Childhood Cancers

louis staudt:so today i'd like to give you an overview, not just of my work but also of much workby others in -- whoops, i just broke the podium, thank you. so this is an exciting time forus in cancer research because, as i'll get into, we have tools that are just jaw-droppingin their power to examine the human genome. and i've been using these for quite a fewyears to study lymphoma, as mentioned. i also now have two jobs, two full-time jobs, theother one being directing all the large genomic efforts that the national cancer instituteis conducting right now. and these, then, hope to do what i've been doing for lymphomafor all cancers and do it on a grand scale that will lift the entire field up.

so i'll begin today's talk giving that overviewbased on what's going on out there, and then i will focus in on my own research and showyou some more in-depth studies in how we think about the genetics of cancer. but the broadtake-home message is that the current ways that we're diagnosing cancer and treatingcancer are changing before our eyes, and we're going from a histological diagnosis to a moleculardiagnosis of cancer as fast as we can. so when i speak of genomics, what i'm talkingabout are the semantic changes that occur in the cancer cell. not in the germline ofthe individual, although that contributes, but what i'm going to be talking about todayare the alterations in the tumor cell that make it different from a normal cell. and,of course, this begins with mutations that

sometimes alter the structure of proteins.there are structural rearrangements of the genome, either copy -- dna copy gains or losses,or sometimes translocations of the genome. there are -- there's a need to look at theactive part of the genome that is expressed as messenger rna. that can be dysregulatedin cancer. there are a class of small regularity rnas called micro rnas that are importantto know about, and then there are modifications of dna such as dna methylation that lead usinto insights into the abnormalities of cancer. and in the latter part of the talk, i'll talkabout -- these i would class as "structural genomics," and then there's another classof studies called "functional genomics" where we aim to find out which genes are importantby directly manipulating those genes and seeing

what happens. the tool that all of us are relying on noware these so-called next-generation sequencers. here's one from illumina. and what we took,"we" being the broad genomics community, took more than a decade to finish one human genome.now two human genomes can be done in 11 days of sequencing on this one machine. so thisis an outpouring of data that obviously requires a lot of computational power to make senseout of. and this is courtesy of the genome institute, this is one of their more commonly-shownslides, i'm sure, showing that here is the drop of cost in computing, i'm sure you'veall heard of moore's law, and here's the drop in the cost of sequencing the genome. so it'splummeting faster than moore's law, and we're

now somewhere around a $4,000 genome and wehope to be soon at a $1,000 genome, really making it very personal for all of us verysoon. so, as i mentioned, i'm directing these programsin genomics for the nci now, and they include something called the cancer genome atlas,which is looking at adult cancers very deeply, there's a program looking at pediatric cancers,and then there's a functional genomics program. and so first i'll tell you a little bit aboutrecent progress in this tcga program from structural genomics of cancer. and the goalhere was to study a large number of cancers because we believe that there's this greatdiversity of human cancer, and until we see that in all of its glory, we won't know howto understand any individual patient in what's

going on. so here are the tumors that we'vestudied, and by the end of the year, we will have brought into the program 10,000 tumorsthat will then be finished in their analysis about two years later. and the first thing we have realized is somethingwe knew before, but is something we deal with now, that the mutational burden in canceris very, very high, and some of this is, of course, self-inflicted. we have carcinogensof various varieties: smoking, sunlight, possibly food that we eat, leading to very high ratesin mutation in, for example, lung cancer, melanoma, and stomach cancer way on this side,whereas on the other side, you have the childhood cancers, where very little action is goingon. so out of this rather noisy picture, we

have to discern the biological signals. someof this is just noise, and some of it has lead to changes that make the cancer cellmalignant. so a recent study that was published on tcgawas about kidney cancer, renal cell, clear cell, carcinoma. and in this cancer, thisis a busy slide, but it shows you that cancer information is busy and rich, and it focuseson what was found to be a recurrent signaling -- abnormalities in a signaling pathway knownas the pi3 kinase pathway that promotes the survival and metabolism of cells. and thefirst important point is to say it's not one abnormality, but all -- a large number ofgenes were either mutated, amplified, or deleted such that when you put them in a pathway diagram,you can see that in small numbers of patients,

there were mutations somewhere in the pathwayleading to a grand total of 28 percent of all of this cancer type having some abnormalityin this pathway and some of the point mutations in a key kinase called mtor that -- in thispathway are shown here and they cluster in particular parts of the proteins. an important point is that from a clinicalpoint of view, a clinical study had already been done with a drug that targets mtor, knownas everolimus, and you'll note that, in fact, in clear cell carcinoma, there was a signal.there were responders. it's not a home run, we're not curing anyone, but the drug works,and it's probably because this type of cancer relies and has turned on this signaling pathway,which is -- and, of course, now we have to

figure out, well, who are these respondersto this drug? did they have these mutations whereas the other ones did not? now, this is mislabeled. this is the studyof -- recent study of endometrial cancer that the tcga did. and these are about 500 samplesfrom different patients, and then you classify them by their molecular abnormalities in avariety of ways that i won't go into. but the categories are fairly self-explanatory.there's what is called an ultra-mutated type of endometrial cancer with lots and lots ofmutations due to a defect in a repair enzyme. then there's a hyper-mutated form that's adefect in a different repair enzyme, the mismatched repair enzyme. and then there's two otherforms that are called copy number low and

