. And he pointed out the challenge that personalized medicine presents for clinicians, in that "only about six medical schools have embraced molecular medicine and engineering-based medicine." Those that don't use these new techniques as they become available, he said, "will open themselves up to malpractice suits."

Science and Policy (David Guston)
David Guston, co-director of ASU's Consortium for Science, Policy, and Outcomes (CSPO) and director of ASU's Center for Nanotechnology in Society (CNS) spoke about "cognate challenges in social science" and how CNS has been trying to develop a notion of "anticipatory governance of emerging technology" and devising ways to build such a capacity into university research labs as well as broader society, to allow making policy decisions in advance of the emergence of the technology in society at large. He described three capacities of anticipatory governance--foresight, public engagement, and integration, and described how these have been used at ASU.

Foresight: Rather than looking at future consequences as a linear extrapolation, CNS has used scenario development and a process of structured discussions based on those scenarios with scientists, potential users, and other potential stakeholders, about social and technical events that may be subsequent consequences of the scenarios. This method has been tested with Stephen Johnston's "Doc-in-a-Box" project at ASU's Center for Innovations in Medicine, which Guston said led to some changes in the conceptualization of the technology.

Public Engagement: The "scope and inclusion of public values is important for success," Guston said, and gave as an example the "national citizens technology forum" that CNS conducted in six locations to look at speculative scenarios about nanotechnology used for human enhancement. These were essentially very large focus groups whose participants engaged in "informed deliberation" over the course of a weekend, after having read a 61-page background document and spending the prior month engaging in Internet-based interaction.

Integration: Guston described the "embedding of social scientists in science and engineering labs," to develop productive relationships that help lab scientists identify broader implications of their work while it's still in the lab rather than after it's introduced to the general public.

Guston suggested that there might be other ways of implementing "anticipatory governance" in the form of legislative requirements or standards and priorities set by program officers at funding organizations, but that the lab setting is "the best point of leverage at a university" and can set an example for others to follow.

Clinical Perspective (Larry Miller)
Larry Miller, Research Director at the Mayo Clinic in Scottsdale, spoke about the healthcare provider's approach to personalized medicine. He said that Mayo is committed to individualized care, and that now that we are beginning to understand the power of human variation, these new developments have "to be transformational for providers or they won't survive." He suggested that the future of medicine will move from reactive and probabilistic to more deterministic selection of treatments based on diagnoses. He emphasized the need for education for doctors, and pointed out that "standards of care will become outmoded," which is "disruptive to law and [insurance] coverage." He said that Mayo sees a big challenge of complexity, where what was one disease (breast cancer) is now at least ten different subdiseases. Doctors need to make their treatment decisions on the detail, to predict how the disease will behave, and choose the best drugs possible based on safety, effectiveness, and cost-effectiveness.

Miller pointed out that this requires interdisciplinary work, and said that Mayo in Arizona has a huge advantage with its relationship with ASU, where so much of this work is going on. While Mayo has scientific expertise in a number of areas, these new technologies draw on expertise from beyond medicine, in particular informatics and computational resources needed to build an effective decision support system that will become essential for doctors to use in a clinical setting.

He talked about Mayo's program for individualized medicine, which involves not just incorporating new developments in diagnostics and therapeutics, but in regenerative medicine for repair, renewal, and regeneration of deficits.

Mayo has had electronic medical records for the last 15 years, on 6 million people, but these are kept in multiple incompatible systems and were not built with research in mind. They hope to improve their systems so that it can be used in an iterative process to learn more about the efficacy of therapies, and so therapies can be combined with "companion diagnostics for monitoring progression, recurrences, and response to therapy."

Like Poste, he raised objections to the companies that market gene sequencing directly to individuals, which just "scare people inappropriately," but identified learning about disease predispositions as an important part of these developing technologies. We need to develop methods of risk analysis that can help people correctly understand what these predispositions mean.

He sees the future as having three waves--the first wave will be the new diagnostics, the second wave improvements in clinical practice and therapy, and the third wave embedding the new technology into the healthcare system, with significant changes to policy and education.

