After Ovarian Cancer Surgery

>> good afternoon, we will begun. we're very particularly grateful to one of our speakers today, dr. hasaan, who has been kind enough to come to bat for dr. pastan who unfortunately due to a family problem is unable to be here with us today.

so just broadly thinking, what seems to be the major advances in cancer therapy in recent times, seems to me they fall into two and a half categories. one is the use of genomics and genomic profiling, to detect possible susceptibility genes and also to lead to development

of targeted therapy based upon identification of specific genes, their protein structure and so forth. related to that is development antibodies which seem to be increasingly merging on the scene as being potentially powerful cancer

chemotherapeutics. but the other major area is the what i call taming of biologics toxics to therapy and sort of in thinking about it, it struck me attention that quite possibly birds and mammals are the only two species to my knowledge at least don't elaborate biologic

toxins. we do our toxicity more verbally and physically than biochemically, i guess. but throughout test of biology, the elaboration of toxins which kill by a variety of mechanisms which are not reversible, this is very different from say

drug-related chemotherapy which may affect cell and give rise to alternative ways for drug resistance, for example. these biologic toxins are like the guillotine, there's no way to put the head back on once everything has been killed off when you think about it, reptile

s, bacteria, fungi, plants, they live in a hostile world of battles going on all the time. and toxins are critical to their protection and maybe in some instances to their own biology other than just as protection. in the bacterial world the group

of clostridia, tetanus botulinum, cholera, these toxins are devastating. we understand something about how they affect diphtheria but they affect a cell in a way, this is the important thing, it's all overwith, there is no escape once a cell has been

affected so it's a powerful way but how do you direct it to a cancer cell as distinguished from any other cell because any other cell can be killed by the toxin as well. so with these two big area of let's call it genomics, toxinology in relationship to

cancer therapy, that's really the subject of today's demistfying medicine. we're extremely fortunate in having -- before i get to that, i would ju find out two questions come to mind. should i think about relate to subjects that will be discussed

that you might think about what do these people all have in common that can relate to cancer you will learn the next hour and see how that has -- how that relationship exists. i seem to have lost -- the other one is for example, we have patient a b and c, and they all

have a hidden cancer which histologically looks virtually the same. a is treated with drug one. it gets a fantastic response a is treated with drug two, nothing happens. the disease progresses. a is treated with drug 3 -- i'm

sorry, that's right -- no, b is treated with the first -- the drug the other fellow didn't have a response to, and he or she shows big improvement. the account for this extraordinary difference in cancers which at least in old fashioned pathology all look

relatively the same under the microscope so these two big areas touched on are really the. so this discussion and we're most grateful because here at nci are two outstanding investigators who will discuss these problems with us. so i would like to introduce

both to you at the beginning. first is rafi hasaan, who graduated from medical school in kashmir in india, came to the united states, received clinical training in buffalo. and here the nci for ecology training was in oklahoma for a while, then came back in 2002 as

a tenured investigator at nci and worked in the laboratory of which ira pastan was chief. and dr. hasaan has been largely responsible for this toxin related immunotoxin related treatment of the disease that all those people on that list that i showed you are suspect of

potentially having and that's call mesothelioma which we'll here more about. second speaker is paul meltzer who received medical training i believe in kansas and a ph.d. at cal tech and is widely known for his work in molecular biology, particularly genomics

and here he's the head of the molecular genetics section and the national cancer institute and those of you who read or will read some of the references we put up on the website, realize the major interest is in understanding the use of genomics to detect differences

in responses of individuals with different kinds of tumor,particularly sarcomas but many others and to try and use genomics for the better understanding of perhaps both diagnosis and development of therapy. so the first speaker is going to

be dr. hasaan, thank you very much for being with us. >> thank you very much for the invitation, i'm sure we're all looking forward the hearing ira, he sends his apologies. so i'm going to talk about immunotoxin mesothelioma and this work we have been doing

with the last 20 years or so. this to get you up to speed on mesothelioma, it's the thank you ma of the tear tone yum, there are 3,000 new cases diagnosed in the u.s. each year. and this disease disease has looked at treatment options for prognosis.

this we call it a cat scan you inject radio label glucose to patients and as you can see the tumor is pretty diffuse and takes up the raid quo labeled glucose. it's pre-- radio labeled so mesothelioma can go to different parts of to body

lining of the lung, the tumor is surrounding the lung, it can involve the abdominal cavity where you can so the diffuse tumors all over the -- occasionally we see patients we get per pericardial mesothelioma because it line it is pericardium and patientspatients with

mesothelioma of the tumor originalis, the lining of the testes so mesothelioma can involve different parts of body which have mesothelial cells but the most common part is the plural mesothelioma. in 1960, the a pathologist in south africa noticed an increase

incidence of mess these owe ma among -- mess these owe ma abong asbestos workers in south africa and showed that this increase was because of asbestos exposure. and in the previous slide there's increase of mesothelioma in minors, family members of

minors because these minors come home with long coatses contaminated with the asbestos fibers when they go home the asbestos dust would infect their family members. we also see a lot of patients who work in different submarines many years ago because asbestos

was a fantastic insulator and widely used in ships. there's long latent period between exposure to asbestos mesothelioma, usually 20 to 30 years, so many patients have remote history. but there are many patients who do not give a gad history of

expoe -- good history of exposure to asbestos so there are other etiologic factors for getting mesothelioma, sv 40 is controversial and in general p the thinking is that it's by itself does not lead to mesothelioma. however there's increase indense

of mesothelioma in patients with hodgkins disease or non-hodgkins lymphoma who have got radiation in the past and we see young patients cured of their hodgkins disease or lymphoma but present with mesothelioma. recently gene called brca 1 associated protein is implicated

in development of mesothelioma. about 40% of spotted mesothelioma have deletions or mutations of these genes but there's also germ line mutation in which leads to increase susceptibility to mesothelioma so a genetic reason for developing mesothelioma and

there's a lot of interest in that. plural mesothelioma, the mesothelioma lining the lung is the most aggressive tumor and it has poor prognosis. surgery has a very limited role because most of these patients present with advanced disease.

most chemotherapy drugs do not work. the only drugs that work is a combination of methotrexate, an anti-folate with a cisplatin. even with the best therapy that we have today, the median survival of patients is only 12 months and almost all the

patients are dead by two years so there's clearly a lead to develop better therapies for this patient. work over the last 20 years or so focused on a protein called mesothelin. it's a tumor presenting antigen identified in ira pastan's lab

many years ago, please thee lynn is a -- me sew thee lynn is a glyco protein, the normal tissues is limited to the mesothelial cells that line the pleura and pericardium, that's important to develop a targeted therapy because you don't want it to be expressed in important

organs. the mesothelin gene has 71-kilo dalton attached to cell membrane by gpi linker and it is clear the protein that 31-kilo dalton we call mpf and the membrane bound follicular dalton fragment we call mesothelin. that is the target for therapy.