copy number high, meaning that there wereeither extra amplifications of dna here or not as many events over here. now if you look at this classification withrespect to -- you know, that's not shown here yet. i'll get to the next slide. so here -- sothis leads us to the classification. what's now shown is, how does this relate to thehistology as a pathologist would normally do it? and the pathologist sees there's anendometrioid type of this cancer and a serous-like type cancer. and that's in either light blueor dark blue. and you'll note that in this copy number lowand these others, mostly it's the endometrioids, so things are lining up okay. but here, wehave a problem. here there's a mixture of

histological diagnoses within one of thesecategories. and, in fact, it would look like there's some things that are really serousby their molecular profile, but the pathologist call an endometrial. do we care about this?well, yes, we do, because endometrioid has a better prognosis and is treated differentlyfrom the serous type. it is given adjuvant therapy with radio therapy, is along withthe surgery, whereas this more difficult to treat serous type, chemotherapy is added tosurgery. where if you look at the survival of the patients classified in this way, you'llsee that, indeed, this serous type in the red curve has the worst outcome, this copynumber low has an intermediate outcome, and this ultra-mutated type at the end, thesefolks do incredibly well. so you can say maybe

all they need is surgery and nothing else.but the important point, again, is that if we stuck to the histological prognosis, it'spossible that some patients who really had this molecular serous type would've have beencalled endometrioid and maybe gotten the less aggressive therapy and not done as well asif they had gotten chemotherapy. so now can we find new therapies for our patientsbased on structural genomics? and here i'll highlight just one example from the pediatrictarget study that we've been looking at a large number of these childhood cancers shownhere. and this was a 10-year-old boy, and he had a refectory form of b-cell acute lymphoblasticleukemia. of course it's pretty treatable, except when it's not, and in this case, itwasn't. and rna sequencing was done, another

way to look at the rna, not the dna, but nonethelessit showed that this boy's tumor had a translocation that involved a receptor tyrosine kinase knownas the pdgf receptor. so, fortunately, there's a drug for this, and it's called imatinib.you may have used it or heard about it as a treatment for chronic myelogenous leukemia.but it also is a very good inhibitor of the pdgf receptor. and he started this treatment,got immediate clinical improvement. within one week there was a morphological remissionin the blood, and by two weeks, there was essentially no molecularly detected disease,and that's shown here in this pcr assay. before treatment, this is the translocation mrna,and on treatment, you just can't detect it anymore.

now this is a very new unpublished work, sowe don't know how this individual is doing long term, but it's a dramatic response. andit's this kind of in-of-one experience, anecdotal experience, that we've been seeing with someof these targeted agents that have gotten a lot of people excited. but, of course, anecdotesare not how we do good clinical medicine, so the cancer institute is trying to do thismore seriously. and essentially the problem is how we find the right drug for the rightpatient. and there's two ways to approach that. the first is in this genotype to phenotype.that is, i discover that this genus has this mutation, i got a drug i think is good forthat aberrant protein made by the mutation, and let's see how it works.

and we're not going to do this in just onepatient; we're going to do this in 1,000 patients. and to find these thousand patients that havesome targetable molecular abnormality, we'll screen about 1,500 to 3,000 of them in clialabs to get these molecular genetic changes. we've gotten drugs over 30 of them committedby pharmaceutical companies that made target one or the other of these mutant proteins,and they are going to give them to the nci for this trial, and then we are going to assignthis sort of algorithm, if you have this abnormality or the other, you get drug a, these otherabnormalities, you get drug b, and so on. and then we're going to see how often it isthat we can, as i say, predict -- read the tea leaves, predict a response. is this asuccessful approach? and then we'll go through

the clinical trial design. the second way to look for the right drugand the right person is so-called phenotype to genotype. here the doctor is the investigator.the doctor sees this remarkable response in one individual that the same drug in otherindividuals, seemingly of the same cancer type, did not respond. and so these are theso-called exceptional responders. and here's one celebrated case from memorial sloan kettering,a 73-year-old gentleman, and i think it's -- actually, i don't know that -- a 73-year-oldwith bladder cancer, and this person had been put on, failing a lot of other therapies,had put on the same drug that i introduced to you, this everolimus that targeted themtor protein, and had a remarkable complete

response, and it was -- it's been quite durable,no evidence of disease 24 months after starting the therapy. and this is just abominable scanshowing large clusters of tumor masses that then resolved and are still gone at 18 months. so why is it that this one patient had a response,whereas the clinical trial that he was in was declared a failed trial? the statisticalendpoint was not met. well, whole genome sequencing was done on this patient, and here's a waywe depict all the molecular abnormalities where all the chromosomes are arrayed in acircle, 1 through 22 x and y, and then these various marks refer to where there are abnormalities,and some of them you can't see. these little dots are point mutations in the genome. andi've circled one green dot which indicates

