 Good morning and welcome. There's still a few seats left in the back, so please help yourself. My name is Mark Pigram. I'm the newest kid on the block, arguably, on the faculty. I joined Stanford about three months ago now in the oncology division, and I'm the new director of the breast oncology program, so it's really a treat to be able to meet with you this morning and give you a little bit of an update. I have kind of a hard act to follow after Ron Johnson, and I'm not quite in the same vein as him in terms of speaking style. I asked my wife this morning, I said, I'm not sure what I should wear. Maybe I should wear a black turtleneck and blue jeans, and she said, you don't have a black turtleneck. Then I said, well, I should probably wear a hoodie. And she said, well, you have a hoodie, but it says UCLA on it. Maybe you better not wear that, because I was on the faculty at UCLA for many years back in the day. So we're here to talk about breast cancer, and I know not all of you have an oncology background, so I'm going to keep this fairly basic, and then we'll have time for Q&A and discussion at the end. But we have a small group here, so if you have anything that's on your mind during the presentation, I have no problem at all for you to chime in. This is being taped, and so we will pass around a mic for you, so just wait for the mic if you want to raise a point anywhere along the way. Please feel free. This is entirely informal. So I'm going to talk about some genomic applications that have been applied recently to breast cancer in terms of new molecular diagnostics and how those have led to new therapeutic opportunities and what the future might hold for new diagnostics and therapies in breast cancer based on a fundamental understanding of the molecular biology of breast cancer. So as I hope most of you know, cancer is a disease of the genes. It's a disease of acquired genetic mutations. Now some of these mutations can be inherited in a familial fashion, so there are some family cancer syndromes. You all are probably aware of familial breast cancers and ovarian cancers, etc. So sometimes you can be unlucky to inherit a genetic predisposition, but most of the time the mutations that occur that give rise to the most common types of breast cancer are acquired mutations that happen during your lifetime, probably as a result of exposure to environmental carcinogens, to ionizing radiation in the background, whatever. But these are acquired events that happen during normal cell division. And so this is a so-called chromosome painting experiment. The technical term is a spectral karyotype. These are all the normal human chromosomes that all of us have in every single cell in our body. And the chromosomes are where the DNA reside, and that is the instructions given out to all cells, programming those cells into some physiologic function or another that they must execute for us all to stay healthy. So you can see that there are pairs of chromosomes, they're highly organized, very tidy indeed, very efficient packaging for all of the genomic information. But what happens in cancer? In cancer the genome is highly disorganized as a result of all these mutations. You can see that in some cases some chromosomes are missing altogether. In other instances there are too many copies of the chromosomes due to duplication of whole chromosomes. In many cases there are rearrangements of the chromosomes where the chromosomes simply are scrambled one with another, and this puts genes into new environments, into contexts that they are not normally supposed to reside, and that can cause the expression of those genes to go crazy. And this can lead to abnormal cell growth and give rise to a tumor. So highly disorganized, highly mutated genomes are the rule in breast cancers as well as other tumor types indeed. So it's this accumulation of genetic alterations that give rise to cancers. At first maybe there are only a few mutations that occur and that can give rise to areas of abnormal tissue called hyperplasia or dysplasia. These are just abnormal growths, abnormal collections of cells, but they're not full cancers yet. These are just pre-malignant conditions. And then you can actually get to a situation where the cells in these collections can sometimes resemble a cancer and they're actually called a carcinoma because they look so much like tumor cells under the microscope, but they don't have the ability to invade. So this is a pre-cancerous condition. In breast cancer this is called ductal carcinoma in situ. And these are actually readily cured, usually with surgery and or surgery plus radiation alone. Just a simple lumpectomy often is all that's needed to cure this condition because it's incapable of spreading. It doesn't invade into the normal surrounding tissues yet. And with the advent of imaging tools, this is the ideal. If we could catch lesions early, then we could prevent them from evolving into a full-blown cancer. And with new technologies of detection, more and more we're finding smaller and smaller lesions, many of which are now in the pre-malignant stages. One of the pluses of mammography, for example, is a stage migration to these earlier stages, which are much more amenable to successful treatments and far superior outcomes long term. So this is one of the advantages of early detection. Now with enough genetic alterations, you can reach a stage where these tumor cells acquire a capacity to invade into normal adjacent tissues and they can gain access to vascular channels. They can actually break into the bloodstream where they can move around to other parts of the body and set up shop and other vital organs. And that's what at the end of the day really causes the