both npf as well as mesothelin is detected the the serum and can be good-bye markers but i'm not going to talk about that today. we and others have shown that mesothelin is highly expressed in many human cancers. almost 100% of epithelial

mesotheliomas as well as pancreatic adenocarcinoma highly expressed mesothelin, it is also expressed in 70% of ovarian cancer, about 50% of lung adenocarcinoma and other tumors including gastric sighnoid and bill area carcinomas. so though work so far has been

focused on mesothelioma, the high expression of mesothelin in many common solid tumors makes therapies applicable to other tumors. the biologic function is not known, many years ago (indiscernible) in ira's lab made mutant mice which lacked

the mesothelin gene but the mice has no phenotype and reproduced normally. we have shown recently mesothelin is a novel cl-1 binding protein and play a role in tumor implantation in the pure ra peritoneal cavity and -- pleura peritoneal cavity.

in the peritoneal cavity of mice. so this will suggest that mesothelin may have a role in tumor metastasis or suppression. however, focus is on targeting mesothelin for cancer therapy. the high cell fate expression in solid tumors makes a good target

for antibody therapy. expression of the normal pleura per tone yum and pericardium was when we decided to explore for this is one of my first papers and i was a post-doc in high ra's lab where we had in monoclonal ann body to mesothelin called k-1, these

were nude mice that -- mesothelin negative tumor in one flank and tumor of the flank, was radio labeled antibody to mesothelin k-1 with nd 111 as you can see there was preferential uptake in mesothelin expressing tumors. and this was one of the studies

that set us on this path of exploiting mesothelin directed so it took us about 15 years to show in patients that antibody to mesothelin does in fact look at localized to tumors, this is a clinical trial that we did often anti-mesothelin meataxe has been, radio labeled with md

111, this is a patient with mesothelioma, this is a cat scan with mesothelioma, pet scan inject radio labeled glucose, you can see uptake tic in the tumor, and if you did a biopsy you see diffuse please thee lynn expression. when o mesothelin expression.

when with we inject the radio label antibody, this is proof of principle in patient that an antibody targeting mesothelin does localize to mesothelin expressing tumors. so we are using many different ways to target mesothelin but today i'm going to talk mostly

about work using immunotoxin. as you know, very few antitumor antibodies by themselves are good at killing cancer cells. there are some notable examples such as trastuzimab an antibody to her 2 for treating breast cancer, once it kinds to tumor cells it leads to cell death but

by and large tumor antiguys tumor antigen they don't mediate cell death. so to take antitumor antibodies useful we can on with different substances so they become more potent. one way is you take an antibody to a tumor antigen and attach a

toxin to it, that's an immunotoxin. the other way is you take about antitumor antibody and attach a potent chemotherapy drug to it. these are the antibody drug conjugates where you can deliver highly potent chemotherapeutic agents to the cancer cell and

the third approach is that you take antibodies and radio label them with radio isotopes which are used mostly for treating blood cancer. so to immunotoxin consists of antibody or antibody fragment linked to toxin. so this is an antibody so you

take an fv and you take a toxin t. toxin with we're working on is pseudomow in this toxin a made by the bacteria pseudomonas domain. the binding that binds o the cell membrane, it has what we call a processing domain whose function we don't know.

ain't has a catalytic domain in red, adb loungings leading to to make an immunotoxin, we get rid of the binding domain of the toxin and replace with the of interest so fc directs the toxin to the cancer cell and mediate cell kill. once immunotoxin binds to

receptor in this case mesothelin it is take up up by a class coated pip, endosome, the toxin is processed, goes to er and ultimately to the cytosol where it leads to cell death. so ss-1p is an anti-mesothelin immunotoxin, it has high affinity for mesothelin, it is

cytotoxic to mesothelin expressing thank you more cells with patients with mesothelioma and ovarian cancer an regression of mesothelin positive tumors many mice. so i'm not going to go with the clinical data but we did a clinical trial ss 1p in patients

who had advance cancer that failed all standard therapy so we treated patients with tumors expressed mesothelin, we established what was the maximum dose of the antibody that you can give to the patient which was 45 micrograms per kilogram, the dose toxicity that is

toxicity where we cannot give ss 1p was pleuritis which is inflammation of the pleura. this is something we expected because mesothelin is present on the normal pleura. but we saw minimal activity, we didn't see tumor regression. the main reason we did not see

any regressions was ss 1p was immunogenic. so it's a big foreign protein and when you inject to patients they develop antibodies to the drug so you give more drug the antibody play the drug so that's a big problem. and about 90% of the patients

developed this antibody after only one cycle so most patients got one cycle and did not get any more drug. so the immunogenicity problem immunotoxin has limited development for solid tumors, 20 years ago there was a lot of excitement about using

immunotoxin, as targeted but they failed in the clinic because of this problem and previous efforts to overcome this problem have failed. and many felt it's a huge problem that cannot be overcome. so that's why academic center as well as pharmaceutical

industries gave up on studying immunotoxic. we have focused on this problem using two different approaches. one is preventing human immune response to immunotoxin. and the second one ambitious approach, is to make an immunotoxin inherently less

we and others have looked at different ways to prevent in human immune response to immunotoxin and agents typically immunosuppressive agents such as cyclosporin, single age toxin retuximab target cd 20 and depletes b cells do not work. all these patients develop about

antibodies. however one colleague at nci dan father who works on bone marrow transplant has shown that a regimen of pentastatin plus sigh toxin which can deplete t and b cells, results in host immune depletion without myelosuppression and that it was

safe and patients with immunocancer so that was exciting so give a general in the lab and we thought that we should evaluate this approach whether this will work so we did an experiment where we took immunocompetent mice which have a normal immune system and

immunize him with ss 1p so we gave it weekly for nines toes, that's a -- doses. that's a immunogenic dose of ss 1p and all the mice shown here in red developed and ss 1p antibody. that was no surprise. but when we took a second group

of mice that were treated with pentastatin and sigh toxin, none developed antibodies to ss 1p, that's an amazing result something we had never seen before that these two drugs could block immune response in so we decided to see if the same would hold true in patients.

