Geoffrey S. Ginsburg directs the Center for Genomic Medicine, one of five centers that form the Institute for Genome Sciences & Policy at Duke. The Institute’s overall goals are to advance the revolution in genomic research and address its larger implications for society. To accomplish this ambitious mission, each center brings a particular expertise to the whole, from research, to computation, to legal and social acumen. The key to Duke’s approach—and to the overall advancement of personalized medicine—is to integrate the best of many disciplines into an entirely new field that will reflect state-of-the-art healthcare in the genomic age.
Ginsburg discusses his laboratory’s recent breakthroughs in the context of larger clinical, economic, and policy studies about personalized medicine now being conducted at Duke.
Q: How do you define personalized medicine?
A: To me, personalized medicine is the use of all kinds of clinical and molecular information to optimize treatment and health outcomes for individual patients. Personalized medicine aims to answer precisely, “Who should be treated?” and “How should we treat them?”
Q: How can we expect medicine to be so individualized, given that researchers cannot conduct studies on each and every person in the world?
A: While our goal is to individualize treatment, it is a misnomer to think of personalized medicine as the study of individuals. We study populations of people with a particular disease or who are taking a particular drug. We then measure the course of disease for individuals within that population and/or whether they respond to treatment or not. My view is that studies in large populations and at the level of the individual are ultimately going to define personalized medicine.
Q: Can you give me an example of a population for which personalized medicine would be beneficial?
A: Yes. Let’s look at one of the most deadly cancers, lung cancer. In the early stages of non-small-cell lung cancer, patients have their tumors surgically removed. If the tumors measure less than 3 cm, they are called Stage 1A and typically patients are not treated with chemotherapy following surgical removal of their tumor. Unfortunately, in about 30 percent of these patients, the cancer recurs and the patients usually die within five years.
Q: Why not, then, give all patients more aggressive treatment rather than simply observing them after surgery?
A: The commonly prescribed “cytotoxic” therapies for cancer, as most people know, have a plethora of nasty side effects. And oftentimes, the treatments do not have any tangible benefit. If we look at this at a population level– that is, looking at all the patients with Stage 1A lung cancer–there is no compelling reason to give everybody chemotherapy, especially when therapy overall doesn't show any long-lasting benefit, but is virtually guaranteed to cause suffering.
Q: So you have a dilemma: give everyone chemotherapy and subject 70 percent to an unnecessary drug with toxic side effects, or give no one chemotherapy and risk 30 percent dying.
A: Exactly. It is an awful predicament. Ultimately, it’s a question of both the kind of disease each patient has and how each will respond to the various chemotherapeutic agents available.
Q: How can personalized medicine address those questions?
A: We now have an approach to identify the type of lung cancer each patient has and, therefore, which patients will respond to chemotherapy. We studied the molecular characteristics of patients’ tumors using DNA chips or microarrays, which analyze patterns of gene expression. From that data, we have developed a novel “prognostic gene signature” that can identify which patients with Stage 1A lung cancer will have recurrence after surgery and which will not.
Q: What do you mean by “prognostic gene signature?”
A: It’s a specific group of genes whose activity defines the type of tumor. We recently reported in the New England Journal of Medicine, that we have used such a signature to define which tumors are “aggressive” or are “high risk” for recurrence, and which are not. Patients who are in the genetically favorable prognostic group are not likely to experience recurrence. Those in the high-risk group are. We also showed that this information is highly reproducible in several groups of patients with lung cancer. Thus this prognostic gene signature or “Lung Metagene Predictor,” as we called it in the Journal, can identify individuals who would be candidates for more aggressive therapy after their surgery.
Q: What are next steps?
A: The immediate one is a clinical trial, which will be the first of its kind in lung cancer. The study will mark the first time that a genomic test (in this case involving 2100 genes) will be used to guide treatment in a prospective study.
Q: So this research is different from a retrospective study: for example, looking at tumor tissue after the patients either experienced recurrence of cancer or not?
A: Yes. About 1000 patients will be tested and get a score for their tumor. Individuals who receive a low risk score will get the usual care for Stage 1A lung cancer–surgery without chemotherapy. And individuals with a high-risk score will be randomized to receive adjuvant chemotherapy. This is a ‘first of its kind’ clinical trial in which a microarray test will be used to assign patients to one therapy vs. another.
Q: If successful, what will be the eventual impact of such a study?
A: If the patients who get chemotherapy turn out to benefit from this prognostic marker, it will be huge, because now we are talking about tens of thousands of patients throughout the world, who would potentially benefit from more aggressive treatment and would have otherwise been left to die.
Q: Given your test and others like it in the works, where should we be focusing our resources to make personalized medicine, as you’ve defined it, come to pass?
A: At the Center for Genomic Medicine, we have a couple of organizing principles that dictate where we will focus our research. One, as with the lung cancer research, is to address the question of “Who do we treat?” The second is to do so in a reasonable timeframe so that we bring the technology into the clinic. We are focusing on more short-term areas that impact medicine sooner rather than later.
Q: What are those areas?
A: One is patients’ responses to drugs, pharmacogenomics, which answers the question, “How should we treat individuals?” Another is quality of life. In our lung cancer study, for example, the question of survival is obviously important to the patient, but so is living with the disease and not knowing if you are going to live or die for two, five, or 10 years. That, obviously, impacts quality of life.
Q: What about the economic impact?
A: One of the hot topics among health economists and policy scholars, and rightly so, is who will pay for reimbursement of personalized medicine testing as a whole? It is the big question and it is quite open.
Q: Given that the center that you direct is part of Duke¹s larger Institute for Genome Sciences & Policy, are you trying to address those questions as well?
A: Most certainly. We have a number of economists and policy analysts in the health sector at Duke. They are working closely with us. Most of the economic analyses that I have seen in the area of personalized medicine have been retrospective, and while those give us a little bit of a flavor for what could be done and what needs to be done, I still think that the major insurance companies and Centers for Medicare and Medicaid Services (CMS) are looking for prospective trial data.
Q: Are you referring to studies such as your lung cancer study?
A: Absolutely. In clinical trials such as ours, we hope to show both health and economic outcome benefits; third-party payers and CMS are really looking for these types of data as proof. These studies can and should have economic analyses embedded within them so that one can show that having genomic-guided medical care actually leads to cost savings at the healthcare systemic level. That would at least, in part, address some of the barriers that I see in convincing third-party payers to get behind reimbursing these sorts of tests.
Q: Are there other social or policy issues that you are considering that will help make personalized medicine a reality?
A: Indeed. Among them is an assessment of people’s attitudes toward genetic and genomic testing. There are real extremes. For example, in the Durham community, we are assessing attitudes and perceptions about genetics and genomics in groups of lower socioeconomic status and educational levels versus the other extreme, members of Duke’s Executive Health Programs who tend to come from very high socioeconomic backgrounds. Ultimately, we want to be better prepared so that when we do have a personalized test, we will know what challenges and community values will make it easy or difficult to deploy. We can't just drop personalized medicine in the community’s lap without first doing our homework.
Q: Why do you see these social implications as so important?
A: Personalized medicine is not simply a promise–is becoming real today. But people play a critical role in how quickly or fully that reality shapes itself. People have to take charge of their healthcare and get educated about where the opportunities are for them to be proactive. Eventually, we hope they will see the need to practice more preventative healthcare based on molecular information. That participation is part of the definition–and will ultimately determine the future–of personalized medicine.
|