damage in cancers is its ability to spread. It's called metastasis. And that's the deadly form of the disease, actually. These cancer cells can also gain access to the lymph channels as well as the bloodstream. So both types of channels can be invaded by the growing and invading tumor. And so breast cancer has a particular propensity to spread to the lymph glands and that's why oftentimes we'll sample the lymph nodes nearest the breast to see whether tumors have picked up that ability to spread as well as grow locally. So there are hallmarks of a cancer phenotype that occur along the way as a result of this accumulation of more and more genetic mutations. And one of the things that happens early on is an ability to proliferate more rapidly than normal. That's a hallmark of cancer. Normally cells grow, they divide, they perform some natural physiologic function, and then they senesce and they die off and then they're replaced by new daughter dividing cells in a balanced situation. That's how it's supposed to happen. But cancers can pick up a phenotype where they forget to senesce and die. The term is called programmed cell death. That's a normal physiologic function that all cells are supposed to go through. But cancers can evade programmed cell death. The Greek term for that is apoptosis. I think it's great, I didn't study Greeks, I don't know. But there's also a phenotype of insensitivity to normal anti-growth signals that keep cells in check and keep them from accumulating aberrantly. So cancers can pick up this phenotype. Cancers can actually recruit new blood vessels because for a tumor to grow beyond about two millimeters, it has to acquire a brand new blood supply to deliver oxygen and nutrients. So really, you must think of cancer really as a true organ. It has lots of normal cells mixed in with it. Blood vessel cells, blood cells, immune cells, support cells like fibroblasts. It's a collection of a variety of cell types. It's not just the cancer. A tumor is very complex, very heterogeneous. It's an intermingling of many normal cell types mixed in with the abnormal cells altogether, acting truly as an organ with its own blood supply. Then there's this limitless potential for cell division where these cells live beyond their normal lifespan and they have limitless potential for replication. Stem cells are like this. Then tissue invasion and metastasis is the hallmark of cancer. So these hallmarks have been cataloged in the literature and in textbooks. These half dozen or so phenotypes are the clinical hallmarks of all human cancers. If you think about it, each of these processes is an opportunity for development of new diagnostics, also an opportunity for development of new therapeutics. For example, if we can block the ability of tumors to recruit a new blood supply, then we can have successful anti-angiogenic therapies, and that in fact is the case. We do have drugs that are used for other cancers that do just that. There are drugs that block growth factor signaling. I'll show you an example of that in just a minute. There are drugs that can block invasion and metastasis, for example. Each of these is really a paradigm for diagnosis as well as treatment in the future. Next, I'd like to talk about some technology that was pioneered here at Stanford and is now in all the textbooks in breast cancer. You can take a tumor in the laboratory that's been frozen in liquid nitrogen usually, and you can grind up that tumor and you can extract the DNA and the RNA out of that tumor and then analyze it. And in this fashion, you can measure which genes are turned on and which genes are turned off in the whole genome in a single experiment using a gene chip that has a representation for each gene on the chip. And by assigning different colors for relative abundance of the expression of genes, you can actually get a molecular portrait of all the genes that are turned on and all the genes that are turned off in a particular patient's tumor. Now, for an analysis of a particular spot on this array, remember I told you that tumors are highly heterogeneous. They're admixtures of normal cells mixed with tumor cells, and so there can be dilutional artifacts because of all the normal cells in there that will have normal expression of genes. And so a particular spot on this array may not be incredibly accurate because of this dilutional artifact, but it's really not the specific gene that is important for this technology. It's the overall pattern that's important. So as long as we capture the pattern faithfully, we can do so whether we have precision on any one gene or not, and that's still okay. Let me show you an example for that. Here's an example where the pixels are not truly faithful. So each one of these pixels is not particularly accurate. Nevertheless, we can still get a glimpse of what the overall picture is, and this is me enjoying the Grand Canyon on my summer vacation. So this gene chip microwave technology can do the same thing by looking at expression of gene patterns throughout the whole genome in a single experiment, and you can do this on each individual patient's tumor, readily, in the laboratory. And so in published data from laboratories here at Stanford, we had for the first time the molecular portrait of human breast cancer. And this was incredibly revealing because it revealed that there are approximately five intrinsic subtypes of breast cancer. This is not just one disease. Until this time, we had just sort of treated all breast cancers the same, because under the microscope and in the clinic and on the mammograms, they resemble each other and they start in the breast, so we call them breast cancer. But we knew, based on history, that certain tumors