so we decided to do a small pilot study, evaluating ss 1p with pentastatin and sigh toxin in patients, the rationale was depleting the host t-cells with pentastatin and host b cells with sigh toxin decrease formation of anti-ss 1p antibody, and we know

pentastatin has no escaping solid tumors, and that sigh toxin has no antitumor activity in patients with mesothelioma. so in this clinical trial because this was very this was a very different target we're using combination of three different drugs the only patient

we decided to enroll in this study would be patients who had got previous treatment with treatment refractory disease and progressive disease when getting the study, the primary end point was to see if it was safe and feasible to use a regimen. the primary end point was to see

if using these two drugs we can decrease the antibodies to ss 1p the clinical trial team was very similar to the mouse experiment where we take patients give sigh toxin, for 12 days give pentastatin, three doses of pentastatin with the goal to knock off t and b cells and then

you come with the ss 1p on day ten and give them three doses. if the patients do not develop antibodies after one cycle, a cycle is four weeks. then they could get second cycle of this regimen but sigh toxin only for day. very similar to the plan in

so in this pilot study we enrolled 11 patients the median age was 54. most were male, nine of the patients had plural mesothelioma, two patients have abdominal mesothelioma and the median number of therapies these patients received before coming

on the study was three. the side effects that we saw is great for lymphopenia, meaning lymphocytes decrease, this is exactly what we wanted to do, we wanted to decrease the lymphocytes so they did not make and we saw some anemia and increase liver enzyme.

and the side effects that we saw with ss 1p was what we expected that they would be punitive chest pain because of targeting as shown here the absolute lymphocyte count of patients each is individual patients. this is a lymphocyte count before treatment and after just

one cycle all the patients had a decrease in the lymphocyte count. but at the same time there was no change in the neutraphil, the neutraphil help us fight ine so that was good so the -- infection so the regimen was selected to lymphocyte but not

neutraphil because we worry patients will be at risk for infection. shown here are the t-cells that cd4 cells as you can see all the patients had a significant drop in cd4 t-cell as well as significant drop in their b cell looking at cd 19.

so both regimens did what they expect to do. but what was surprising was that this regimen decreased the formation of antibodies to ss 1p. as i showed you earlier in the phase 1 trial ss 1p by itself, 90% of the patients had

antibodies after one cycle. however in the study only two out of ten patients developed antibodies at the end of one cycle so this regimen for the first time could allow us to give cycle of ssp. but what was even better was that we saw some very good

responses in patients who had a lot of disease. three out of ten patients who could evaluate the response had a partial response, the partial response in clinical oncology mean you look a at the tumors measure them up, if you decrease the tumor 30% or more it's a

partial response and complete response that you see nothing on the ct or pet scan so patient two had a very good partial response and still alive almost three years out. patient 3 again had a nice partial response more than three years.

both patients as well as patient five had three regimen in hospice care at the time we enrolled in the study but they have an exceptional response and more than three years. and we also had a couple of patients who had stable disease but when they got chemotherapy

to which they had not responded before that partial response. so just giving an example of one of the patients, patient 3, this is a baseline ct scan at day 12, three months and 24 months so this is the left lung this is the right lung, you can see top to the bottom.

you don't see any on the right lung encircled by the tumor, but if you look at three months basically the tumors are gone, cat scan at 24 months, ct scan at three years looks about the same. this is a very nice direction in tumor.

patient five, the patients with abdominal mesothelioma with widely metastatic disease as you can see here subcutaneous lymph nodes and this is after two cycle it is lymph nodes are gone. this is a ct scan looking at the chest shown here is the tumor,

this is before treatment, this is about men huge burden of disease, this is the huge burden disease burden. at 1.6 months you see the tumor starting to shrink, they're more or less gop gone at day 21. if you do a pet scan, you inject raid yes label glucose, cancer

cells take it up and they wear black, a lot of uptake neck, chest and about abdomen and at 1.6 months while getting response there was worsening but look at eight months no or less all the ftg uptake is -- similarly patient 2, this is the third patient who had extensive

mesothelioma, before treatment a lot of disease burden, there was some worsening initially but significantly improved later on. so in summary, ss 1p plus pentastatin and sigh toxin in this trial was safe. it decreases the formation of antibodies to ss 1p, we saw

dramatic durable tumor responses in patients with treatment refractory mesothelioma. right now we're doing a small biomarker study to better understand what's going on in this tumors. so after years of working on ss 1p we're very happy at least we

have seen some responses in the patient w we saw recently, extensive response. so this is good but better to have an immunotoxin that is less immunogenic so you do not need to use other drugs. this is the work done primarily in ira's lab to make a less

immunogenic immunotoxin. so we have been using protein engineering to make such an immunotoxin, just to bring you back to ss 1p that's the drug we use in the clinic, it consists of an anti-mesothelin fv linked to a mutated p pseudomonas toxin in domain 2 and 3, domain 3

kills the cell, domain 2 we don't know what the function is. couple of years ago tom weldon a post doc in ira's lab did an amazing experiment where he got rid of entire domain two. we thought cytotoxin would never work but what he found is you can remove the domain two of the

toxin and make a toxin that we call ss 1 lr. doing so you remove most t-cell epitopes which are present in domain 2. but the toxin still retains its activity. so getting rid of significant portion of tok sin gets rid of

the epitopes of the toxin is more active. the next also post-doc in ira's lab, he identified what are the b cell epitopes in domain 3 of the toxin. and mutated them to less immunogenic epitopes, we made a molecule that we call ss 1 lo 10

we can retain the cytotoxicity but is less immunogenic. so over lead molecule consists of this portion ss 1 lo 110 links to an ants mesothelin assay b we are using an sav with less immunogenic toxin we call p-24 so all known human b cell epitopes in major t-cell

epitopes of pseudomonas toxis is mutated. compared to ss 1p and we can give much higher doses man ss 1p and know we are left. just showing an example, looking at the new molecule rr 1, this is cytotoxicity against primary mesothelioma cells we get from a

patient. shown in the green is the new molecule, rg 7787 compared to ss 1p, this is threefold more active than the parent ss 1 molecule. and just to summarize a lot of the pre-clinical work with rg 7787, it's a less immunogenic

anti-mesothelin immunotoxin. it has cytotoxicity against several primary and established cell line cancers that highly express mesothelin including pancreatic cancer and it has single agent activity against mesothelin positive in mice and there's mark synergy

with chemotherapy and we have established creda with clinical development so this is the rg 7787, developed and we treated first patient was treated last month. the second patient of the molecule was treated yesterday so we're very excited about this

new molecule. just to summarize our work, ss 1p with pentastatin and sigh toxin was produced significant tumor regression in tumor refractory mesothelioma patients. new generation immunotoxin, we believe should have more given

at higher doses or more treatment cycles have pure side effects. so ira and i summarized our work over the last 20 years about mesothelin and exploiting target for immunotherapy and i would like to acknowledge a lot of members in the lab and clinic

from my group, working on mesothelioma and mesothelin and ira pastan with the new a cca in ira's lab and collaboration with other pumps. thank you. [applause] >> thank you very much. that was wonderful, we have

plenty of time for questions, comments and encourage you to please ask anything you didn't understand or something you might do understand more. >> nice, so in your new trial where you have treated with this immunotoxin, these patients are also treated earlier like your

