‘Decoding DNA: The Future of DNA Sequencing’: Dr. Lee Hood Appears on Australian Radio Show
Dr. Lee Hood was interviewed by Dr. Norman Swan, of the Australian Broadcasting Corporation, for the "Health Report" show. The interview aired on radio on Feb. 24, 2014.
Listen to the radio segment.
(Editor's note: ABC identifies Institute for Systems Biology as part of the University of Washington in the audio and in the transcript. ISB is an independent nonprofit organization and NOT a part of the UW.)
Norman Swan: Welcome to the Health Report with me, Norman Swan. Today, using the private sector to get the most out of our genes, a blood test for posttraumatic stress disorder, new findings in brain cancer and a new concept, P4 medicine—the way of the future as envisaged by Leroy Hood. He’s one of the people who made mapping and sequencing the human genome possible by developing the first automated gene sequencing machines nearly 30 years ago.
Lee Hood is President and Founder of the Institute for Systems Biology, Seattle. Systems biology tries to analyse as much information about ourselves as possible, from our genes to substances in our blood stream to our environment. Lee Hood’s latest project is a huge study starting this week using systems biology to follow the health and well-being of large numbers of people for many years. I spoke to him last week at the American Association for the Advancement of Science congress in Chicago.
Leroy Hood: We’re going to start a longitudinal study of 100,000 well patients over a period of at least 20 to 30 years. We’ll be able to get two interesting kinds of data; with 100,000 well people we’ll be able to follow a set of patients that remain well or even get weller, but from 100,000 we’ll see wellness to disease transitions for all the major diseases, and because we’re creating data at multiple time points we’ll have data on the very earliest points of transition from wellness into disease, and the idea, of course, is to get very early diagnostics and understand the earliest mechanisms of disease so we can change the disease trajectory immediately back to the wellness trajectory.
Norman Swan: So what are you measuring?
Leroy Hood: We’re going to determine the entire genome sequence. We think there are at least 300 actionable variants, so if you have one of them you’ll want to know about it and what you can do to improve your health.
Norman Swan: And when you say an actionable variant, there’s lots of genetic differences between us but there are not many yet that you know something that you can do about it.
Leroy Hood: There are about 300 variants where you really can do something about them. For example, a friend of mine from Microsoft got early onset osteoporosis at about 35, and that’s a disastrous diagnosis for a young man. Had a SNP analysis done, found he had a defect in a calcium channel transporter. He took 20 times the normal amount of calcium for the next year and a half and he reverted his bone density back to normal again. So that’s an example of a variant and an action you can take to improve your health.
Norman Swan: So you’re going to do their complete gene sequence?
Leroy Hood: We’ll do the complete gene sequence.
Norman Swan: And what else are you going to measure?
Leroy Hood: We’re going to measure the gut microbiome.
Norman Swan: So the bugs inside you?
Leroy Hood: The bugs inside you. We’re going to measure heart, respiration, weight, blood pressure, sleep quality, sleep duration, calories used, things like that. We’re going to measure clinical chemistries; it’ll be very focused toward nutrition. We’ve developed a series of organ-specific blood protein fingerprints that will allow us to follow disease transitions from wellness to disease in the brain, in the heart and in the liver. And what we will do then is integrate all of these types of data together. The data alone have actionable possibilities, but when you do data integrations you can increase the number of actionable possibilities, and my contention is, in the first project which will start March first where we’re going to have 100 volunteers doing this, my contention is each of them will have multiple actionable opportunities that they can determine whether they’d like to improve their health or avoid disease.
Norman Swan: So, a couple of things about this approach, which is this massive data approach to the health and illness which is on our doorstep—in fact, you would argue it’s here now—is many of the things that will happen to these people will not be because of what they were born with or because…well, partly because they were born with but what they are exposed to in their general lives, and that’s very much more complicated than even just studying the genes and the bugs in your bowel. How are you going to know what’s happened to them in their lives and what’s made a difference to sll measures that you are making?
Leroy Hood: So a sixth type of data that we’re going to get is kind of a personal record of any major changes in life—depression, nutrition, exercise, changing drugs, all of those kind of things—and that’s why the model is going to be assessing the two major types of biological information that exist in humans—one the genome and two the environment.