a frameshift mutation in a tourist suppressorgene called tsc1. and tsc1 is a negative regular of this mtor signaling complex. so when youremove the brake on this complex, you get spontaneous signaling, and that is presumablywhy this patient has such a good response. now, with this knowledge, if you go back tothe clinical trial, and look at the response of patients on this trial and ask whetherthere were mutations in this tsc1 disease gene, you find here this big -- this is thedepiction of the tumor volume, so going over to the left is a loss of tumor volume. thisis the patient that we've been talking about with the remarkable response. but here's someother partial responses, or actually didn't even make it into partial response, but theyare the ones that have mutations, whereas

the ones that don't have mutation, all thosetumors grew, so this has now sparked more work, there's another clinical trial to evaluatewhether this is indeed true. and so to capitalize on this sort of clinicalobservation, we're encouraging physicians from all over the country, all over the worldfor that matter, to send us your exceptional responders. that is, we would like to studythese unusual responses if the patient is willing to participate, have their genomesequenced, and then see how often we can figure out the puzzle of why we've had such a clinicalresponse. so in the last part of the talk, then, i'lltell you about how we have approached a particular really aggressive type of lymphoma known asdiffuse large b-cell lymphoma.

and i will introduce first the idea of functionalgenomics, and we depict this by achilles here being felled by the arrow to his heel, andi modified this to include, on his shield, a histological picture of diffuse large b-celllymphoma. and the metaphor is so rich i can't believe to tell you, but if you remember theiliad, then, the shield of achilles, which is described in several pages in that book,basically, if you read and understood the shield of achilles, you would know how theworld worked at that time. there were depictions of the various cities, and there were depictionsof wars, and there were mountains, and there are -- and rivers and oceans. so same way,we feel that if we can understand, in a molecular way, this cancer, then we will find this vulnerabilitythat we can target therapeutically.

so the way we do this is do a genetic screen.of course, genetic screens have been done since mendel and famous people studying fruitflies, and we just couldn't do that with the cancer cells because, of course, we can'tbreed our organism of choice, the human, at will. but we had new techniques that allowedus to do genetics without heredity, in a way. and what we do is we have tools that we callsmall hairpin rnas that are able to surgically inactivate a particular one gene at a time.and in so doing, we can see what that gene does for the cancer cell, and that's depictedhere. so we have a library of these shrna vectors, each one in different color targetinga different one of the 24,000 human genes. and we introduce that library into a cancercell line, so you can see that every cell

growing in this flask in my laboratory wouldhave a different shrna and therefore have inactivated a different gene. so this is the beginning of a genetic experiment.you have genetic diversity in the population of the cells. you now apply a selective pressure.and the selective pressure is simply, "can any cell in that flask grow, proliferate,and survive for three weeks in culture?" so you'll note that in this population of cells,one is missing, right? and presumably that is because this shrna, this turquoise shrna,knocked out an essential gene in that cancerous cell that it cannot live without, and therefore,it is depleted from the population. and so then all we have to do is compare the populationof shrna vectors at the end of the experiment

to that at the beginning of the experimentand figure out what the essential genes for proliferation survival are. so these are ourso-called achilles heel screenings. important point to make is that it's not atodds in any way with the structural genomics i just got done telling you about. in fact,what we find over and over again is when we perform an achilles heel screen and then lookwithin the pathway we've -- this essential pathway, somewhere in that pathway is a mutationthat tells us why that pathway was active. and so this is the goal, finding these essentialcancer pathways. so diffuse large b-cell lymphoma, i'm sureyou all had some experience with it in one way or the other, is the most common typeof non-hodgkin's lymphoma. it's aggressive

but obviously can be treated, and the curerate currently with chop chemotherapy regimen plus rituximab is on the order of 50 percent.so, great, 50 percent, much better than other solid cancers, but we're trying to figureout something to do for the other 50 percent. and this leaves us with a large number ofdeaths every year, 10,000 in the united states alone. and so we thought that perhaps the problemwas that by looking under the microscope and describing the cells as large and diffusein spread, maybe that was just not sufficiently accurate to perceive the differences thatwere leading to different responses to chemotherapy. so we used molecular profiling to find sub-typesof diffuse large b-cell lymphoma. and the

molecular profile and the available technologyto me, 10 years, 13 year ago, was not this fancy next-gen sequencing, but rather abilityto profile the levels of messenger rnas of various genes within cell, so-called geneexpression profiling. and the way we usually depict this is if agene is very active and there's a lot of messenger rna, tumor is given a red color. and if there'svery little expression in the gene, you get a green color, and black is somewhere in between.and then what you see in this diagram are lymphoma biopsies, probably a couple 100 ofthem, each in a different column from a different patient, and then we have a number of genes,probably about 200 genes here, and you can see that in tumors of this variety, whichwe call the activated b-cell, or abc, they