first trial. >> yes. these are patient whose failed standard therapy. >> what you see, mutated immunoassay just reduce the purpose of reduce the immunoactivity. why make the molecule more

potent, that was surprising we don't understand that we thought that we might lose activity, more actually than processing with how it goes into the tumor cells. >> i have two questions. >> it looks like you work with the antibody fragment but you

were saying how it can be linked to the full antibody, i was wondering why you chose to work with -- >> we work with fragment that has in solid tumors with fragments because if you use in whole antibody about 150-kilo daltons and you tie the protein

about protein so big molecule, macro molecule and problems with it going into the tumor itself. and so if you have a smaller molecule such as an fv it is about 60-kilo daltons that allows better penetration comparing chemotherapy drugs like tack solve which is one

kilo dalton, big molecules use the whole ants body in solid tumors, you may not get enough into the tumor. >> also for your linker, does it have to be cleaved under cell or toxins does it matter? >> so the pseudomonas toxin has the cleaving site so once it

goes there it is cleaved at serine. so when we make modification we make sure that serine titer is there. that's true. >> you have any hypotheses why the patients got worse and then eventually got better?

>> that's a good question. these patients are depleted of lymphocytes so it may not be it's possible once we give the drug causing inflammation that maybe recruitment of macrophages in tumor microenvironment and that's why there is increase uptake on the pet scan.

so that's what we're doing biopsies and so we can look at what the different cell populations are out there. >> why did you -- >> wait a minute. why did you choose pseudomonas toxin? >> that was made many years

before, there are two major toxins, two groups of toxin, one is the planned toxin and the other bacterial toxin, the plan toxin characteristically causing vascular leak syndrome so most have stayed from them, in terms of bacterial toxin there's two major toxins, one from

diphtheria and one by pseudomonas, the pseudomonas toxin we know well for the last 30 years how it works, that's why we start with pseudomonas path but there (inaudible). >> 50% replication (off mic) >> what i meant is about 50 --

30% of the patients have no history of [expletive] poe hour to asbestos but there could be other reasons why they get mesothelioma but that is not related to the toxin toxins. >> so we think that some is infiltration macrophages or other cells causing an

inflammation within -- >> you have antibodies to reach, you cannot touch immunotoxin -- i mean toxins, you cannot touch drugs. chemotherapeutic drugs or -- so which of the directions in just your opinion is more interesting?

because immunotoxin as we saw are immunogenic but drugs or nanoparticles would not be. >> we also -- looking at very different approaches, one is a naked antibody, other approach is antibody drug conjugate, highly potent drug. to be honest the problem with

these patients or other cancers is that these patient versus developed resistance to chemotherapy drugs, even if you develop a highly potent drug to the tumor, those cells i think are resistant. and there has been a lot of work, maybe 50 antibody drug

conjugates over the last five, six years tested in the clinic. only two of them are approved. one is using herceptin which by itself is a good drug, other is lymphoma. so i'm a little disappointed with the antibody drug conjugate.

we don't know until we see it. the immunotoxin unique mechanism of action by inhibiting protein synthesis so it could be useful for patients for resistance of chemotherapy drugs. >> mesothelioma if i recall is also associated with fibrosis that occurs in the area.

so what happens to that patients treated with immunotoxin? >> that's what we think but look at pet scans of the patients, it's a highly metabolic tumor. but there is a individual underlying inflammatory cell population of fibrosis, so when we look at some of these tumors

in these patients responses that are huge but ma what is remaining is inflammatory component we may not get rid of and we see that in the mouse model where we take tumors grow them in mice, where they stay at a certain level, it's possible some underlying fibrosis.

>> this h is a tumor that develops the in mesothelial cells anywhere in the body. >> so the most common is the lining of the lung. and the abdomen but i see a lot of patients who have mesothelioma of the lining of the heart and also mesothelioma

of the lining of the testes. o because those have my sowthelial cells. but how asbestos gets there i don't know. >> are those metastatic lesions or asbestos? >> they' primary. they start out, i have a

16-year-old patient right now who just presented with -- tetanus around the heart, nowhere else. -- thickness around the heart, >> how limiting is the pleuritis you have seen with the patients? the -- as a side effect? >> how we manage it?

>> how limiting has it been for the patients? otherwise you have no toxicity -- >>s that the main toxicity then we went up on dose now we're at a dose which we feel can be given safely. patients still get pain but it's

managed easily. the other side effect that we sometimes see is that patients have increase in the way they have swelling in the legs, because of what we call vascular leak syndrome but that's not limiting us in giving the drug because it's pretty easily

managed. and self-limiting. >> i think particularly for the non-mds in the audience, must gain some appreciation that this kind of clinical investigation being applied to patients with lethal toxins, it requires, nots just a matter -- it's not a

cowboy issue, you go out and you shoot. you can only imagine what it took to get proposals like this through an institutional review board, for example. let alone the enormous amount of work that has to be done to provide safety at every single

step of the line. and i suspect that's wherea lot of that ten years went, correct? >> sure. >> we'll have time for more questions, thank you very much. >> so after that really elegant talk of patiently developing one treatment for one disease

improving it, i'm going going the to go the other direction to the global world of cancer diagnosis. real challenging to figure out how to do that and still be (inaudible) cancer is such a complex huge family of diseases with so many issues, it's hard

for me to think of the right way to do this without obfuscating so i'm going to try to give you the 50,000-foot view of the field that i work in. and try not to get bogged down into a lot of technical details but rather the concepts that i think are driving the field

forward. which is unbelievably rapidly moving. this was incredibly quickly evolving field that has had a huge amount of interest in it right now. so i think that anyone even non-clinician understand it is

concept of making the diagnosis of the disease. so if patients have a problem, whatever kind of progress, in order to come one the appropriate treatment for the problem, the appropriate advice for the patient, you need to know what it is.

imagine a patient with cough and difficulty breathing you might do x-rays and bacterial logic test and decide they have pneumonia and you want to know what kind of pneumonia they have and give appropriate treatment for that. so that's where we'd like to be

so if you were to go back in history, to probably any time before the 18th century, you find that cancer was thought of as a tumor, which is nothing more than swelling. people get swelling all over their body for lots of reasons. but i'm sure that astute clinics

even in ancient time knew certain kinds of lumps in certain places with a certain feel when you touch them, and certain symptoms will likely be a bad kind of swelling people die from. but it's remarkable how slow this was to develop into more

precise ideas. so if you go back initially cancer was thought of as an anatomic problem where we're really what one would want to do is say well a tumor of the brain would be brain cancer, tumor in the lung could be lung cancer. and so on.

