Bonnie LaFleur has over fifteen years of experience in statistical and cancer research, and in teaching both physicians and degree seeking statistics/biostatistics students. She is currently a senior manager of biostatistics and data management at Ventana Medical Systems, Inc , which is the tissue diagnostic arm of Roche, Inc. Her primary research interests are in the interface between biology and quantitative methods, specifically the application of statistical methods to cancer biology and novel biologic technologies. Her current work focuses on statistical methodologies used in biomarker development and verification of tissue biomarkers used as companion diagnostics for cancer therapy.
+ Full Transcript
Bob Long: We all know the medical field is constantly changing and one of those changes may have a major impact on you. I'm talking about the field of personalized medicine, where doctors are in the early stages of tailoring medical treatment to your individual needs. It's already impacting care for diseases such as breast cancer, melanoma, or heart disease. I'm Bob Long; welcome to Stats and Stories, a program that looks at the statistics behind the stories and the stories behind the statistics. And today we're going to discuss how medicine is getting personal. Before we talk to our special guest, Stats and Stories reporter Lucy Borchers tells us more about a company that's working to make individualized medicine a reality.
Lucy Borchers: Imagine having a family doctor who understands your genetic makeup and prescribes medicine to meet your specific needs. We're still a couple of decades away from making that a reality, but researchers have made tremendous strides in the last 10 years. Stat King Clinical Services President Dennis King is heavily involved in research on what we call personalized medicine. The idea is to use the human genome to develop drugs, medical devices or treatment therapies aimed at the individual.
Dennis King: That's the long-term goal is that each and every individual would have a very specialized treatment regimen just for that particular person.
Borchers: Dennis King says researchers currently are trying to develop treatments for small groups of individuals. That's a huge change from the past when medicine was designed for much larger groups of people battling a disease like lung cancer. King believes researchers are gaining a better understanding of the human genome.
King: Certain genes being turned on or turned off can trigger a disease in a group of individuals. So currently, we're at the level of looking at groups of individuals that exhibit a certain gene or lack of a certain gene.
Borchers: Stat King Clinical Services is a product development company in the pharmaceutical and medical device field. King says they're involved in several types of diagnostic tests.
King: We're trying to develop a diagnostic that identifies a certain gene in a group of people and we know that that gene can have a big effect on how effective a drug or a treatment strategy could be on the group of individuals possessing that gene or lacking that gene.
Borchers: For example, Stat King has dealt with medicines to battle depression. In the past, doctors tried several different types of drugs and different dosages to fight depression. That lengthened the treatment time for the person suffering from depression. King says his research can help shorten that.
King: What the diagnostic tool can do there is identify a certain gene that's turned on, and that will trigger the doctor to provide a specific depression drug for that individual. That eliminates this long search for the correct drug and immediately directs a physician to a drug that would be the most effective in treating that individuals' depression.
Borchers: Dennis King feels the next step for researchers is to develop new drug compounds that are specific to certain combinations of genes.
King: Right now, we're identifying subjects with a certain gene and then giving a particular medicine. I think as we go down the road, we'll see drugs that are specifically developed and compounded for a specific gene set. That's where we're headed. I don't think we're way away from that. We're probably looking over the next 25 years or so, we're going to see some great advances in that area as well.
Borchers: The idea of developing drugs for a small set of individuals would revolutionize drug development in America, and King hopes it saves money.
King: The way it's done today is we try to develop a drug that works for as many people as possible, and that's the way the US regulatory system is set up to approve drugs in that way. So it'll be a completely different regulatory set-up that we will need for this individualized medicine approach, and that of course may eliminate some of the federal regulations that are currently in place and make drug development cheaper.
Borchers: But Stat King President and CEO Dennis King believes it is too early to know how individualized medicine will impact health care costs overall since government regulations will have to change. For Stats and Stories, I'm Lucy Borchers.
Long: Thanks to Lucy for that special report. Joining me on Stats and Stories for our discussion of personalized medicine is Miami University Statistics department chair John Bailer, our other regular panelist Media, Journalism and Film chair Richard Campbell is not with us today. Our special guest is Bonnie LaFleur, a senior manager at Ventana Medical Systems. Bonnie has more than 15 years of experience in statistical and cancer research. We want to stress the views expressed by Bonnie on Stats and Stories do not necessarily represent those of Ventana Medical Systems. So Bonnie, what do we mean when we talk about the word "personalized medicine?"