have this signature of genes being very active,and these genes being inactive, and just the opposite in the germinal center type, gcb,you have a different signature of gene activity. and these different activities didn't justarise from nowhere. they're actually a signature of the type of normal b-cell from whence thislymphoma derived. so, as i mentioned, one of our goals was tofigure out which of our patients were doing well with chemotherapy and which weren't,and it certainly is the case that these abc tumors are the bad actors. here, we're onlycuring at best about 40 percent of the patients with r-chop chemotherapy, whereas the germinalcenter type is coming in at about 75 percent. so we're not done worrying about the germinalcenter type, but we've spent most of our work

in the lab on the activated type. so how do you know that this is reality? well,that is, maybe there's an infinite number of ways to use all this complex data to divideup tumors, and why is this the right way to divide them up? well, i've shown you thatthere were some clinical differences, so that's one thing to hang your hat on. the other thingto hang your hat on is that then you look in those tumors classified as the abc or gcband ask what are the genetic abnormalities; which mutations, which deletions amplificationsof genes? which structural genomic changes are found in one type or the other? and whatyou find is that all of the abnormalities shown on the left side here are only foundin the activated b-cell type, where as all

these others on the right side are only foundin the germinal center type. so this yin-yang appearance tells you that these are pathogenicallydistinct diseases. and why are they distinct? as i mentioned,these tumors come from a different starting point, a different normal b-cell, and probablybecause of that different start point, then it has to take different steps to become malignant.and the way that we think about how this particular abc malignancy arises is boiled down to avery simple molecular wiring diagram. what we noticed very early on was that there wasone signaling pathway within the malignant cells known as the nf-?b pathway, which wasseemingly stuck on in the on position all these tumors. and this is a key pathway thatin any cell type will prevent cell death.

so, of course, that's one of the hallmarksof cancer, that the cells don't die appropriately. now nf-?b does another important thing toa b-lymphocyte. it turns on a transcription factor called irf4 which propels that b-celldown the differentiation pathway to becoming a plasma cell. and if it went all the waydown to the end, then that might be the substrate of multiple myeloma, a plasmacytic neoplasm,but it wouldn't be this abc diffuse lymphoma. so instead there's a second block having todo with a loss of other transcription factor known as blimp1. and these cells pile up ina differentiation stage that's normally only transiently present. so this plasmablasticstage is probably no more than 12 hours in the life of a b-cell. but perhaps now thisabnormal accumulation in this stage allows

for further oncogenic changes to lead to thefull malignancy. what we wanted to understand, though, waswhy this nf-?b pathway that was promoting the abnormal survival of these malignant cells-- what was the cause of this? normally, this is turned on in the b lymphocyte in responseto exposure to an antigen in the environment and it gets the cell activated. but we didn'tknow that there was any antigen here, so this pathway was on all the time; why was that?and, in particular, what upstream signaling pathways lead into the key regulatory kinasein this pathway known as i?b kinase that turns on nf-?b. and what we discovered, which i'lltake you through in a step-by-step fashion, is that it is the b-cell receptor signalingpathway that is important. and the b-cell

receptor pathway, of course, is arguably -- b-cellreceptor is arguably the most important receptor to a b-cell. without it, you can't even makea b-cell from a bone marrow progenitor cell. you have no b-cells. and then even if youhave those b-cells when you were exposed to an antigen, a pathogen, a virus, there wouldbe no response because it's the receptor for those antigens. this is the key receptor,and it does many, many things for the b-cell that promote its survival and proliferation. so what we in discovered in our genetic screen,initially, was that there was a signaling complex downstream of the b-cell receptorand upstream of i?b kinase involving three proteins: card11, malt1, bcl10. and if weinactivated any one of these three proteins

by that shrna technique, then the cells died,our cell-line models of this type of lymphoma. so that was our first clue that this entirepathway, which had been worked out previously by immunologists, was important. then we started sequencing the tumors andfound that in 10 percent of the tumors of this variety, there were mutations that affectedthe card11 adaptive protein. and these mutations were really fascinating because they createdmutant proteins that spontaneously turned on the nf-?b pathway if they were put intoanother cell. so these were the oncogenes, if you will, within these cells. but notethat 90 percent of the cases did not have any mutations in card11, yet we had othercell-line models that with no mutations in

card11, but still, if i knock down card11,they die. so something was turning on wild-type card11, so we surmised there must be somesignal upstream in this pathway that was tickling card11 and turning it on. and, in fact, in our genetic screen, we hadalready observed a strong dependency of those cell lines on bruton's tyrosine kinase, whichi'm sure you remember as important to even make a b-lymphocyte. the boys with this x-linkedinherited disease don't have these cells, and suffer from infections and other things.so this is an important kinase that links the b-cell receptor to the downstream nf-?bpathway. but then we started sequencing -- oh, it must be mutations in btk, so we sequencedthat, and there were no mutations in any of

our patients. we sequenced all the other kinasesin the cascade, no mutations, and finally tumbled to the fact that maybe it's just atthe source of this river, right? maybe it's at the b-cell receptor itself. so we wentand knocked down components of the b-cell receptor and the cells cooked. so -- and turnedoff this nf-?b pathway. so these cells depend on the b-cell receptor,but why? well, in 21 percent of the tumors, we found mutations in two of the componentsof the b-cell receptor known as cd79a and cd79b. this seems to be going in and out,this microphone, sorry. and i'll take you through. these appear to be gain-of-functionmutations. and -- but they're not, perhaps, as strong as that card11 mutation i told youabout, and we'll get into that.