so it was a cancer of an organ. and it really wasn't until this is sort of absolutely amazing to me, it wasn't really until the late 19th century with the development of good microscopes and importantly the dive to enable you to take sections of tissues and stain them and look

at them under the microscope, that the simple anatomic diagnosis became supplemented with a cell level diagnosis tissue level diagnosis, histology. so instead of saying this must be brain cancer because it occurred in the brain, you also

could look at the tumor and the field of pathology, has been devoted to understanding what these histologic patterns represent. indeed the trained eye, of the pathologist can recognize silties that others don't see that helps subtype tumors in

human body. so this is still the most essential basic mechanism for the definitive diagnosis of cancer, it's the basis for the recognition of all subtypes of cancer, and it's been a platform for continuous diagnostic refinement since these

fundamental procedures were developed 150 years ago or so. and it's still the basis of all clinical decision making. so when a patient is referred to the nih clinical center for cancer treatment, essentially the first thing that is done is try the get hold of the paraffin

block which contain as piece of the tissue, we take it o the pathologist and con firm a cancer that actually is the type of cancer that they think they have so it's absolutely the basis of clinical diagnosis and the basis of selection of treatment.

importantly it's the basis of all clinical cancer research so if you want to compare different drugs in 100 patients you need to know they have the same disease or more or less the same disease so this is essential and sometimes when i have a lot of pathologists they thing the

things i'm saying is an attack on traditional pathology, i no way feel that, this is absolutely the essential root of cancer diagnosis combined with clinical and imaging study in a few laboratory studies but it's really the bedrock of cancer i'll wear the question i don't

think i'm going to be able to answer it defin actively. but one thing people think ant now, and i will just -- about now and i will focus try to come back to it later is our progressive ability to subdivide cancer into finer and finer precisely defined entities based

on traditional ways of classifying tumors, distracted from the potential importance of commonnalties among tumors which cut different histotypes. you already heard that, in the previous talk, mesothelin is a great target mesothelioma but it's also expressed in other

and maybe starts out as mesothelioma treatment will turn out more broadly relevant. so this is i think a little bit of a shift in our thinking. so new advances in diagnosis are we hope based on an every deepening understanding of cancer biology and advancements

in biotechnology which led us to see things in tumors that we were never able to see previously. so the goals of this kind of work is to make a biologically informed cancer diagnosis which will create the opportunity for biologically informed clinical

decision making. and hope fluework towards precision cancer therapy, the term used to be personalized medicine but it's supplanted appropriately by precision medicine. i will digress a moment and talk about the human genome which is

the field that i come from i'm a pediatric oncologist and tumor geneticist, not a pathologist and i think about things, i see the world through the lens of the human genome more or less completely. and this has been an incredible process that went from discovery

of the structure of dna and first report in 1953 by watson and crick, to the report of the complete sequence of the human genome in 2001. during that period amazing things happened so tremendous advances in understanding genome function and molecular biology

in little less than 50 years here, but there was also tremendous technology development driven by the research which has built momentum and moving forward to clinical applications rather rapidly. this is the idea we can study

the genome function with what people call integrated genomics. basically the genome is basically the sequence the dna sequence carried by individual organism and in our case the human genome. but it has many properties that we can measure.

so we can look at for example the genes that are encoded in the genome and how they're expressed copy number from individual to individual. how dna is modified by post translational and biochemical processes. we look at sequence variation

phone tween individual, things you ear born with or things that happen during one's lifetime. it's pretty clear to understand how the genome works, we need to bring all of these different approaches as well as others in to play. i'm going to talk very briefly

two slides on technology, i don't think i can spend time discussing these things in detail. but the two technologies that have culminated as we develop the ability to analyze the genome, first arrived really high through put platform was

microarray and basically microarray is something that -- nothing but solid port on which the moderner ra, millions of -- era millions of oligonucleotides are attached. these create chromes used to analyze almost any femur of the genome.

this shows the evolution when we started working with the technology in '96 and we got only about -- so at present where we have in the span of ten years we went in tens of millions of probes and very powerful still useful technology.

rather routine at this pick. -- point. the other thing that happened, this incredible change in ability to sequence genomes. this slide shows how this happens, and a unquestionably this was driven by the human genome project and by the

recognition throughout the whole world of biology, we're talking dna, something that applies to all fields of biology, and microbial plant animal you name it. so there is a huge motivation to get better at sequencing genomes.

this is what happened. this shows the genome compared to moore's law which is the ever falling cost of computing power which used to be so dna sequencing path moore's law abruptly here, about this point, this was with the development of new sequencing technologies and

the goal has always been the thousand dollars genome, as you can see on this graph, we're into the several thousand dollars genome by 2014 but we actually are looking at fall towards the thousand dollars genome in the next year or two there. is a catch.

you have to have a huge sequencing center which can spend huge amounts of money sequencing many thousands of genomes to get that degree of saving. but i have no doubt that disruptive technologies will continue to be developed and the

genome sequencing will be as routine as the blood count. at some point in the future, i can't say when that will happen. but i think it can be predicted with some certainty that it will happen. now, back to how these relate to cancer, back to cancer as a

genomics disease, that of course is what excited me since i was old enough to understand this concept which turns outs not to be that new. this is one of these very humbling quote from the history of science, from boveri, a german scientist in 1992 said

that a statement which was incredibly prescient. in every cell he said there's a specific arrangement for inhibiting and definite chromosomes which inhibit division, tumors would arrive if those inhibiting chromosomes were eliminated so this is the

prediction, correct prediction of the existence of the tumor suppressor gene but what's fascinating is though this idea came out at the turn of the last sentencery, it wasn't accept bud the scientific community readily for quite a long time and i have a contrast in quote here i love

from the nobel prize lecture of tumor virologist (indiscernible), who makes conflicting statement, he goes what can be the nature of the generality of neoplastic changes for the step like alterations they frequently undergo? a favorite explanation is

oncogenes cause alterations in the genes of the body, somatic mutation these are termed but numerous facts when taken together deseesively excludes this juxtaposition. there you have it, 64 years after boveri strong opposition by powerful people in the

scientific community, so that's an interesting juxtaposition. but i can tell you since 1966 using simpler and gradually more advanced technologies, that boveri was proven right over and over again, we have a huge amount of data supporting this concept that cancer is basically

a disease of the genome, many aspects to it that may you need to capture from the genome, but it certainly a disease of the so the way i put it is i feel comfortable now taking it as ax owematic that alterations in the cancer genome substantially determine the malignant

phenotype and characterizing these will hopefully yield correspondingly substantial incites to tumor biology with clinical implication. so just to give you two pictures which i think translates into something that anybody can relate to, so here we think of

genome sequencing as being a modern technology but in fact all the whole genome scan was the chromosomal karyotype. so here you see the normal 46 human chromosomes, the 22 pairs of autosomes and sex chromosomes looking all lined up in itself from normal tissue.