Norman Swan: And how are you going to know it’s scientifically valid, because in medicine the gold standard is the randomised control trial? In other words, you don’t just do it to somebody, you have a control group and everything else is equal so that you know that doing all this measurement and then making an intervention, it’s not going to be cheap, it’s actually worth the money and the effort?
Leroy Hood: See, I would argue that the randomised control trial is wrong in its fundamental features, because it takes, for example, 30,000 patients, give some drugs, some placebo, but what it fails to do is recognise that every single one of those 30,000 patients are different in two dimensions. One, they’re different genetically, and two, they have different environments. What P4 medicine—predictive, preventive, personalised and participatory—does is to analyse each individual patient and after we’ve done that we can then aggregate patients that behave in similar responses to drugs, whatever you like. And we’re basing this medicine on a unique analysis of genetics and environment for each individual. So I’d argue, in the future, we’re going to have a lot of n=1 studies which when aggregated are going to tell us about many aspects of disease that could never be done with clinical trials because there aren’t enough patients out there to do it.
Norman Swan: There’s too much for an average. And when you say n=1 it means a trial comparing…
Leroy Hood: One individual.
Norman Swan: Comparing it to yourself.
Leroy Hood: So suppose that you were in this study that we’re starting in March, and we looked at your genome and we found that there were four variants that made it very likely that you were deficient in certain key neurohormones. We could then give you intermediary metabolites above the block that would replace those things, and we could do an n of 1 experiment on you and see did you lose your depression, did you get more energy and so forth. So, n of 1 experiments, I think, with proper data controls and the ability to do clever aggregations are going to be the wave of the future.
Norman Swan: Given that you’re in a sense experimenting with each individual here, you know, so you’re giving intermediary hormones to boost your hormone levels, how do you guide against harm? Because one of the good things of randomised trials is they…it takes a while to get to the definitive trial because you test safety first. It might not be safe to give the intermediary hormones because it’s not kind of a natural thing to do. How are you going to assess safety?
Leroy Hood: I think what you will always do is start with very low doses and see if they make a difference or not, and frankly it’s what you do with patients in big trials too. And you can escalate to the place where you’ve optimised whatever it is. So I think you just have to be sensible and careful. I think for a lot of the actual variants in the initial phase they’ll be nutritional, and I think there’ll be almost no risk modifying nutritional environment exposure and then assessing how people feel.
Norman Swan: So, with this P4 medicine, in many ways the new element is participatory, implying that the person whose genome it is participates in this process and kind of owns it. Talk to me a little bit about that.
Leroy Hood: Participatory is one of the fundamental driving forces of this new type of medicine, and it really has a lot of different aspects. One, you are going to have your individual data cloud, and that’s going to give us the opportunity to say, here are actionable possibilities. Do you want to explore these in the context of what we know? I think a second idea is, can we arrange to generate a means for taking these individual data clouds and aggregating them after anonymisation so we can mine for the predictive medicine of the future. That’s really going to be key, and I think the third thing on the participatory side is, how do we educate the patients? How do we educate the physicians? How do we educate the medical community? And the key driving force there are going to be patient-activated social networks. That’s going to be how they learn; they are going to be the driving force that is going to demand this P4 medicine.
Norman Swan: Now, what legislation needs to be in place to protect this? Because the problem in Australia is that there’s no protection against genetic information. In other words, if you go and get your genome done and they come up with actionable mutations, you’re obliged to tell your life insurance company. You’re not protected, whereas in the United States you do have some protection.
Leroy Hood: We do have two types of protection. One is the law called GINA, and that’s really focused at employers and insurance companies, and it basically says you can’t use genetic information to discriminate. And of course, ObamaCare is very explicit in saying neither employers or insurance companies can discriminate, so we have laws in place that are pretty good, and I’m not saying they can’t be made better, but we’ve taken a major step forward.
Norman Swan: So for this kind of medicine to go into a country like Australia we need those laws in place, really, or a version of them.
Leroy Hood: I think you need to be able to protect patients against discrimination. Now, people argue about whether anonymisation does that, and it is true. Theoretically you can take genome information and adduce the individual, but you need a lot of data to do that, and most people aren’t going to be doing that.