so how does b-cell receptors signaling work?well, this is sort of the cartoon version, and this is what you would find in an immunologytextbook, that there are the two immunoglobulin chains that interact with antigen. and thenthere's a signaling chain, cd79a and b. within the cytoplasmic part of cd79a and b thereis a series of amino acids, a motif known as an i-10 motif. and this is the signalingmotif that's important. and what happens is some src family kinases phosphorylate thetyrosines there, recruit another kinase known as sic, and then a lot of downstream signalingoccurs. so mutations in cd79a and b occurred in 21percent of the patients with this abc type of lymphoma, and essentially never or werevery rare occurring in the other types of

lymphomas that we had studied. so this saysthat this type of b-cell receptor signaling is a key part of the pathogenesis of the abclymphoma and not these other lymphomas, again, highlighting the molecular differences betweenthese cell types. so what are the functional consequences ofthose mutations in the b-cell receptor? well, here's the cartoon that i showed you. andwhat we found -- the mutations, first, what were they? well, they were fascinating inthat they affected one single amino acid, this tyrosine, a proximal tyrosine in cd79aand cd79b in the vast majority of cases. and basically you could put any other amino acidin there, in one case, actually deleted that tyrosine, any of those appear to be functionallythe same type of mutation.

now at a lower frequency, we found deletionsin the other signaling component, cd79a, that surgically took out the i-10 region of cd79a.so what do these do? what are they selected for? well, the first thing -- way to understandthis is that one of the consequences of engaging the b-cell receptor with this red antigen,this foreign substance, is, like all receptors, it will then be down-modulated off the cellsurface in an endocytic recycling and thereby terminating the response. well, we discoveredthat if you have mutation in cd79b, this endocytic recycling is blocked. so the receptor simplystays at higher levels at all times on the cell service, and thereby promoting signaling. the second mechanism that we discovered hadto do with a negative feedback for b-cell

receptor signaling. so all receptors bothturn on signaling and then have ways of turning it off, because if you didn't, this wouldbe pathological. and in this particular receptor, there's a -- one of the src family kinasesknown as lyn, that phosphorylates another protein, cd22, and recruits a phosphatasethat takes off those phosphor tyrosines and turns the receptor back to its native state.and what we found, that in tumors that have a mutant form of cd79b, that kinase, althoughactive, is much less active and does not engage those negative regulatory elements. so twoways we get more b-cell receptor at the surface and it's more active. now, life was fine at that point, but themolecular biology and genetics of cancer doesn't

have to be simple. and so we had, in our geneticscreen, an equally important survival pathway that seemed to emanate from a signaling adaptorknown as myd88. and it also turns on the nf-?b pathway very strongly. well -- and in 39 percentof our tumors, we found activating mutations in myd88 that make is constituently able toturn on nf-?b. so what is myd88? it's a very famous receptor for immunologists becauseit is the adaptor, rather, for the toll-like receptors that are involved in innate immunity.they just see danger somewhere in the environment. they see a virus with a viral rna or a viraldna, or they see a lipid from a bacteria, and they alert the immune response. theseare the so-called "pattern recognition" receptors. and they utilize the signaling adaptor myd88.and they are powerful receptors, just like

the b-cell receptor that turn on the nf-?bpathway, p30 map kinase, engage a number of cytokines, il-6 and il-10, and interferon. and this is becoming -- this particular mutation-- one of the, i call it, sort of the ras oncogene for late b-cell malignances. so hereis the frequency of this mutation within various mature b-cell malignances. here's our abclymphomas coming in at 29 percent with one particular mutation i'm showing you. but there'sa primary central nervous system diffuse lymphoma that has about a 40 percent occurrence ofthis mutation, a cutaneous type of diffuse lymphoma with a 70 percent, testicular diffuselymphoma, 80 percent, and a type of indolent plasma cell neoplasm, waldenstroem's macroglobulinemia,90 percent of individuals with this type of

cancer, the tumor cells have this one particularpoint mutation. but at a lower frequency, chronic lymphocytic leukemia, which is, ofcourse, the most common leukemia, 3 percent of those cases have that same mutation. sowe need to understand what this guy is doing. so on the face of it, it would look like youhad two parallel pathways leading to nf-?b. but we had a drug, first of all, that targetedone of those two pathways, the b-cell receptive pathway. and that drug is known as ibrutinib.and so we were -- what? male speaker:what was the name? louis staudt:ibrutinib. ibrutinib. good to write this name down because we are all going to be prescribingthis drug, or hopefully not getting this drug,

