Issue link: https://tmcpulse.uberflip.com/i/375565
t m c » p u l s e | s e p t e m b e r 2 0 1 4 14 to you. It will be a standard of practice. That's a reason- able prediction. Q | How much of what you're doing is starting to affect decisions on the therapeutic side? A | Well if you're asking about what we can do today, it really does depend on your definition of intervention. It is growing, but we have some way to go. For exam- ple, we can ask today, if you are a cancer patient and you have your tumor genome sequenced, how often would that data impact your care? That's an important question. The answer ranges from five percent to 25 percent. In five percent of the cases, you can find an answer that is truly directive for your therapy. This is taxol, or herceptin or other drugs that are targeted to specific genetic changes. There's a clear direct link there between what you'll find in the cancer profile versus what can be chosen for therapy. Now, to go from five percent to 20 percent, the definition of intervention needs to be 'useful information.' Then that includes a patient being steered towards a new trial or towards another drug that may not have been in the frontline before. Then there may be a subtle shift in the therapeu- tic regimen, so I think we've got ourselves in a bit of a corner if we paint the picture that this is truly directive for all patients at this time. If you are in the group that is helped, this is already very important. In the next five to ten years, more and more will be in that category. Q | Can you tell us about the team you have built within the Human Genome Sequencing Center? A | We have an accumulation of expertise. If you look back a decade, DNA sequencing was an art form. Now, even though many aspects of the process are 'plug and play,' the process is still challenging. There are parts of the genome that are very hard to understand and physically difficult to decipher. And it turns out that there's actually enrichment of those hard-to-decipher regions that cause disease, so the challenge is even harder than it might seem at first. So simply what you get from some of these new machines with the stan- dard methods analysis often needs to be improved by human experts. Next, there is the general question of data analysis. Even if you have high quality data, how do you interpret it? That's a whole industry on it's own. There are com- plex challenges including those that result from the databases of reference information not being perfect. Q | If a patient came to you with a specific health issue of concern, what are the various techniques that could be employed? A | So the corner stone of this whole branch of science and medicine is genetics and genetic determinism. When you ask 'when is something genetic' and 'what is the evidence that something is genetic,' the best thing to do is to look and see if it runs in families. When you have an aggregation of a disorder with very similar pathology within individuals in a family, the immedi- ate logical jump is that it likely has a genetic compo- nent. Now in some cases, that genetic component is very easy to track. If the disorder is very clear and its pattern of segregation in the family is unambiguous, then that is a simple genetic problem often caused by a single gene, and a single letter change in the genome. You can go work with a family like that, track the gene for a disease successfully. Other disorders just have a loose tendency to run in families. Examples are many of the behavioral disor- ders and cancers. We know there are genetic contrib- utors and we know that understanding those will help us understand the mechanism of the disease. But it is a much more difficult problem to track down the genes for these common conditions. This is the big challenge in genetics and we are making good progress. One of the things we discovered in the last five years is how much natural variation there is between each of us. We have known for a long time that if we sequence any new person, there would be about three million differences between them and the reference database. But we've learned more recently that even in the gene regions—the one-percent that really mat- ters—there are hundreds of individual DNA changes that we will not likely see in anybody else on the planet. So we're very different genetically from each other. Now that's thinking just in terms of population, genetics and the structure of life. But it is also a big practical problem when you try to figure out these genetics stories. For example we can examine different families with multiple siblings, some of whom have heart disorders, to find genes that cause adult heart problems. But when we do that, there are going to be a lot of parts of the DNA sequence that are unique to individual fami- lies, but that are not related to the disorder. So we have an issue of scale. We can't learn things from three fam- ilies, but perhaps we can learn something from 3,000 families. So if we want to solve these genetic problems and we want to apply the genetics tools we have devel- oped, we really need to work with more families. There is a big sea change in the ability to scale these activities. Historically, research and clinical care have been two activities that are well separated. For example, if you told your clinician about a fam- ily history of heart disease, I'm sure that you'd get that acknowledged but it would not have led to you becoming a research subject. If you heard about a heart disease study you might volunteer as a research participant, but not through your regular physician. Previously, the research investigator would have looked at enough families who volunteered in order to make a discovery about a particular gene. Then the researcher would declare that gene important in these families, and perhaps develop a test for and then tell the physician 'You should test for this gene in families like that.' That's the current cycle that we have. What we envision in the future when the act of sequencing entire genomes is more routine for all sorts of reasons, is that we can generate the data to complete that kind of research activity in a much faster and more direct way. So it would be as simple as an investigator could come to the clinical databases that have DNA sequence information, and with the right IRB approvals, and with consent, they could mine that data and say 'Hey look at this. In all the families with that history of cardiac problems we find these genes have these mutations every time.' Hence, the discovery would be catalyzed by better collection of clinical data together with DNA sequence. It will be a faster and more efficient cycle. Q | You've played a really key role in leading the new TMC Genomics Institute design team. What would be your vision for a world-class clinical genomics program? A | Consider some of the things I said earlier about how genomics has changed research. One of the changes is the definition of deliverables in research. The genome project showed that quantifiable deliv- erables can be a part of a dynamic and active and flexible research program. So in the case of the human genome, we said 'We're going to determine this many letters' and we then did that. This basic concept has become a quality of many different projects in biology. Now we can say that with new patients in the medical center, we quantitate the number that we can advan- tage by providing their genome sequence data, in their medical record. We can be specific and quantitative about the classes of patients, the needs of different groups of patients and the speed with which we can deliver genomic data to them as an adjunct to their current care. Q | So for families in the future, when sequencing may be as common as getting an X-ray, where do we begin to set that new standard of care? A | So it's all about risk and familiarization. I think the clearest place to start is in reproductive health and in the administration of carrier testing for couples plan- ning to have children. Clearly you want to be aware of the possibility of some inborn error that is inherited. That's the easiest scenario to conceptualize. But the impact of these technologies actually goes on from there. For example, you can monitor the fetal genome early in development and that can impact care as well. There are really three main fronts in this area. One is these very early predictors to improve child health. The second one is in cancer prediction. Of course, many of us carry cancer predisposition genes and the story of the breast cancer gene is the most dramatic because you have the clear impact of the locus of the gene and the clear clinical follow up that you need if you are at high risk. But there are many other genes that impact cancer and should be considered. The third category of risk is where we're lagging behind the most. You are an essentially healthy adult, you've had all of your cancer screens and you don't have cancer in your family, why would you want your genome sequenced? What we tell healthy people right now, with the current state of knowledge, is that your need to have your genome sequenced today is minimal. But the fact is that there are many adults who are not healthy—or who have family risk they have not consid- ered—or soon will have disease. We can't sequence everybody today. Even at $100 a person it's still too much of a burden. But as we see these methods grow to a greater efficiency and an even greater interpretability, and as we build pro- grams that are based upon new families and upon