Health Informatics (Diana Petitti)
Diana Petitti, former CDC epidemiologist and former director of research for Kaiser Permanente, where she built a 20-year longitudinal data repository for its 35 million members, spoke about the importance of health informatics. (She is now a professor in ASU's Department of Biomedical Informatics.) Dr. Petitti raised concerns about how in the United States we are "loathe to deny anyone anything" in terms of medical treatments, but in fact "we do deny lots of people lots of things." She worried that personalized medicine has the potential to lead to greater maldistributions of healthcare, with the "haves" getting more and better treatment and the "have nots" getting less and worse treatment, unless we plan carefully. She advocated evidence-based medicine and assessing value of treatments to be deployed to the general population.

Dr. Petitti brought up as an example the fact that oral contraceptives result in a 2x-10x increase in the likelihood of a venous thrombotic event, and that the Factor V Leiden gene is predictive of susceptibility to that consequence, but no screening is done for it. Why not? Because the test only predicts 5% of those who will have the event, it's a very expensive test, and we don't have good alternatives for oral contraceptives. These kinds of issues, she suggested, will recur with multiplex diagnostics.

She explicitly worried that "we have dramatically oversold preventive medicine" and doesn't think it's likely that savings from prevention will allow coverage for more extensive treatment. She advocated that everyone in the field see the film "Gattaca," and stated that ASU provides "unique opportunities to train people to think about these issues" using "quantitative reasoning and probabilistic thought." She concluded by saying that we need to "work towards rational delivery of healthcare that optimizes public health."

Law (Gary Marchant)
Prof. Gary Marchant of the Sandra Day O'Connor School of Law at ASU, who has a Ph.D. in genetics and is the executive director of ASU's Center for the Study of Law, Science, and Innovation (formerly Center for the Study of Law, Science, and Technology), spoke about legal issues. First he listed the many programs available at ASU in the area, beginning with the genetics and law program that has been here for 10 years and was the reason he first came to ASU. Others include a new personalized medicine and law program at the Center for Law, Science, and Innovation, a planned center on ethical and policy issues regarding personalized medicine in conjunction with the Biodesign Institute, CSPO, TGEN, Mayo, etc., and research clusters at the law school on breast cancer, warfarin, and personalized medicine. He also gave a plug for an upcoming conference March 8-9, 2010 at the Arizona Biltmore sponsored by AAAS and Mayo, which also has a great deal of corporate support.

Prof. Marchant indicated that liability is the biggest issue regarding personalized medicine, and he sees doctors as "sitting ducks," facing huge risks. If a doctor prescribes a treatment without doing a corresponding new diagnostic test, and that has complications, he can be sued. If he does the diagnostic test, it shows a very low likelihood of a disease recurrence, and advises against the treatment, and then the patient ends up being one of the rare people who has the recurrence, the doctor can be sued. The doctor is really in a damned-if-you-do, damned-if-you-don't situation. The insurers and pharmaceutical companies are at less risk, since they have already developed enormous resources for dealing with the lawsuits that are a regular part of their existence. In a short discussion after the forum, I asked Prof. Marchant if doctors would be liable if they performed a diagnostic test, found that it showed a low likelihood of recurrence or benefit for a treatment, and then recommended the treatment anyway, knowing the insurance company would refuse to pay for it--would that shift the liability to the insurance company? He thought it might, though it would be unethical for a doctor to recommend treatment that he didn't actually think was necessary, and there's still the potential for liability if the insurance company pays for the treatment and the treatment itself produces complications. It seems that this problem really needs a legislative or regulatory fix of some sort, so that doctors have some limitation of liability in cases where they have made a recommendation that everyone would agree was the right course of action but a low-probability negative consequence occurs anyway.

Prof. Marchant observed that the liability issues are particularly problematic in states like Arizona, where each side in the suit is limited to a single expert witness. He said there is "no clear guidance or defense for doctors," and the use of clinical guidelines in a defense has not been effective in court, in part because doctors don't use them.