but it is an unbelievable drug and it's goingto be active -- it is active -- in multiple b-cell lymphomas, but i predict will be -- haveactivity in autoimmune and inflammatory diseases. so we hooked up with a company that makesthis called pharmacyclics and forged ahead, learning a little bit about this pathway,the myd88 pathway, whether it would somehow mitigate the effect of this drug, but nonetheless,decided to go ahead with its study. the drug is fantastic in how it works. it makes a covalentbond in the active site of the kinase, and thereby inactivating that protein kinase forits lifetime. so this gives you wonderful pharmacodynamics, it gives you single dosingwhere you virtually complete kinase for 24 hours, and it gives you great seal activity.there are only 10 kinases encoded in the human

genome that have this particular cystine towhich the drug attaches in that position. so as you should be aware, these kinase inhibitors,of which there are many, many of them are multi-kinase, they have lots of differentthings. this one is more specific. male speaker:what does it do to infections? louis staudt:it doesn't seem to cause any problems with infections, but it's an excellent question.in the lab, this was highly potent in this dose response curve in killing some of ourlymphoma cell lines that rely on b-cell receptor but had no effect on some other cell linesthat don't. so we started a clinical trial across theroad in the clinical center with my colleague

wyndham wilson, to test this out, and it'ssort of an extension of the company's phase i trial. and this is a 52-year-old woman whosetumor, by molecular profiling, was of the abc type and had a mutation within the b-cellreceptors, cd79b. she had had -- all the patients i'm going to tell you about are relapsed refractorycategory. they have, on average, had three prior therapies, usually two or three. andthis woman had had two chemotherapies and had had a complete response to each, but ineach case, relapsed. and then started taking ibrutinib by mouth. it's a single white pillyou take once a day without any discernible side effects. by week eight, she went intoa complete response shown here. here are some abdominal tumors. the others are her kidneys,of course. and these abdominal tumors in this

pet scan overlaid on the ct have gone awayat week eight. and she is our star. she is our exceptional responder. she is now out3.3 years living without any sign of disease, taking this drug once a day by mouth. no infections,no discernible side effects. so this is a remarkable way to get your cancer treated.it's like a blood pressure pill. here's another lady that had -- 59-year-old-- same abc tumor; in this case, though, she did not have a mutation in her b-cell receptoror in myd88. she had had a stormy course. she had what is called primary refractorydisease where they tumor never responded once to chemotherapy. and she came to us in extremis.she had had a very rapid rise in her ldh, an indication of tumor volume in her serum.then she started getting a ibrutinib, and

you can see this dramatic fall in the ldh.here's her tumor that was there by imaging when she arrived. she had so much tumor inher abdomen that she couldn't eat because her stomach was being compressed. and here'swhat her scan looked like after three weeks. so this is a tumor that never budged withchemotherapy that seemed to melt away here. this is great, i saw her at this moment, shewas doing fine, and then about two weeks later the tumor came roaring back, and so she hadsome form of resistance. and so it's not all, you know, all roses here. it's the beginningof a new type of therapy. so to take a serious look and get away withanecdotes, we did a clinical trial, phase ii clinical trial, multi-center around thecountry, led by wyndham and myself, and this

enrolled 70 patients, all relapse refractory.here we took all patients of all molecular subtypes and determined whether they wereabc or gcb by profiling. and then they started getting this same drug at the same dose. andwe've now finished the clinical trial, and here are some of the survival curves, wherehere's our lady that is our exceptional responder at the top, and here are a number of otherindividuals, some of which are still on study with these arrows. and now this is a littleold, this slide, about two or three of them are out past a year, but you can see thatmany of abc tumors that are relapsing but at a longer time point than the gcb tumors. and this is another way to depict the results.this is a waterfall plot looking at loss of

tumor volume, and you can see we have a lotof complete responses here, and partial responses, and most of them are in the blue abc typerather than the gcb type. so overall, response rate was around 40 percentin the abc type with -- counting both complete and partial responses, whereas only 5 percentof the gcb type responded. so this is evidence that in response to a targeted therapy, there'sa need to know about this molecular distinction between these subtypes. and this translates into a statistically significantincrease in progression-free survival. and at overall survival -- our overall surviveof the abc type is 10.3 months, and the gcb type, 3.3 months. so, you know, it's a benefitto these patients. obviously, we need to add

something to this single-agent therapy tomake it really work, but it's certainly a start. so can we understand which of the abc patientsare responding and which are not by looking at the mutations that i've been telling youabout in this various pathways? sort of. so if you look at tumors that have a mutationin the b-cell receptor, and the numbers are small here, that six out of nine of thoseresponse, so a somewhat higher response rate than in patients that have a wild-type formof the b-cell receptor. while that's significant and indicating that the mutations are important,i want to emphasize that 30 percent response rate in patients without a genetic disease,genetic cause, is an important signal and

tells us that there may be non-genetic waysto turn on b-cell receptor signaling in these cases. and to drive that home, here's one other clinicalvignette: one of our patients, a 48-year-old gentleman with abc lymphomas. by our sequencing,had wild-type b cell receptor in myd88, had relapsed from several agents, went on therapy,went into a complete remission; here's a thoracic tumor that went away at week 10. i believehe had a four-and-a-half-month response to this drug and then relapsed. important pointis that we sent the tumor off for some very deep sequencing by a company called foundationmedicine. and they found, using their next-generation sequencing, something we hadn't found by conventionalsequencing. they found, at a 5 percent level