Norman Swan: Without wanting to rain on your party, whenever new technology historically has come into healthcare, whether it be in the United States or in Australia, it’s expensive. It’s been promised to be cheaper, but it’s expensive. So we spend a lot of money on medical research, we find these genes, we develop targeted drugs, and the drugs cost $100,000 a year or more per patient. Is your vision going to be affordable in a modern healthcare system?
Leroy Hood: So, let’s disambiguate your question. One aspect of it is about drugs, one aspect of it is about technology, and they’re really two different things. So, about technology I would say this: I think the classic example is, we developed in 1985 the automated DNA sequencer. The sequencing machines today are somewhere like $100,000 less expensive to actually determine the sequence. My prediction is in a five to eight year period with third-generation sequencing, single molecule sequencing, the human genome will be $100. And I would argue, most of the assays I talked about in this 100 pioneer project are on Moore’s law decline. So I would argue in five years what now costs $100,000 per patient might cost two or three per cent that amount. So, I think you have to take into account how technology is changing. And we’ve got concrete examples of where it really has changed and become less expensive. And again, I think most of the important technologies are in this Moore’s law decline. But the second question is…
Norman Swan: Are the interventions which arise from the genome studies?
Leroy Hood: Yes. The second question is, drugs are really expensive, and I think drugs are really expensive because drug companies are really great at making drugs. They’re terrible at choosing the drug targets. What systems biology is going to give us are powerful new ways to identify drug candidates really inexpensively, and if we blend that together with drug companies’ effectiveness at making drugs, I would argue we could bring the cost of drugs down by an order of magnitude or more because we can get rid of most of the drugs that fail.
Norman Swan: Gene sequencing and systems biology pioneer Leroy Hood, and you’re listening to the Health Report here on RN with me, Norman Swan. Lee Hood hasn’t stopped working, though, on very specific health problems, one of which is one of the wickedest forms of cancer, malignant brain tumours known as glioblastomas.
Leroy Hood: We’ve collaborated with Terry Van Dyke at NCI who’s constructed four different mouse models for glioblastoma that very accurately replicate the four grades of human glioblastoma. But what she’s also done that’s really clever is with each of these gene manipulations she’s put an activator on it. So nothing happens in the mice with these genes until they get through to adulthood and then you can turn on the cancer process any time you want, and you can follow from day one exactly how the disease-perturbed networks in the brain are changing. So we’ve analysed that information and the results have been absolutely spectacular. One, they explain virtually every aspect we know about the pathophysiology of the disease, but two, we can take the eight classical landmarks for cancer and we can map them in a time series as to just when each of these comes into being in this cancerous process. So…
Norman Swan: So what you’re talking about is the cancer cascade, the sequence of genes that have to mutate?
Leroy Hood: That’s right. It’s the sequence of networks that have to be disease-perturbed that encode all of these different aspects of cancer and everything.
Norman Swan: And any intervention points that would be more effective than current treatment?
Leroy Hood: So, we can make hypotheses about the intervention points and the really important point about that is most pharma are really reluctant to get into glioblastoma, because it’s a small cancer, and small in terms of numbers of patients. But if we can give them great drug candidates we may be able to persuade some to take on the task of generating the drugs to see if we can successfully intervene.
Norman Swan: You mentioned earlier, in fact, that one of the basic premises of your P4 medicine and this 100,000 person study that you’re proposing is the…is an emphasis on wellness rather than disease, and also on prevention and early intervention, and you’ve walked the talk on this in terms of what are called biomarkers, in other words, much simpler tests to detect disease and follow it through, and one is lung cancer. Tell me what you’ve done with lung cancer.
Leroy Hood: With lung cancer, about five years ago I created a company called Integrated Diagnostics, and we transferred to them the systems approach to how we can identify in the blood biomarkers that truly reflect the disease. And let me just say, when you look at the bloods of normal and the bloods of people that have lung cancer you see an enormous number of protein differences. Most of the differences are noise that don’t accurately reflect the disease. Systems biology lets you pick out the signal. So what we’ve been able to do is ask the following question: in lung cancer in the US there are 3 million lung nodules a year, 600,000 people get surgical procedures to determine whether their lung nodule is normal or neoplastic, and 70 or 80 per cent of those 600,000 carry benign nodules. An enormous waste of money and morbidity.