what happens if we look at a typical adult carcinoma, might be this where you see all sorts of terrible things. chromosomes that obviously are structurally abnormal or even forming rings. some that even geneticists can't tell by looking at them.

some missing because they're scrambled to the other chromosomes so you can say this genome is really screwed up, abnormal, got a lot of change, so the challenge has been out of all of this to figure out what's important in the cancer because there's some stochasticity to

the process, there's no doubt about that. so that then we come to the field of cancer genomics and instead of looking at the normal genome we look at the same variables with technologies that are suitsable for seth studying the cancer genome, i don't have

enough time the walk through all of it but i will take a couple of examples, gene expression. dna methylation and sequence variation is something that i can talk about today. and i'm going to keep this general, i'll give examples from very specific diseases that i

have worked on because i think they're fun to try to understand. so the first question we faced in the early genome era was can cancer types by gene expression profile, this seems terribly obvious now but it wasn't at that point so are all the genes

expressed in the same way every cancer or are they different? and we started working on this using array technology. and i will show you a couple of slides from a study we did in 2001, which was one of the first to classify similar looking tumors, so these four types of

cancers, all look about the same under the microscope and they're frequently viewed. and (indiscernible) a fellow in my lab at that time decided to set about separating these technically called small round glial cell tumors using a microarray, and this is just

shows the result of that kind of analysis. the separation here between different tumor types shows their pattern of gene that's four different tumor types, a kristin barrett in the lab, later looked at 11 tumor types and here what she's done

is for all the tumors, labeled at top here, identified the genes which most characteristically define each of those tumor types. what was exciting and misleading at the time, we look at one of the tumors and i will to cuts on that for a few slides,

gastrointestinal stromal tumor. when you look at just what's in that list under the column headed this here, if we look at the genes that are in that list, we see the top one is the kit oncogene which turns out to be both a diagnostic marker and ultimately a therapeutic target

for most that occur in adults. right at the top of our weighted gene list of the genes that most specifically characterize. in fact, all sorts of good-bye markers have come out of these -- good bio markers that come out. p p p

floss one hope the therapeutic targets as well. that's turned out to be challenging because it isn't easy specific genes just when you want. so think of it as a possible target. a lot of implications from this

type of work which is many thousands of publications in the literature now, probably most importantly it's a powerful tool for class discovery which still leads to recognition of molecularly defined subclasses of tumors. the data can be used to develop

diagnostic and prognostic clinical tests and you can identify individual genes redsly trank transitioned to clinical assays such as immunohistochemistry, flow cytometry or rt pcr. so i will talk about gastrointestinal stromal tumor

and how we have been able to further fractionnate these on unusual subsets with integrated genomics. for those who haven't heard of this tumor they're derived from interstitial cell a pacemaker cell in the lining of the gut. so this is what causes sin crow

in thisty of peristall tick contractions in the gastrointestinal tract. it's been a model for targeted therapy because as i said, most of the adult cancers have activating mutations in tyrosine kinase, if they don't they have one in another kinase edgfra and

impressive clinical results are found in some patients. we were interested in not all these varied mutations. particularly we got interested in what we call pediatric and wild type, they didn't have those mutations so there's a subset of patients that lack the

egfr mutation, they're often pediatric. and hereditary, so we help set up within w a number of colleagues and collaborators, interdisciplinary clinic here which brought patients to nih to study the -- this rare disease. interestingly they're mainly

gastric tumors. hereditary cases for recognized pretty quickly they can be associated with other tumors, particularly paragangliomas which are tumors of the peripheral nervous system. and they may be caused by mutations in genes which encode

a enzyme and energy metabolism that occurs in the mitochondrion called succinate dehydrogenat. some cases a triad or patients that both just have a previous position to just periganglioma and rare tumors called pulmonary chondromas. so we sets up this clinic at nih

to try to study these patients specifically. this just shows a slide from a terrific pathologist, the clinic is fun to participate in because we had a great team of experts, here at nih and mark is an expert pathologist. what this slide really shows is

the disappearance of sdhv staining in a tumor that's std deficient. this is the simple diagnostic test to recognize, most of these now, lets me introduce the concept of epigenetics so genetics is information that's transmitted from parent to child

through the dna code. epigenetics is something that is the property of particular cells based on their lineage development and this plot from wattington, a clever guy, i love the illustration, easiest to understand, what this is, imagine this is a cell here

occuring in the state of a fertilized egg, it's pluripotent, it could develop to any tissue or organ in the body but as development pro seeds, it finds its way towards the specific valley and which may continue to branch into the different cell types that are

found in the human genome. these have different epigenetic states based on modifications of dna and chromatin. and modifications ultimately in gene expression. so we decided the look at that angle, the dna methylation angle as well as the dna sequence

angle, put these things together with the gene expression in our clinic patient. so i'll show you what we see here, if you look at the left panel first, each of these little spots is the -- basically represents the epigenetic measured by dna methylation.

you will see two clumps of spots, these down here at the bottom, and these at the top. these are all just tumors and what we found, these are all the tumors that have mutations in one of the four sdh subunit genes, and these are all tumors that have mutations in one of

the signal transduction pathways genes which would be cdgfra, braf, nf-1 neurofibromatosis gene. they're completely different. this was kind of a shock. i honestly was not expecting, this is the most clean binary separation within a closely

related disease group that i have ever encountserred so far. these are tremendously different. immediately we can diagnose n molecular level in a patient with the sth deficiency based on this phenomenon. this is just a two dimensional

heat map which shows the stains and shows it primarily what we see is hypermethylation. so red means more dna methylation, each represents a different point in the genome. although there's a little bit of loss of methylation in some places, mainly hypermethylation

phenomenon. this is showing the two subtypes of tumors which we're now trying to diagnose, and i will show here what that group of tumors with hypermethylation look like compared to normal tissue. the point to make here is that these different color blobs

represent different normal tissue and they cluster closely together. no tumor looks exactly like a normal tissue when you look at the epigenome, they're always they resemble no normal cell in the body. this is absolutely fascinating

thing. so keith a terrific staff clinician in the lab did this work but we were left puzzled. as we were doing this we were developing next generation sequencing technology so that we can sequence all these genes at one go on -- the new kinds of

sequencers, we figured out almost all of them the mutation was, in almost every one of these but we were left with fairly good sized faction of this group which there were about a third, not quite a third of the cases which we could find no mutation for any genes we

knew were connected to this pathway. so we did something different. and we decided could we split this group of patients even further. and i'll show you how we did that in a moment. in terms of the mechanism of

this, into say we don't really know for sure, it's hard to study these telomeres which we have no good model systems in cell lines or animals. but this shows here on the left the sdh tetramer embedded in the inner mitochondrial membrane, its main job biochemically is to

convince -- to convert succinate to fumarate in the krebs cycle, and it turns outs there's an enzyme and there are several enzymes related to epigenetic which use -- produce succinate as a by-product and one is debt 2 because it's known to be a true tumor suppressor genome.