in that biopsy in that patient, there wasa mutation in the b-cell receptor that we had missed because of our techniques. andthat when we looked at that tumor, here's before treatment, this looked like a wild-typesequence. but when we started looking at the tumor on therapy, in the pressure of ibrutinib,now you see this sort of noisy line. this is another sub-clone that is that rare sub-clonethat didn't show up very well here. there's a second band, so there is a rare sub-clone,says this tumor signaling there's a mutation the b-cell receptor, but remember, the wholetumor went away. it's a complete response. but only 5 percent of the tumor had this mutation.so you don't need the mutation to be responding, but this is obviously telling us somethingimportant. now what about the other --

male speaker:was this study done on the tumor prior to treatment, or [inaudible] relapse? louis staudt:this -- well, yeah, right, good question. this is -- the first one is beyond study butbefore ibrutinib treatment. this second one is three days of ibrutinib, only three daysof ibrutinib. so within three days, we start selecting out this sub-clone that has a higherlevel b-cell receptor signal. all right. what about some of these othermutations? card11, if you remember, the wiring diagram is way downstream of btk, so you mightsay this might be a resistance mutation. numbers are small, only had two patients; they didn'trespond. not going to make too much out of

that. what about this myd88 mutations? would theysomehow circumvent the ability of ibrutinib to work? well, not really. you had responsesif you were mutant or wild type. but the picture got much more interesting when we asked whetherthey had concurrent b-cell receptor mutations. and in four out of five patients that hadboth the myd88 and the b-cell receptor mutated, they responded, whereas zero out of sevenpatients that had only the myd88 mutation responded. and why is this? well, we thinkit has something to do with the fact that genetically, there was an overlap amongst155 tumors in about 10 percent of the abc tumors had both mutations, and that was morethat we would expect by chance. so we think

that there might be a synergy between thesepathways in some patients, whereas the zero out of seven response over here may say, well,maybe there's another way to become this abc type of tumor that has nothing to do withthe b-cell receptor, will not respond to ibrutinib, and we need a new therapy over here. okay, and since the hour's late, i'm goingto sum up, saying that ibrutinib, the latter part of the talk, is showing you this irreversiblebtk inhibitor is inducing a high rate of response in previously refractory tumors; that themutant b-cell receptors have a more frequent response to this drug, but you don't absolutelyrequire that. and here i will highlight some recently published work on this drug, ibrutinib,in chronic lymphocytic leukemia, and in mantle

cell lymphoma, two new england journal papersyou may have seen, where there's a 70 percent and 68 percent response to this ibrutinibdrug, and by genomic sequencing, there are no mutations in the b-cell receptor, but thosemalignancies are responding to this drug. same sort of thing, this sort of non-geneticdependence on the b-cell receptor. i didn't show you this for the sake of time,but ibrutinib combines very favorably with a lot of other drugs, including conventionalchemotherapy, leading to a lot of combination trials that we're now setting up at the nihand around the country. and to sum up the whole talk, i would liketo emphasize, from the research point of view, that we can combine this next-gen sequencing,find the structural genomics of cancer with

a functional genomic study of the essentialpathways, getting some good computer scientists involved to deal with this avalanche, thisbig data problem, and arrive at these essential cancer pathways. but going forward, and somethingi'm going to be focusing on in my new job, is that the patient will be at the center,and we will be wanting to do all of these genomic studies in the context of clinicaltrials where we have the outcome and we can relate the response or not to the predictorof molecular abnormalities. and very soon, coming to a hospital near you,will be the following paradigm: that the patient will show up, they'll then undergo some sortof genomic profiling of sequencing the dna or looking at expression levels and profilingin that way. there will be some phase of data

interpretation. then you'll make a decisionto employ a particular drug based on what you found. you'll image to see whether therewas a response. if and when drug resistance occurs, you'll have to get a fresh biopsy,figure out whether there were any new mutations that occurred in that tumor, and then if youfind an actual mutation, employ -- you have a different drug. i don't think this is sciencefiction; i think this will be reality in the very near future. so, just to sum up, i just really want togive most credit to wyndham wilson who helped lead all of these clinical trials of ibrutinibwith me. thanks very much. [applause]

male speaker:so how will this incredibly [unintelligible] science find its way into the clinic? howwill oncologists learn of this [unintelligible]? louis staudt:well, it's already happening, and for better or for worse, so i'll take this opportunity.so for, i think, about $4,000, you can order a test from foundation medicine, and theremay well be other companies, so i'm not endorsing just that one company, but they will sequenceover 200 genes for you and find all sorts of abnormalities. then you can see, oh, well,there's maybe a drug targeting this particular abnormality. that goes under the rubric ofthis in-of-one type; i think that may work sometimes. i showed you some examples whereit does work. it may fail more times, we just

don't know. so i think one of the virtuesis that we have the data available if we want to do that, but we still need good clinicalscience to see how informative that data really is. male speaker:yes. male speaker:thank you so much for this talk. i've been waiting for this for a long time. but i wantto give you another concept to reverse things, and that is that stage 3 and stage 4 rheumatoidarthritis is a disabling disease with an increased incidence of lymphoma. louis staudt:yes, it is.