behave differently from one individual to the next. Some patients relapse, some patients are cured. Why is that? Well, it turns out that there are intrinsic molecular subtypes that were identified here at Stanford that can distinguish one type of breast cancer from another. For example, these types, called Lumil-A, are almost always estrogen receptor positive. In fact, this first branch point in this analysis is basically taking all the ER estrogen receptor positive genes and putting them next to each other because they resemble each other, and then all the estrogen receptor negative tumors actually cluster together when you analyze these expression patterns. And so these Lumil-A tumors typically will respond to anti-estrogen treatments like tamoxifen, for example. This group of tumors here in pink have amplification of an oncogene called HER2, human epidermal growth factor receptor number two. And there is a treatment for that called Herceptin that can vastly improve the outcome of patients with this subtype. And now we're working in the laboratory to identify new therapeutic targets in these other intrinsic subtypes. And as I mentioned, these subtypes did correlate with long-term clinical outcome, with some subtypes having a particularly poor prognosis. This is the relapse over time, and lower is worse. Here are the Lumil-A's. They tend to fare better in terms of molecular subtypes. So these phenotypes are biologically and clinically relevant because they can track prognosis and they can suggest and inform therapeutic decisions because of the targets that are known to be expressed in some of these subtypes. And it provides an opportunity to discover new targets in these unmet needs for other tumor types for which we do not yet have targeted therapeutics. And so that's a very intense focus for our research now at Stanford. Now I told you that these chips, these spots on the chips are not particularly faithful or accurate in terms of their gene expression on an individual experimental run. But there are other technologies that can actually make it incredibly precise and actually quantitative. And it turns out that you probably don't need to measure all 20,000 human genes in order to get these patterns into a useful format that can be applied clinically. In fact, our colleagues at Genomic Health Institute nearby here in the Bay Area, they have developed a panel of 21 genes and they use a technology called PCR, which is more precise than these chip approaches. And they can precisely measure the expression of several genes that are involved in uncontrolled proliferation, several genes that are known to be influenced by estrogen receptor expression, genes that are responsible for invasion. Here's that oncogene that I mentioned called the HER2 gene and associated genes. There's a suite of reference genes that are provided as internal controls. And then a few other genes that are associated with prognosis and breast cancer from the literature. And you can quantitate these and give them a score and give them weighting factors and an algorithm and you can come up with a quantitative measure of risk of recurrence. It's called the recurrence score. And if the recurrence score is low, then they have a low probability of having a relapse event in the next 10 years. There's an intermediate zone and then a high risk that we think especially are needy in terms of treatment. So this has been applied not only for prognosis, but also for prediction of response to therapy. In particular, response to chemotherapy. Back in the day when we didn't have this understanding of intrinsic subtypes and how diverse and unique each breast tumor is, we used to treat them all the same. If tumors grew beyond a certain size, we recommended chemo to all those patients in the 90s and the early 2000s. It didn't matter because we weren't sophisticated enough to understand who really needs chemo and who didn't. We didn't want to miss the opportunity for somebody to get an advantage. We didn't want to miss the opportunity for somebody to treat with chemo back in those days. And what the NCI in collaboration with Genomic Health found looking at large historical data sets is that if you have a low risk recurrence score and you randomize patients to receive chemo or not, it didn't really matter. These are statistical p-values and for a p-value to be significant in a biological or medical field, this is much greater than .05. So these differences are not significant. You can see that these survival curves are overlapping one another. So this suggests that chemo does not work in patients with a low risk recurrence score. What about high risk? And high risk recurrence scores, if the patients don't receive chemo, they have a poor outcome. Only about a 60% chance of this is a distant relapse-free survival. So that's the absence of metastasis. Only about 60% of patients in this study had the absence of spread of their cancer after a decade later. Whereas if they got chemotherapy, they had nearly a 90% chance of being free of distant metastatic relapse 10 years later. So this is very powerful technology. It can discriminate who needs chemotherapy from those who don't need chemotherapy. That's a huge advance. So this has been applied now routinely since about 2004. And here's data from University of Pennsylvania. And this has been mimicked in clinics all over the country, mine included. In the pre-oncotype era, that's the recurrence score trademark, oncotype DX. Before that, back in the early 2000s, chemo was given to about 40% of newly diagnosed breast cancers presenting with early stage, usually stage one or stage two. So it hasn't spread to other parts of the body, for example, those early cases. But