Bonnie LaFleur: Personalized medicine is generally seen as a way to ensure that the right patient gets the right treatment at the right time. And we have various ways of measuring biological processes in people's bodies and we would like to tailor the treatment that they get for diseases to those processes.
Long: So for example, I'm thinking, let's take something like breast cancer. There would, I suppose, be a protocol that most doctors would follow, the kind of medicine that somebody would receive, but what you're talking about is everybody's body is a little different and how they react therefore might be different.
LaFleur: Exactly, and in part what we need to do is be able to measure these individual differences so that we can tailor these treatments for those individuals.
Long: John Bailer, we'll go to you.
John Bailer: Well, sort of in contrast to that, how does that differ from historical practice in terms of how treatments were assigned?
LaFleur: Well current practice in many cases, take cancer for instance, if a person were to have breast cancer and they had a certain stage of breast cancer then they would get a certain chemotherapy. Sometimes, in addition to that chemotherapy, they'll also be given other treatments based on the characteristics of their tumor, and these additional treatment regimens actually reduce their risk of occurrence, sometimes up to 35%.
Bailer: So what do you mean by treatment regimen?
LaFleur: So chemotherapy and radiation and surgery are all three treatment regimens and in addition to that we sometimes give patients hormonal therapies which modify the body's ability to react with certain proteins; estrogen being one that many breast cancers feed on estrogen, as it were, and so you'd like to reduce the body's ability to feed on those, so they might get a certain therapy to help their body stop that process.
Long: Now as far as you personally and how you fit into all this, I mean you're involved on the research side, kind of talk to us a little bit about what you've been doing because I know cancer research is one of the key areas for personalized medicine.
LaFleur: Yes, well at Ventana, we are interested in both finding these new targets and finding ways of measuring those targets and we're also interested in evaluating the diagnostic itself to determine whether a patient is positive or negative for a particular treatment that might be targeted to their type of tumor. So we have two types of statistical studies, at least, that we work on.
Bailer: Do you want to say more about what that might be?
LaFleur: Well yes, so one of the types of research that we do is finding targets that will stratify patients into high and low risk and high and low risk can then be used to determine whether or not a patient would get a standard therapy and so that would be something that's already being used in practice like a specific chemotherapy. Another new medicinal program is the companion diagnostic program and this is a very pre-specified, targeted therapy based on patients that have their tumors express certain characteristics and these targeted therapies are designed solely for patients only who express these proteins and they wouldn't necessarily work at all if patients do not have these targeted proteins.
Long: Does this also affect, I'm assuming, the length of time that the treatment takes place, because if you're discovering things that are unique to one person, it may change their regimen completely compared to somebody else.
LaFleur: It could. I think currently most of the targeted therapies are being used in addition to standard therapy so you would get the standard therapy and you could get additional treatment. The most successful of the companion diagnostics that we see right now are of that type. The future, or what we hope is going to be the future, is that the individual's entire treatment might change based on a variety of individualized sort of markers.
Long: You're listening to Stats and Stories where we always talk about the statistics behind the stories and the stories behind the statistics, and we're focusing this time on the whole issue of how medicine is becoming more personal. I'm Bob Long; our regular panelist with me today, Miami University Statistics department chair John Bailer and our special guest, Bonnie LaFleur: senior manager at Ventana Medical Systems and a woman who has more than fifteen years of experience in statistical and cancer research. We also wanted to find out what the people out there on the street know about our topic, so we asked them, what do you think personalized medicine is?
Man on the street #1: Personal medicine is medicine someone uses to correct something that's wrong with themselves. It's something that's like individually you take this medicine to help something that's wrong with yourself.
Woman on the street #1: Well everyone's bodies are different so everyone's bodies are going to react differently to different medication. So I think a doctor tries to do their best and kind of like gauge it to your body and how you react to certain things.
Woman on the street #2: Obviously different genders have different characteristics and different needs as well as different body types. When you look at how people react to different things, I would say it would be accommodating to those differences.
Man on the street #2: I think that personalized medicine is medicine that's directed in some way toward the specific patient and whatever their needs are obviously.