it's immuno. and what happens is succinate builds up when you don't have a functioning sth complex in the cell, and it poisons debt 2 and blocks reaction which is the conversion of cytosol 5 hydroxymethyl site scene for maintenance demethylation.

so we think methyl marks bill up in the cell from this biochemical perturbation. so our conclusion at this points was these which i call sdh in any four subunits so these sdh x mutant tumors have deleterious mutation and one of these genes, they have frequent loss of

hetero zygosity of the second allele so they have no functional sdh complex at all. and which by the way, amazing thing, cell doesn't just die but activated to develop into a they have a pervasive remodeling of the epigenome that i showed you.

and they have relatively few copy number changes. they're simple compared to other cancers. so as i said, we found that most of the sdh deficient tumor were classify by dna sequencing but there was this group with no molecular diagnosis and we

were trying to find out what's going on with these patients, resequenced and looked manually at every bit of data to figure out and we did manage to find subtle mutations in a couple of cases but most we couldn't find anything. so we decided to look at the

global level again, and we did in this case a differential methylation analysis on the dna and gene expression data to -- we have had enough samples of these skh wild types so ones we can find no mutation to those which had mutation. and we have gotten amazing

results. asbestos, really the reason we can find it is we have the right patient to look at. we did this, i'll show you a plot we made up for this setting, we call janice -- it looks two ways, like methylation and expression of genes at the

same time. so here is a paper that we just published, and what we have found is looking across the whole genome there was one point in the genome that showed increase in methylation, which is thought to be in many cases a silencing mark.

slows down gene expression. and in this case, we see at the same point in the genome a loss of gene expression. so we have exactly the signature that you would find in a very focal dna methylation which silences a gene, anybody care to guess what kind of gene it is?

this was i guess one of those things that you might have guessed ahead of time, it turns out to be right on top of the sdh, the gene. so you expand from this point and see this region of locally increased methylation is on top of the sth gene as promoter end

of the gene and spreading into the gene. when we looked carefully at the gene expression data, we found indeed convincing evidence and that's shown here in the -- these are different probes, commercial microarray. that show loss of fdhd

expression in a series of samples that shows hypermethylation. what was fascinating, we found this was essentially in every patient who in whom we could not find dna level mutation, so we term this an epimutation, and so we were able now to make a

definitive diagnosis of what was going on at the molecular level in now almost of the patients we have seen in this clinic. different diagnoses that maybe clinically rather different how they evolve. in these closely related tumors. so this is splitting to the

inth degree. but we found that in the tumors which left a mutation in sth subunit, s thc is typically inactivated by what we term an epimutation. so we don't understand the biochemistry of this, why it happens, in detail but it's

clearly pushing these cells down the same final common pathway of pathogenesis which loss of fth function. then finally we did also realize these patients with sdh mutation are also the ones susceptible to other related cancers paraganglioma and chondroma and

this is the cause of the triad that most likely will require some time to validate. so i will switch gears and talk about protein alteration mutations in cancer in engine. and the problem with cancer we have a ton of data, even has heard of the tcga project

literally thousands upon thousands of tumors through that project and other international projects sequenced, there's a huge amount of data available at this points so things we thought impossible are routine and massive quantities of data. one thing we have learned is

tumors vary in terms of number of protein altering mutations. from tumors that are exposed to carcinogens such as uv exposure or cigarette smoke or the really thousands of mutations, whereas pediatric tumors as a rule contain less than 20 mutations. there are some cancers that

appear to have only one or two mutations and post line ones we can't find anything yet. i just gave an example where we couldn't find anything. by sequencing of the dna alone. so there is a huge variety in complexity butly tell you a story of one simple example many

of you may have heard about, that's chronic myelogenous leukemia. so going back to the looking at the chromosomes themselves in 1960, nolan hungerford found tumor as chronic leukemia that evolves slowly, they all carried unusual chromosome, by just staining the

chromosome, they called it the philadelphia chromosome, they worked in philadelphia, this was before (inaudible) was invented or perfected. and they couldn't tell what it was and then janet rowl, university of chicago in '73 figured it was a translocation

that is a fusion which joined two chromosome, nine and 21, then gerard grossfeld in mid 80s showed this was a fusion gene, the two break points created a chimeric gene that never occurs in any no mall cell, cancer specific gene that fused two genes one called bcr

and one called able, able is a tyrosine kinase. this information was exciting because it gave us the specific molecular pathogenesis for rare but important leukemia. that information sat around until 1996, when brian druker who made this his cause, said look if

it's caused by the disease is caused by a kinase there should be a drug that lost that kinase. and he worked to come up with collections of drugs about he found one, that actually is effective in inhibiting the bca able fusion onco protein and the result is that that arise in

treatment and patients with resistance and so on but it's transformed this field. so the reason i go through this in some detail is well, the cml to assign a paradigm for cancer cure, a lot of people thought that many of us really desperately hoped that that

would be the case. so what is that paradigm? how would we know when we're encountering it? are there other examples? one specific driver gene need ford the cancer cell survival growth of the tumor, you need a therapeutic agent which targets

that gene. simple. so to put it in a graphical format, the idea specific tumor biology to determine the outcome of the leukemia, we're go tag make a therapeutic intervention to prevent that biology from expressing itself.

so of course then and 90 days when these are starting we knew all cancers wouldn't have the same driver mutation that we have to come up with different kinds of biology and biology specific intervention, so bcr able might be one, egfr mutation epidermal growth factor receptor

might be another. that was the simple idea. well, have we been successful in doing this? i can definitely point to success in some of the more important ones are listed here. there are many more in various specific situations.

so there are other things that are close to the cml situation. but this comes with buts an ands so the but is move common adult cancer clml like results are now achievable. we know when a drug l target gene is present so something special about this leukemia

case, that makes it easier to treat. a large proportion of human cancers as this is the more depressing aspect. a large proportion of cancers do not have a gene activated by mutation. which fits the cml paradigm.

that's a tougher problem to think how you're getting around. so in addition to not getting terrific responses with single agents, lots of instances of resistance developing by many different mechanism, mutations in the target gene activation and alternative pathways, you

name it. this is now a relatively sad sobering saga. but there is an end to this story. first of all, many of the altered genes successfully targeted in cancers whether frequently are also observed at

low frequency in diverse other cancers and/or rare cancers. so this gets back to the point i made at the very beginning. so known targetable mutations to useful as molecular marketers to guide precision therapy out of the disease context which there were originally validated.