chemo was still given to about 40% of those patients back in the early 2000s and 80s and 90s. Within a year or so, after the advent of this technology and its commercialization and availability nationally and even globally, the use of chemotherapy was cut virtually in half at this institution. And it's similar data across the globe at this point. So this is a huge advance, eliminating unnecessary toxic chemotherapy. And it's based on a molecular understanding of the molecular portraits of different subtypes of breast cancer. Extremely important advance practice changing indeed. So this is just the likelihood of recommending by the oncologist. It fell from 56% to 26%. Obviously, even though the oncologist recommended it, 56% of the time, only about 40% of the subjects had something right there. They knew that there was a propensity for overtreatment even back in those days. So an important advance. Now, this whole cell growth and senescence and cell death paradigm is highly regulated in normal physiological circumstances. In order for something to be that fine-tuned, it has to function as a network. And here's just an example of a schematic that was created by cell biologists. Showing the normal physiological circumstance of signaling receptors. There are ligands. An example of a receptor would be insulin receptor. And an example of a ligand would be insulin itself, for example. So there are a number of receptor families. This is the human epidermal growth factor receptor family, which is important in a variety of cancers. EGF receptor, the parent of this family, is very important in lung cancer. For example, HER2 is very important in a subset of breast cancers, as I mentioned. HER4 may be mutated in melanomas, for example. So this is just one example of one signaling receptor family. And you can see all of these downstream signaling intermediates in the circuit in a highly regulated and coordinated network. So this is how cells are supposed to function. This is the normal physiological circumstance in normal breast cells. This is what happens every day. Now, what do you suppose would happen to this network if all of a sudden there were an accumulation of genetic mutations for the genes that control all of these network signaling intermediates? The network would likely become deranged and look something more like this, I would argue. This I snapped on another vacation in Ho Chi Minh City a couple summers ago. So this is a deranged network, but it is highly functional. It gets the job done, however disorganized. And cancers are just like this. They're highly disorganized, but they still are dangerous because they're still highly functional. But I will argue that even though this is a highly functional network, it points out a vulnerability that we could potentially exploit. For example, what do you suppose would happen if you chopped down this telephone pole? The power would go out in probably half of South Vietnam. I mean, you've seen those things even in North America, the blackouts on the East Coast because somebody hit a telephone pole in Ohio or something. So this is a vulnerability. And if we could identify the telephone poles in cancer cells, we might really have something. And we've done that. And one example would be the estrogen receptor. That is something that works. We treat breast cancer patients who have estrogen receptors that are overexpressed. We treat them with anti-estrogen drugs and their tumors can melt away. It really works. HER2 is another example. You can treat a HER2-positive patient with herceptin and other treatments in some of those cases. That will bring down the network. So HER2 would be another example of another telephone pole. So now the mission in research is to find more of these common nodes in an otherwise disorganized network so that we can bring the house down in breast cancer. That's the goal. So I've talked a lot about HER2. This is what a lot of my research has been focused on since the early 90s. Here's a mutation where you can actually see the HER2 genes using a colorometric probe. This is a fluorescent probe that binds directly to the HER2 gene. And this type of a mutation in these HER2-altered breast cancers are an example of what's called gene amplification. Normally you have two copies of every gene in your gene, one from your father, one from your mother. That's the normal situation. You should have just two. But in these tumor cells, during cell division, instead of making just two copies of the HER2 gene, it's like a Xerox machine getting stuck in the on position. It made bunches of HER2 genes by mistake. This is a mutation, a gene amplification mutation. And so there can be 20 or 30 or even 50 copies of the HER2 gene. And if there's too much of the HER2 gene, it'll make too much of the signaling receptor. And the receptor will become aberrantly activated. It's always turned on. And this is a growth factor that signals the cells to grow in an uncontrolled fashion. Here's an experiment we did in the laboratory when I was at UCLA. These are normal cells that lack over expression of the HER2 oncogene. And you can see a few little colonies here in this colony count assay growing in some soft auger. But if you introduce this gene and over express it to the same degree as you see in clinical tumor samples with HER2 gene amplification, you can see a marked increase in tumor growth. And you can even see depletion of nutrients in this Petri dish, because this pink color here is a reflection of the pH in the soft auger. And you can see that it's being bleached here because these cells are growing so fast they're depleting the nutrients and altering the pH in these dishes. So this is a striking change as a result of