A few additional points of interest from the Q&A sessions (some of which has already been combined into the above summaries):

Dr. LaBaer pointed out that most markers for diseases don't seem to have any role in the cause of the disease, such as CA25 and ovarian cancer. So his lab is looking not just for biomarkers, but for those that will affect clinical decisions. 4 out of 5 positive results in a mammography for breast cancer are actually cases where there is nothing wrong and the woman will not end up getting breast cancer, but some procedure ends up being undergone, with no value. So he wants to find a companion test that can tell which are the 4 that don't need further treatment.

George Poste pointed out that baby boomers are going to bankrupt the system as they reach the end of their lives, and about 70% of the $2.3 trillion in healthcare spending is spent in the last 2-3 years of life, with many treatments costing $60K-$100K per treatment cycle on drugs that add 2-3 weeks of life. The UK's National Institute of Clinical Excellence has been making what are, in effect, rationing decisions by turning down all of the new cancer drugs that have come along because they have such great cost and such minimal benefit. He asked, "how much money could you save with a 90% accurate test of who's going to die no matter what you do?"

Prof. Marchant said more about legal issues involving specimen repositories, including a case at ASU. The developer of the prostate-specific antigen (PSA) test, William Catalona, had a specimen repository with 30,000 tissue samples at Washington University, that he wished to take with him to Northwestern University when he took a new position there. He began asking patients for permission to move the samples, and 6,000 gave permission. But Washington University sued him, claiming that the samples were property of the university. Patients pointed out that their consent agreement gave them the right to withdraw their samples from future research and they had only consented to research on prostate cancer, but federal judge Stephen Limbaugh ruled in favor of the university and that patients had no property rights in their tissue. This ruling has reduced incentives for patients to consent to give specimens for research.

A current lawsuit against ASU by the Havasupai Indian tribe involves blood samples that were given for a study of diabetes by researchers who are no longer at ASU. They wanted to take the samples with them, and samples had also been given to other researchers for use in studies of schizophrenia and the historical origins of the tribe, even though informed consent was apparently only given for the diabetes research. Although this case was originally dismissed, it was recently reinstated.

Other cases involve patent protection of genetic information. About 25% of the human genome is patented, including Myriad Genetics' patent on the BRCA1 and BRCA2 genes which are predictive of breast cancer and can only legally be tested for by Myriad. This case is likely to go to the U.S. Supreme Court regarding the issue of whether human genes can be patented. The courts so far have ruled that a gene in isolation outside of the human body is patentable, even though (in my opinion) this seems at odds with the requirement that patents be limited to inventions, not discoveries. There has already been a legislative limitation of patent protection for surgical procedures for the clinical context, so that doctors can't be sued for patent infringement for performing a surgery that saves someone's life; it's possible that a similar limitation will be applied on gene patents in a clinical context, if they don't get overturned completely by the courts.

These gene patents create a further problem for the multiplex tests, since they inevitably include many patented genes. Prof. Marchant observed that someone from Affymetrix spoke at an ASU seminar and stood up and said they were building their GeneChip DNA microarrays for testing for the presence of thousands of genes, and were ignoring gene patents. They were subsequently sued. Dr. LaBaer stated that his lab is doing the same thing with cloned genes--they're cloning everything and giving them away, without regard to patents.

The session was videotaped and will be made available to the public online. I will add a link to this posting when it becomes available.

If you've read this far, you may also be interested in my summary of Dr. Fintan Steele's talk at this year's The Amazing Meeting 7, titled "Personalized Medicine or Personalized Mysticism?", in my summary of the Science-Based Medicine conference that took place just prior to TAM7, and in my short summary of Dr. Martin Pera's talk on regenerative medicine and embryonic stem cells at the Atheist Alliance International convention that took place earlier this month.


tom said...

(Off Topic - couldn't find an email address on your blog):

Richard Carrier will be giving a talk nearby on Nov. 8; I thought I'd drop you a note in case you didn't know:

Lippard said...

Thanks, Tom.