male speaker:and so for the methotrexate failures, the treatment options are an anti-tnf drug, whichmay increase lymphoma, and with the problems with big pharma, we don't know if it increaseslymphoma, or [unintelligible] there, or we can just use adabadosec [spelled phonetically],which is a toll receptor, or we can use rituxan. and so how about thinking about using genomicsto determine aggressive therapy in rheumatoid arthritis based on your genomics? male speaker:we are recording; could you paraphrase the question? louis staudt:the question is, can this way of turning the

crank on a molecular view of disease be applied,in this case, to rheumatoid arthritis. and it's certainly what i've been trying to promotefor years. the problem -- the cancer is special. we know where the disease is: it's where thetumor is. rheumatoid, where is the disease? what should we profile? can we look at theblood? well, maybe not, maybe the relevant cells are not floating around the blood, maybethey're in the joints. got to get a joint biopsy. did you hit the lesion that is informativein the joint, because it's all spotty? not so simple but not impossible. there, you know,there's -- so i think there -- and of course, there's an underlying genetics. you can alwayslook at the germline component. and i will emphasize what you said, or restate it, isthat there is indeed a genetic increased risk

of non-hodgkin's lymphoma in patients thathave rheumatoid arthritis. there is a genetic component that we have yet. it's about a twofoldrelative risk, but it's there. so there could be -- and as you've seen, because the b-cellreceptors can be -- b-cells can be involved in both these autoimmune diseases and thesemalignant diseases, the same receptor pathways could be used, and we don't know yet, andit'll be fascinating to see how ibrutinib works in this patient population. male speaker:some oncologists won't let us use an anti-tnf in a patient that has pre-existing lymphomaand rheumatoid arthritis, and some will, you know? and do you have any comment --

louis staudt:it's so hard. it's -- well, i don't, because tnf is hooked up in so many different ways;depending on other things that are going on, can either promote the survival or the deathof cells. it's way complicated, and you have to be talking about which immune cells, whichsubset are you talking about, so i don't have any deep insights there. i think that -- imean, it's not every patient that gets these lymphomas. i mean, as i've understood, they'repretty darn rare. it's hard to even collect a series. so they're reported, they probablyhappen, but as you alluded to, we don't have a real study of what tnf -- anti-tnf-inducedlymphoma looks like. i'd love to study it, i just don't have the tumors.

yeah. male speaker:there are decades of failed clinical trials on drugs that don't meet statistical significance.how do you go back and look at trying to find those exceptional responders in those drugs? louis staudt:so that's exactly what we want to do. so the cancer institute has -- moderator:question? louis staudt:i'm sorry. the question was, we've had decades of, let's say, empirical trials. i mean, theywere a new drug tried in cancer that didn't

work. and can we salvage anything by the molecularanalysis of tumors? well, fortunately, we kept a lot of those tumors in tumor banks,and now one of the things i would like to do is turn this heavy-duty technology on thoseolder tumors and see if those two patients out of 45, which is what their experiencewas in that bladder cancer trial, which was a failed nci-supported trial, now we can see,well, those were the two, and now if we -- you know, because when you get into common cancers,if there are 1 or 2 percent of patients that have a targetable lesion, you want to knowabout that, and so we have to care about cancer in all of its heterogeneity. so when we'vedone our clinical trials, we've lumped everybody together. so we couldn't possibly have seenit.

male speaker:the example there is kras in colon cancer, i guess. louis staudt:absolutely. male speaker:one more question, and then we'll break. male speaker:just a quick question around the future because, in the question, if you think about it again,from a clinical micro epidemiology perspective, in the future of the clinical trial [unintelligible],in the age of electronic medical records and big data, there's been some recent articlessaying that the use of registry-based real time study -- is that something that --

louis staudt:yes. that -- i -- less fancy way, i call it a non-clinical trial clinical trial. that'swhat i've been doing for all of the work that i described in lymphoma. these were not patientson a trial, these were patients that received standard of care at good academic institutionsaround the country, and we just got their tumors, figured out what their response was,and made sense out of it. i would say that there's still a need to conduct things properly.but there might be a proper way to know with electronic records that a patient had actuallyreceived a particular regimen, you know, hadn't stopped early -- there are problems with doingwhat you suggest, but it's not at all insurmountable. and it's -- and it could well be -- and avision also that we have is that genetic profiling

will become so inexpensive that patients willjust get it as a course of their care. and then what we would like to do as researchersis if the patients are willing to donate their cancer information, we would then like tojust take the -- their tumors at that point, and take the data and then look at not just10,000, how about 100,000 patients, how about a million patients? and then really understandmore that way. male speaker:are you able to do expression studies in preserved tissue -- louis staudt:yes. that's a very exciting new development. we can do perfectly fine rna sequencing fromformal and fixed tissue.

male speaker:one of the best lectures we've had here in years. thank you very much. louis staudt:thanks.