Woman on the street #3: It's just basically having the choice and like having medicine tailored for you. If you have like two people with the same diseases, you might not take the same approach with them.
Long: Well you can see people have a wide variety of views on what personalized medicine is about, so we'll go to John Bailer for a question relating to that.
Bailer: So Bonnie, you no doubt hear this variety of responses to this and potential misunderstandings, what would you say is one of the most common misunderstandings of what personalized medicine is that you hear in your business?
LaFleur: I think one of the most common misunderstandings is that doctors make the decisions for patients, when in reality physicians generally will provide patients with possible scenarios or possible given a whole host of characteristics of that patient of which these new, novel markers might be just one. And many times it is that decision that's made at the patient level by the patient for the treatment that is most difficult because the physician has to present all the different scenarios to the patient with the amount of risk that might be involved, both with a therapeutic risk as well as the risk of the treatment not working.
Long: I know some of the articles that I've looked at on this whole topic kind of build on what you're just saying, that sometimes a doctor will say to a patient, well you know there's this new medicine out there that's available, but it's sort of in the trial stage, and do you want to do this? That's kind of where we going with it because so much of this is new, correct?
LaFleur: It is and the availability of this is becoming more widespread, but it's still very small segments of the population that might have a higher likelihood of being offered new treatments versus say, rural, clinical patients who don't have access to the larger research institutions might not get the offer of some of these novel, new treatment regimes.
Bailer: You know one thing that I'm hearing in your description is a lot of knowledge and breadth of contact in terms of how you talk about this, so I'm curious about your role as a statistician in personalized medicines and the developments related to it.
LaFleur: So I have a very mathematical and statistical role in that I help examine evidence and I present other researchers and other scientists with the level of evidence that we're measuring, but I also actually have experience with translating that into individual level - so there's a language of statistics and a language of science and then there's the language of risk, which is what individuals need to know, and I think statisticians have a vital role to play to be able to translate our statistical metrics into personal metrics which most patients would see as, "What's my risk of this happening?" And that's something that statistics does and something that statisticians are uniquely educated and able to present.
Bailer: To sort of follow up, you know, when you're communicating risks, there's also the chance that you may be saying, you have this risk of this, but you might be wrong. So how do you communicate some of the uncertainty in some of the predictions and projections of this risk?
LaFleur: That's a great question and we generally don't present our level of risk, or the level of error that we're estimating and that would be a very difficult type of a metric to explain to most patients, so I think that we really don't do that very well at all right now, but one could consider levels of risk, perhaps as a potential metric - to say that we actually do have data that suggest that some risk measures have an interval them, so your risk could be anywhere from 9%-10% of a recurrence if you do not take this chemotherapy treatment. Well a level between five and ten or twenty percent is fairly large and people might have a different idea of what error that they're willing to accept, so that should be part of our conversation when we're describing risk of recurrence or risk of a treatment side effect, we also should be talking about the level of error or the level of certainty we have about that risk estimate.
Long: Something that concerns me about all the drug advertisements that you see on television because you hear so much about, well you could get this or that, but you don't, like you're saying, you really don't know what's the percentage of risk, because that may govern what I'm willing to try to do and like you say, John might be willing to do something that is a 10% risk and I might say I only want 4% or 5% and I guess that's a lot of what you're dealing with because this is all so new.
LaFleur: I agree and I think we are sort of changing, that process has been changing and I think we're much better at that, in defining in our drug pamphlets, when you take a drug, they actually do separate out serious side effects with those that are probably less likely to occur or occur less often, so they're already starting to change, but I think that what we've been doing in the past is the manufacturers assume that risk and they will not present a product until they're sure that those risks are low. And so I think that we can sort of use the FDA and treatment manufacturers as our guide in that, to some extent, but I think it's a great question for patients to ask.
Long: You're listening to Stats and Stories; we're focusing this time on the importance of how medicine is becoming more personal. I'm Bob Long; our regular panelist that is with us today, Miami University Statistics department chair John Bailer, our special guest, Bonnie LaFleur, senior management at Ventana Medical Systems and she has more than fifteen years of experience in statistical and cancer research. Also for our topic today, we had another question for folks on the street and we're asking them how doctors decide what treatment the patient receives.