an accumulated alteration due to a mutational event. And this happens in about 20% of all breast cancers globally. And so we identified this as a target. This would be a target for both diagnostics and also a target for therapeutics because if we could interrupt this process perhaps we could have a successful therapeutic intervention. And that in fact has been the case. Here's an experiment in a large study involving thousands of patients that were randomized to conventional chemotherapy. These are the initials of the various chemotherapy drugs. I'm not going to bore you with the details of the names. If you treated patients with newly diagnosed early-stage breast cancer that had not yet spread to any other organs you can see that over time these patients still had relapses and had a relatively poor prognosis in the absence of HER2 targeted therapy. But if you added a drug that blocks this aberrant HER2 signaling in this case Herceptin, the generic name is Trastuzumab for those of you who are aficionados and generic nomenclature. This made a vast impact on the probability of a recurrence of the cancer. These patients had far fewer relapse events. In fact the event rate was cut about in half. This is called a hazard ratio and it's the fraction of patients who have not yet had a relapse event. So it was cut nearly in half by simply adding Herceptin to the standard treatment recipe. And if you looked at their overall survival over time the hazard ratio for the overall survival is 0.67. That means that 33% enjoyed improved survival a reduction in mortality of breast cancer just due to this targeted therapeutic. And here I told you earlier about P-values here's a P-value that's significantly less than 0.05. So these are highly statistically significant results. And this was practice changing. This was published in the New England Journal of Medicine and is now the standard of care for treatment of HER2 positive breast cancer globally everywhere. So one of the next projects that I took on in the lab is a project that was to build on this concept of mixing Herceptin with chemo. The problem with Herceptin mixed with chemo is that the patients that are treated with chemo still all have the side effects of chemo. Upset stomach, hair loss, low blood counts, infections. Many of you are aware of the common side effects of chemo. And those are very troublesome. Some in fact can even be dangerous. So this is not ideal still. And we wondered is there some way that we could exploit the ability to target this receptor with Herceptin and get rid of the chemo side effects. And so one idea that we had was instead of just giving the chemo separately and giving the Herceptin separately, we thought what if we actually chemically bound the chemo directly to the Herceptin drug itself. This Y-shaped glob here is actually Herceptin. It's an antibody protein. It's a protein that normally is part of the immune system that fights infections. And we simply redirected it so that it could bind to her, too. So in this case, we're using Herceptin not as necessarily a way to block signaling, but we're using it as a vehicle to merely deliver a toxic payload by chemically linking the chemo directly onto the Herceptin molecule. So when this reaches the tumor, in theory, hopefully the chemotherapy would become freed up through some chemistry that happens in the hostile environment within tumors. And as the Herceptin binds to the target, hopefully the chemotherapy will be liberated and that will specifically attack the tumor. And then the chemo isn't spread around the rest of the body. That was the idea. And so we tested this in experimental systems. And here are some of the results. Here's a control human, her too positive breast tumor that's untreated, essentially. They got some control vehicle solutions. Here is chemo and Herceptin given separately, the old-fashioned way, in the maximum tolerated dose of these drugs. And you can see it's effective. It reduces the tumor burden by at least half. So quite effective. Well, here's this so-called antibody drug conjugate where we directly coupled the chemo onto the Herceptin. But this had similar activity to the free chemo and the free Herceptin, but it only one-tenth the dose. So at one-tenth the dose, we got all of the effectiveness. And at one-tenth the dose, you have only one-tenth the toxicity. Now, the problem with this project was that it was in collaboration with a small biotech company at Atlanta. And this particular molecule, the first one that we published, this was the first publication of this concept in literature back in the early 2000s, this molecule did not survive the so-called valley of death due to lack of funding. We couldn't get this into human trials because to do that, you have to satisfy all the federal regulatory authorities to do all the manufacturing products, stability testing, purity, non-clinical toxicology and pharmacology testing. All that stuff is expensive and it's not covered typically in NIH grants. They don't fund that kind of work. They expect industry to do it. So this little biotech company didn't survive. But there's another larger biotech company here in South San Francisco that had deeper pockets than we had. And so they did the same thing, borrowing on the same idea. They coupled a different chemo than the one that we published. And they picked one that's even more toxic than the one we were working with. They picked one called metansin, which is a very potent toxic species. It's so toxic you can't even give it as a free agent. It was tried and it's not tolerable. And here's Herceptin again. They coupled it directly to Herceptin. It binds to the HER2 receptor on the cell surface. This is the surface of a