Norman Swan: I should just explain just a bit of background for the listener, is that there’s almost an epidemic of CT scans of the lung, for various reasons, and by accident they find these little lumps and then everybody goes into a panic because they don’t know what to do with them, and this has generated huge expense and concern, and what you’re talking about is can we actually know without doing too much whether or not this is a nodule we need to be worried about it.
Leroy Hood: So what we’ve done is set up a procedure that now gives us a 13-protein biomarker panel that allows us to unerringly rule out more than 70 per cent of the normal nodules and distinguish them from their cancerous counterparts. And the really interesting economic calculation is just by doing that we’re going to save the American healthcare system north of $3.5 billion.
Norman Swan: So in other words, a translation of what you’ve just said is, when you diagnose a nodule as normal, you can be really sure it’s normal, but there’s 30 per cent that you’re not sure about, so you’ve at least cut down the biopsy rate by 70 per cent.
Leroy Hood: That’s correct.
Norman Swan: And this is mind-blowing, you’ve managed to come up with a biomarker for posttraumatic stress disorder?
Leroy Hood: We have. It’s a very preliminary analysis; we looked at 43 soldiers from Afghanistan that were normal by all criteria and 42 that had extreme forms of PTSD and did not have traumatic brain injuries, so it was a pure population. And we did the blood protein and other biomarker analysis to create a panel that gives us roughly 95 per cent sensitivity and specificity, and this has never happened before on any neuropsychological disease, and the reason this panel is important once we validate it is, one, this allows us to distinguish normal from PTSD soldiers, and previously that was all psychological and pretty fuzzy psychological. Number two, we can do very early diagnosis of PTSD right after they come off the battlefield, say. Number three, we can follow the progression of the disease beautifully. Number four, we can follow the response to therapy, and number five, we believe that actually we’ll need to stratify patients into different subgroups, because genetic constitutions may predispose patients to be extremely susceptible or more resistant. So the biomarkers will be able to do all of these things, and the final point I would make, which is really critical, is we believe we can apply this approach to all of the major neuropsychiatric diseases for which there are no quantitative measures today—schizophrenia, bi-polar disease, autism—and achieve the same kind of results.
Norman Swan: And the biomarker is what? Is it a stress hormone, or is it a set of genes? What are you actually?
Leroy Hood: It’s… The markers that we’re measuring for PTSD are of two types: one is proteins, and the proteins really relate to the parts of the brain that are affected by PTSD, and number two it’s a type of nucleic acid called microRNAs that turn out to play a really important regulatory role in disease, but interestingly enough get secreted into the blood and they’re protected in the blood by coats of proteins that shield them from enzymes that could cleave them up and everything. So they turn out to be quite a good blood biomarker just as the proteins are a good biomarker.
Norman Swan: So what are you going to say to the Department of Defense when they come to you and they say, ‘Well you, Dr Hood, you’ve managed to find the genetic profile of people who are highly susceptible to PTSD. We want to use that to screen people out.’ So, if they want to join the Marines, they can’t join if they’ve got this thing.
Leroy Hood: So, I’ve already talked to the Department of Defense at some length about this, and I’ll tell you, the Department of Defense isn’t sure at all it wants to know about the stratification of patients for the reasons you’re implying in your question. This kind of assay would really be great for rangers that undergo this special training that’s a couple of million dollars per soldier. If you could tell them who’s going to wash out before they ever get started—and quite a significant number wash out now—so you’ve wasted one or two million dollars on these guys that can’t cut it. Suppose you can do that all front-end. They’re not even sure they want to do that, and that’s their choice. Frankly I think they and the soldiers themselves would be foolish not to have these studies. If I were going to undergo that and I had something genetically that wasn’t going to let me do it, I don’t think frankly I’d want to go through the humiliation of being a failure. Now, that’s a rational point of view, and I recognise that in ethics rationality doesn’t necessarily hold sway always.