Woman on the street #4: The doctors look at what has worked in the past and what your complications are. And then they with those two with their wisdom from the past choose whether they're going to give you procedures they've done before or if they want to try a new one that's like new and coming, and they might do that too.
Man on the street #3: I actually don't know how they would go about it. But probably looking at your symptoms and then looking at other people who have gone through the same thing and kind of trying something and then going with it, and if it's not working, moving on to something else.
Woman on the street #5: They definitely look at things of the past. They also look at current and modern treatments and look at how that's been successful or not. They look at the individual person and their type of cancer because sometimes that can be super-individualized. Talk with the patient, and other doctors and other specialists, and colleagues and things like that to find the best way to help them.
Man on the street #4: Doctors will probably go look back at the patients they've had before and see how similar they are to their patient at the time and then make their treatment plan, I guess.
Long: As those responses suggest, a lot of people think doctors do make all the decisions on medication. John Bailer, I'm going to let you follow-up on that.
Bailer: So earlier you mentioned that the patients, ultimately, are deciding what treatments they receive, but the doctors are looking at kind of what the advice is or what's the standard of care, the standard of practice. How do the new ideas that are surfacing with personalized medicine get integrated into those recommendations?
LaFleur: Currently, most of the time it's the physicians that are guiding that practice, so unless there is a targeted therapy, so we need to separate out again those therapies that are specific to only people whose tumors that have certain characteristics versus those markers that might decide whether people are at higher risk for something versus another set of patients. So the risk type of markers are completely at the discretion of physicians. If a marker has been established and accepted into the profession as a potential risk marker, then they are going to decide to use them or not depending on how comfortable they are with the evidence that's been provided on the quality of that risk classification. For drug therapies and for treatment therapies that are based on diagnostic tests, those are much more regulated and a physician is going to be much less likely to use a treatment based on that diagnostic, unless the FDA has said that these two therapies and markers go together.
Long: Cost is another issue and I thought that we could get into that and I'm sure John has some other questions he wants to ask, but I'm just kind of curious because it sounds to me that everybody is so worried about skyrocketing costs, but I could see where this, down the road, probably a ways away, but it could really help reduce costs, do you agree with that?
LaFleur: I do because the way that patients are treated now for cancer is that they get a treatment, it works or it doesn't work, if it doesn't work, they get the next line of treatment and if that doesn't work, then they get yet another line of treatment and at every one of those decision points, there is an amount of risk for those therapies that is very costly; people end up in the hospital, people get very sick. So having the right therapy for these patients ahead of time would mean you wouldn't have to go through this linear sort of a process of trying this, and if that doesn't work, try that. You can go right to the therapy that theoretically is supposed to work for you.
Long: What we're talking about is what we do in standard medicine is more trial and error - this has worked elsewhere; we're going to try this. What you're saying is that if you have this individualized medicine you can avoid, possibly, a lot of those kinds of things, so therefore, the cost would go down.
LaFleur: Yeah, I believe that, and there's also a reduced risk for the patient for having all of the potential side effects that go along with it. Because part of those therapies also are some patients shouldn't get certain therapies and so that's also a very important potential marker that we can give patients, that these patients should probably stay away certain therapies that might be more toxic to them.
Long: John Bailer.
Bailer: Can you give examples of some success stories when you think about the use of this, I know this is fairly new work and a new world of medical practice, but are there particular examples that you think are worth highlighting?
LaFleur: Well the most commonly cited positive case study is the use of Herceptin in breast cancer. So women who have breast cancer that expresses a protein, HER2, receive Herceptin in addition to their standard chemotherapy and it's actually reduced the rate of recurrence up to 35% in those women.
Bailer: That's outstanding. It sounds like there's a lot of development and testing that goes along with this you know, from the point at which something is identified as a candidate until it would be accepted in practice, what kind of time frame are we talking about?
LaFleur: For treatment sort of biomarkers, we're probably talking 20 years. For screening biomarkers, where you're hoping to catch people who might manifest disease without actually having the disease yet, it's much, much longer and the level of testing is much different so it sort of depends on what type of markers we're talking about and how long it takes to actually manifest the disease.
Long: We've talked a lot about for example, breast cancer, are there some other areas, I'm wondering for example, heart disease, things like that where you're also doing the research on that in today's world too.