cancer cell. This is the HER2 receptor. And it's actually drawn to be anatomically correct based on the X-ray crystal structure in this particular cartoon. So this cartoon is actually fairly accurate. What happens when Herceptin binds to this receptor is that the receptor is internalized in a process called receptor mediated endocytosis. I'll give you a quiz after this. So it goes into this compartment in the cell called the endosome. And the endosome is a very hostile compartment. It's got low pH, has tons of enzymes. And these enzymes will actually degrade the antibody and that frees up the toxic chemo and that will kill the cell. The beauty of this is that this particular Herceptin-based antibody drug conjugate has no nausea, no vomiting, no hair loss, no drop in the white blood cell counts. So far safer than chemo plus Herceptin. So we put this to the test. This drug is called TDM1. It doesn't even have a name yet, it's so new. We put this to the test in a large, randomized trial of HER2 positive breast cancers with metastasis. With a thousand patients, we randomly assign them to the standard FDA approved treatment for this condition, which is another HER2 targeted agent called Lepatinib. We don't have time to get into that today. Lepatinib plus a chemo called capesidobene. That's the standard. And we tested this new antibody drug conjugate head to head versus the FDA approved standard treatment. And I can tell you that this study is positive. That's in the public domain because there's already been a press release. And this study will be presented for the first time at the American Society for Clinical Oncology meeting in about a week and a half in Chicago. And I'm pleased to be a co-author on that paper. It'll be presented in the plenary session. So anything that's plenary at ASCO, you know that it's going to be significant and practice changing. And there's already a paper in progress to be submitted for publication. In the interest of time, I'm going to fast forward here and wrap up quickly. You all know that we now have sequencing technology so that we can sequence the entire genome of an individual. And this has now been applied to sequencing individual cancers. And when this is done, you can actually catalog all the mutations on all the chromosomes. This is called a circus plot. And it has each of the chromosomes numbered in this circus around the outside of the cell. These are all the chromosomes. And then wherever there's a mutation in the chromosomes, these little black lines indicate point mutations. Wherever there are amplifications, those are shown in green. And areas where genes are deleted, those are shown in red on this part of the circus. And then wherever there are rearrangements where the chromosomes are scrambled and fused together, those are in these arcs in the middle showing connection of one chromosome to another across, you know, large areas of the genome. So you can catalog all these changes in a breast cancer. And you can see that it's extremely complex. In this one particular breast cancer case, there were 157 different gene rearrangements alone and many more point mutations. So extremely complex. When you do this in a large cohort of breast cancers, this is the first 500 cases that have been analyzed in this way, just presented, not yet published, just presented in December for the first time at a national meeting. You can see that there are many, many different genes that are mutated in human breast cancer and no two are alike. Every single tumor is unique. In fact, within one patient, in different parts of the tumor or in different metastases, there can be different mutations as well. So highly complex. This is both an opportunity, but also a challenge. The opportunity is, many of these might be potential therapeutic targets. That's fantastic. And we're really excited in the lab to take a critical look at these, to see which ones are drug-able and come up with new therapeutic strategies that can be personalized and individualized to a particular and tailored to a particular patient's alterations. But it is a daunting challenge because that means a whole lot more work. This will take an army of post-docs and grad students and lots of thesis projects to get this work done. So that's where you come in because this is going to take a lot of support. So any of you with college-age kids can send them over to our labs, please. We'll put them to work. Here's an example of a cancer cell line from my laboratory, her two-positive cell line. And we selected clones that were resistant to her two-targeted therapy, in this case Lepatinib. And so we asked the question, well, what happens to the genomes of these cells as a result of drug resistance? Well, they just go crazy. There were 37 new point mutations around the circuit. There are dozens of copy gains and losses that are new and new rearrangements that occur. So this is just underscoring the complexity of just a simple experiment giving rise to drug resistance. This is happening in patients all the time and we need to understand this and embrace it and exploit it. So in conclusion, we have now this challenge of this relevant genomic information. What we know today is pretty simplistic. The only things that are routinely measured in human breast cancer in the clinics right now are estrogen receptor, progesterone receptor, HER2, and then that 21 gene assay that I showed you add a few more to the list. But what we need to know is much more dynamic and complex than that and will require informatics, bioinformatics, new research tools, tons of computing power. It's staggering how much computing power you need to be able to dissect these complex networks, especially when they're drained