the biggest most frequent example of this is the braf gene which is kinase that's activated frequently in melanoma, it turns outs the exact same active -- activating mutation which is a b 600 e br abf occurs in lots of kinds of cancer. some of these there's reason to

think that braf targeted therapy may also be effective. so this is where we are go what cuts across lots of different tumors and goes down histologic cancer. so one parting shot here, so as we have been sequencing more and more tumors, it's now become

practical to sequence dna from small clinical samples quickly and accurately. you can do it on any scale, whether from the grand scale whole genome to individual genes. sequencing of a panel of so called driver genes which could

be targeted for therapy or which would aid definitive diagnosis itself clinically accessible and routine. the broad utility of tumor dna sequencing is currently a subject of ongoing clinical research but sequencing is standard of care in specific

clinical situation. i want to make a quick comparison of this model tumor progression which is an old model that start off with normal cell, which then gets excessive mutations in genes until you finally get invasive cancer. we know now this linear way of

thinking about cancer evolution is not really correct. a better way to think of it is in terms of branching polyclonal disease that's the normal cell, tumors pick up changes and bifurcate in the clone some which may die out. here is in blue is one that

becomes dominant in tumor diagnosed at the tumor grows, there are some other secondary clones which are actually not that rare. if you treat the patient you may find the tumor recurs and that a clone is different when it comes this concept which is not that

new, i think, is turning out to be actually really important. why? because the clonal heterogeneity of tumors can be quantitated really well by tumor genome sequencing. this is best achieved with leukemia, but i'm willing to bet

it will happen in other cancers as well. this is my last slide, and i'll conclude with this, this is my vision of the future of the molecular characterization of which is that at the time of tumor biopsy when a patient presents to the oncologist, in

addition to the usual things that i showed you that we were doing 100 or more years ago, immunostaining, will also be gathering nucleic acids for tumor normal genome sequencing. rna sequencing. methyl sequencing. and from that we'll get all this

information accurate classification. we'll sound outlier gene expression good for targeting one way or another, we'll find out about genome amplification, small mutation, we'll learn about neoantigen which is a cool topic now.

for antigens that never exist in nature but occurs in tumor cell, all the other things i talked about translocation, we'll try to be able to predict pathway. and carry out this clonality analysis which will enable us to -- not cancer as something you have got, the day it's diagnosed

but as a process, that's going to change as patients live longer with the tumor. so at recurrence, this aspect of this diagnostic process will be repeated. the tumor is re-evaluate and retargeted based on the act of proliferating clone at that

time. this a bit of a vision. we don't have most information to do this, it's still too costly to do but i think there's little doubt in my mind this is going to happen, already happening in the leukemia world. i think we'll see this expanding

more and more, so that's what i had to present, i'll be grad to take questions. thanks. >> thank you very much, paul. slide into the brave new world. >> when you work with cells in vitro, it's clear you can sequence because they all

identical clonal. but now you take a tumor which has stromal cells infiltrating lymphocytes, and the endothelial cells and tumor cells which could be highly heterogeneous. my question is how are you sorting out which signal is coming from which type of cell

and how you establish signature. >> right. so -- >> i have second question. >> just really quickly, you're asked a question, it opens a big world of other questions. so what i can say is certain things you can tease apart so

that in fact we do this routinely for aspects of the signature of normal cells that are mixed in. that's one thing, at least the gene expression level. at the mutation level, we try to estimate the percentage of tumor cells and you can overcome that

heterogeneity by sequencing more. the normal cells generally do not carry mutation of burden. mutation of burden isn't zero but it's easy to spot. and so there are some strategies, a bigger problem i think than local heterogeneity

or microscopic is the heterogeneity geographically in a tumor that's large. no reason to assume that a tumor here is identical to the tumor thathe few centimeter ace way. that's a big challenge. we know that's true already for some tumors, that's just a

relativery new field. >> my second question is the -- is there any genetic signature on mesotheliomas? because the first speaker just mentioned bat 1 protein involved but what about other changes? what is known? >> i think the third thing is to

say we're work on that. and we don't have all the answers yet but there are some genes that are going to be mutated in mesothelioma but we're early stage. >> mesothelioma, very interested in looking at young patients with mesothelioma, you have less

than 20 which frequently see young kids with the mesothelioma so that might hold water. >> might be important. >> >> p-53 is very uncommon in mesothelioma but we see i-16 present in 80%. then we talked about at that

point. nf 2. >> regarding the ss-1p treatment, what is the immune reconstitution like for these patients? do they receive trans-- did you receive transplant afterwards? >> the patients get leukemia but

no e decrease in neutraphil and lymphocytes recover within about -- it's variable, how much treatment they get but three to six months their lymphocyte count is normal there.s no transplant. it's just a lymphodepletion but lymphocyte recovers on their

own. >> i think that in fairness -- in fact there already have been commentaries on this, that paper which is a very nice paper for giving the overview of the disease, he didn't really say that was bad luck that was the press translating that paper in

to for the -- manage for the public consumption. because it gives the misimpression because you make your luck. to some extent with things like what cancer risk factors you expose yourself to notoriously, like smoking, lots of other

things. so there is a clearly a stochastic element to this but it's one that can be affected by genes, it can be effected by lifestyle. so you have to be -- there's luck and there's luck. that's a complicated topic to

simplify, well it's just bad luck. floss you maybe a smoker because you have a gene that predisposes you to smoke, that's bad luck. >> i just wanted to tack on a question to her question earlier, have you had any patients that were already

immunocompromised hiv positive patients for example? >> in general -- so the question was did we have patients with immunocomemize because of hiv? but in general for this is clinical trials we exclude patients because we worry if we further (inaudible)

>> we heard -- >> we have been hearing a great deal in the past five years or more about the cancer genome project in which large number of cancers are being sequenced. can you tell us what the general status of that project is and what have wearerred from it?

>> let's see. well the general status is that funding for that project is being concluded. and the projects are being published. there's already a huge number of papers that have come out. most of the major diseases have

publication, one of the other nice things is the data is available publicly, the process data and raw data is also available >> there's a few more hoops that we have to go through. so this is the kind of data you really can't squeeze everything

out of it in that first overview paper. so there's a lot left to discover. what i would say is that there are a number of instances where diseases that were considered to be one is now clearly split into subtypes this is particularly

true in some of the brain tumors, really good example is medullar blastoma which now can be broken into several subgroups based on gene expression profile which is related to mutational profile. there are many such instances i would say the other thing that

comes out is that when you find a gene that is unknown, to be mutated in cancer and you find it at 15 or 20% of some cancer you take it much more seriously when mutated at 1% or .5% in another cancer so that is tending to build these bridges of thought across the whole

field which will talk tame for us to sort -- time for us to sort out. we don't know how this will add up. in terms of potential therapeutic options. or the opposite knowing that a therapy won't work, sometime it

is presence of mutation is a good example being -- >> that's an ongoing project. >> so that's ongoing. from z >> all right. any more questions? if not thank you both very much for very exciting talks.