LaFleur: Yes, the use of statins and high risk for coronary heart disease has also been in the press a lot lately; who is really at risk for coronary heart disease, who's most likely to benefit from statin use, whose potentially risks of downstream or down adverse events might suggest that maybe they shouldn't be taking statin medication?
Bailer: What kind of changes in research and in understanding of biology has impacted this world? I mean in your practice in statistics, in biostatistics, what kind of new things have you had to come to understand as a consequence of these changes?
LaFleur: Well I think one of the things that the new personalized medicine has focused on is the real importance of understanding biology. Biology, medicine and quantifying all of those things, so if I could point to one thing that's really evolved over the course of my career, it's the interdisciplinary teams, or the cross-functional teams that it takes to actually go through the scientific process has just expanded immensely. The idea that a biologist can sit in a lab and discover a new process that they think someday might be useful is no longer. It now is a fairly large group of people who together decide which biologic processes are most important and which ones can then be pushed further and faster to get out to the population that it needs to get out to, and I think that that's something that's been evolving over the past ten or twenty years.
Long: I think John was kind of asking about this earlier, but it seems like the timeframe issue, especially when it comes to government regulation, things like that, that's one of the biggest obstacles, isn't it sometimes because you come up with something that you think is going to be really good, but it may take a long time before it can actually get to the people who need the help.
LaFleur: Well I think the regulation; personally, my experience has been that the FDA actually is very onboard with this. They have a definition for personalized health care, they have put out pamphlets on it, they understand its importance and I think that they're actually being pretty responsive to science. I think what's taking the longest amount of time is just the level of evidence that's required for individual physicians and patients and for everybody to adopt it because everybody hears about these things in the news with either a large amount of skepticism or a large amount of hope that I think that just the level of evidence is just increasing, the requirements.
Bailer: It's one thing, as you say, to start with a proposal and it's another thing to validate this candidate as something that's going to be useful.
LaFleur: And I think standardization, a lot of this has to do with the technology, it's driven by technology. We have the ability to measure things in our body that we were never able to measure 20 years ago. Well that technology comes at a price and that price sometimes is precision and you don't know about that price until after you've done a very lengthy amount of studying and testing.
Long: Something we always ask on this show, because you've mentioned the media aspect, how's that factored in? Has it been a positive thing in terms of that we're kind of telling people out there that there's this new treatment that's being developed, how would you asses how that is being done in the media today, has it been positive, or negative, or what?
LaFleur: Well as with most things, I think that it has been complicated by the fact that there's a certain level of knowledge that's expected, what's happening is people are reading scientific studies, they're then synthesizing and broadly exposing the general population to them and it's not always the case that it's synthesized in the way that it was intended to be synthesized by the scientist, so there's a translational piece that's actually, could potentially be adding some miscommunication to the process, sometimes they do it well, sometimes they don't. But it certainly is the case that it's very difficult for the person who does not come from a scientific background to actually understand some of the information and so that translation is really, really, really important.
Long: John Bailer, we've got time for one final question here today.
Bailer: Thanks Bob, one thing we often ask guests here is what they would like the public to know more about. So if there was an idea that you think it's important for the public to understand about some statistical concepts, whether that's uncertainty or variability or errors of prediction, what kind of ideas would you hope that the public would have better understanding of to help them as they make decisions about personalized medical care in their futures?
LaFleur: Well I think that the variation is, obviously, at the core of most statisticians' livelihood. I'm characterizing variability as what we do, that's what we're taught to do, it's probably one of the more fundamental things that we do. I think having a public understanding about that means to them is really important. I think that there is no certainty in this world, there's always uncertainty, and hopefully you'd like to be able to feel fairly confident that the decision that you make is one where the uncertainty is in your favor, but that isn't always going to be the case and so I think that as statisticians recognize that people making decisions is difficult because we're presenting them with evidence that they're using to make the decisions and we actually recognize that all of these decisions that we're allowing people to make are going to error on one side or the other.
Long: Bonnie LaFleur, we want to thank you very much for sharing your insights on the whole idea of personalized medicine on our Stats and Stories show today.
LaFleur: Thank you, Bob.
Long: If you'd like to share your thoughts on our program, you can send us an email at email@example.com. Be sure to listen for further editions of Stats and Stories where once again, we'll talk about the statistics behind the stories and the stories behind the statistics.