and especially when they're unique to each individual patient. So we've talked about conventional cancer therapy and the limitations there are particularly toxicity and the one size fits all model that we used to have for breast cancer employing surgery, radiation, chemotherapy. These are still tried and true and they're going to be around with us for a time. But these are old school. Molecularly targeted personalized therapy is highly sophisticated. It's complex. It exploits new technology. It's sexy but it's expensive. And so I don't know that this is necessarily going to be the only solution. It may be that in the future we have integration and synergy of this new translational personalized medicine into the traditional methodologies and we have to be smart to come up with the right balance of integration of new treatment methods in the clinic and hopefully do a little bit better than this fiat. So with your help, it's my hope that we will accelerate the pace of discovery here at Stanford. Sorry this came out a little bit blurry. I was drinking a Starbucks and talking on my cell phone and I drive with a manual so it was a little bit blurry. Thank you very much for your attendance this morning. I really appreciate it. I can't thank you enough for taking time out of it. All of you have very busy schedules like the rest of us and it's really exciting and thrilling to see a packed house upstairs today and all of you coming to the sessions were really grateful. Thanks for your participation. Questions or comments? Yes, please. We have a mic for you. Good. I'm interested in what company is producing TDM-1. That would be Roche Genentech. When I said a large biotech company in South San Francisco that was the code for Genentech. They're the 900 pound gorilla in the room usually. Thanks for a great talk Dr. P. Grapp. The translational aspect, we know there's been tremendous strides and you've sort of shown that on your last slide. So what do we need to get to that next step where we bring what's in the bench or in the computer I guess to the clinic? What do we need next? There are a number of barriers. As you know, one of the barriers is patient participation. We need patients to understand what the research needs are, what the needs are in the pathology laboratories for sophisticated diagnostic approaches including genetic approaches which has a whole layer of ethical and social issues with regards to insurance and the legal system etc. We need patients who are willing to volunteer for new therapeutic strategies some of which we know aren't going to be successful. Not all of the experiments we do work. For everyone that works like TDM-1, there are nine others that don't make it. So it's frustrating both from our point of view and also from a patient's point of view knowing that when they are volunteering to participate in an early phase of testing that it may not work for them. They have to be altruistic. So patient participation is critical and nationally in adults, adult patient cancer participation in clinical trials is single digit as a percentage. It's very low. Whereas in pediatrics it's about 95% plus and you'll notice that the advances in pediatric oncology have far outpaced those in adults and that's one of the reasons. The next limitation is this whole translation issue that you bring up. There has to be collaboration between so many different types of experts because there are physicists and biochemists and cell biologists and imaging people and physicians and statisticians and all these people have to work together and use a common language and be focused on the same problem. The problem is the same but you're looking at it from many different points of view and that's extremely difficult to pull off and it requires enormous resources space, expertise, recruitment, just all of the things that are so rich here and we're so fortunate to be here at Stanford to have access to these kinds of tools and the expertise but it's difficult to maintain and that's one of the other limitations. The next thing is just the complexity of human cancer. I think nobody imagined that there would be so few genes that are frequently mutated in breast cancer. What we were hoping to find from those first 500 patients is maybe 20 other genes that are mutated half the time or 70% of the time or even 20% of the time that would be the next estrogen receptor and the next HER2. That's what we were hoping to find. What we found instead is dozens and dozens of genes that are only affecting 1% or 2% of patients and that's going to challenge the current therapeutic development paradigm from the regulatory point of view from the FDA's point of view and from industry's point of view because to get a new oncology product to market now costs more than a billion dollars. If you go to the boardroom at Roche and say, well I want to invest a billion dollars for a market that's 1% of breast cancer they'll laugh and then kick you right out. So this is going to have to be a revolution in new paradigms to bring these discoveries to practical application and the regulators are going to have to be satisfied with something less than these thousand patient randomized placebo controlled trials. You can't do that with 1% of breast cancer. So there aren't enough patients on the planet to do a thousand patient study necessarily. So the regulators are going to have to... Please join me in giving Dr. Pegram a big hand. Thank you very much. Please stay and ask a few questions. Please do. We need to make room for the next session that's coming in. Thank you Dr. Pegram. The preceding program is copyrighted by the Board of Trustees of the Leland-Stanford Junior University. Please visit us at med.stanford.edu.