Causes Of Ovarian Cancer

today our speaker is christina annunziata, she got her m. d. at university of georgetown medical school, subsidiary wently she joined nci and worked in lou spout and work wide elease cohn, now she's head of translational genomics section, her title ovarian

cancer in the genomic era. >> okay, thank you very much so i was just looking back at my previous presentations and i realized this is the eighth year that i'm giving this lecture but it was not've've i was rnlt always called the genomics era. >> yeah, so this is definitely

updated because it wasn't previously genomics. anyway, so thank you for inviting me back to speak again. i wanted to focus on the topic of genomics and functional genomics. to start out with and i'll get to ovarian cancer in a little

bit. so what is cancer genomics. genomics in my view is the study of the whole genome so not individual genes or individual mutations but study of the genome in general, either chromosomes, gene expression or some sort of global analysis.

so i would say the genomics era fullly started back in 1959, with nowell and hung erford and they identified a abnormality called the philadelphia chromosome that was occurring in patients with this chronic form of alualu keepia. and this is what it looked like.

so in 1959 they looked the chromosome from multiple of leukemia and they notice thad everybody had or most of the patient his this little thing here with the arrow. where's my thingy. i don't kow where the--there. a little extra chromosome.

it looked like an extra chromosome, so they studied this, eventually in 1973, janet rowely discovered that this little chromosome was a reciprocal translocation between hromosomes nine and 22 in all these patients which was at that point called roggic mileog nows

leukemia or cml. and then in 1984, groffen et al, was a chromosome refusion at 22, and actually at 19 but here he found it was a recurrent fusion between bcr and able. so the gene bcr, break point cluster region and a kinase, and then it wasn't until 1996 that

we actually came up with a drug called amat nib which here it's called cgp 57148 with a drug that blocks the able kinase function. so that was quite a time between 1950--what did i say 1959 and 1996 before we came up with serra pine for this particular

phase. so, that points us to what is actually functional. so, the therapies we use are blocking a function of something. so how can we use functional genomics, not just observational denational library of medicinics

like they did in 1959, can we use sort of a functional genomic experiment to find out what part of the genome is actually doing sing. what is causing an effect and what can actually transform normal cells o cancer, these so called driver alterations.

so in 1981, i think shih et al, performed this research. and they found transforming genes of carcinoma and neural blastoma into mouse fibroblast. so they took the entire dna of neuroblastoma, that's why it's called new, chop today up put it i nontransformed fibroblast and

saw which part of the genome caused an effect which was to transform the cells into cancer cells. that could actually form tumors in mice. so they found a gene they had called new and then in 1984 schechter the neu gene was

close to the egfr. this is a southern blot. in 1985, coussens and group found that this was on chromosome 17 and it was amplified and can you see on that blob on the side that there means there's a ton of, so there's a lot of it.

so it's amplify indeed in region of chromosome 17. and then of course in 1987 dennis slamon discovered it was a in breast cancer cases and this is chromosome 17 in all of this region of chromosome 17 in all of these breast cancer cases and in some of them it is

extremely amplified and that eventually led to--and that eventually led to the development of an antibody to block this receptor and the approval of herseptorsin or the dug that we use today in breast cancer that are known to have hertwo amplifications felt so

this is the first functional genomic experiment. so what about ovarian cancer, can we use genomics to study ovarian cancer and do we have any driverin ovarian cancer. so now i'll introduce ovarian cancer so getting to the actual tub ject of this particular

traco lecture. ovarian cancer as you may know is the most lethal gynecologic malignancy in the united states. it causes greater than 16,000 deaths per year and the fifth most common cause of deaths in women. this is because 70% of women are

diagnose wide ovarian cancer. it's unusual to find ovarian canckener early stages. less than 35% of the advanced stage patients are alive at five years. and this is what we mean by stage, stage one is early stage and that occurs in about 20% of

the cases but the survival is quite good, the survival is 90%. if you can find stage one confined to an ovary, the survival is quite good. stage two is--has left the ovary but is still in the pelvic region. stage three as spread or to

lymphnodes and that like i said is the most common stage and unfortunately this level is transpired to 45% at five years. and then stage four, the survival is even worse, les than five% at five years. so ha is the treatment for newly diagnosed ovarian cancer, of

course the first one is complete surgical staging and i will show a bit about that. optimal reductive surgery which we'll talk about followed by chemo therapy and then of course clinical trials. so complete surgical staging--sorry, complete

surgical staging involves looking through the entire paro tonem and for lymphnodes and looking--as can you imagine the pelvis and abnormalities domin are sort of contiguous, so the difference between stage two and stage three and different anatomical intrusions and not a

full hard line of separation. and the pero ton eel fluid circulates freelyy throughout the abdomen. so the surgical staging needs to include everywhere in the abnormalities domin because of the way the pero toniel fluid is.

that is the goal and now the goal is actually to leave as little as possible because as can you imagine the survival of ovarian cancer improves with the less and less disease left at the time of surgery. so the best survival of ovarian patients is if they can get to

zero disease visible in the abdomen. chemo therapy, the basis of chemo therapy is platinum in combination of attacksain and pero toneal, and optimal meaning less than 1 centimeter at the time of surgery. the combination of platinum and

tax an, can you see the introduction of various chemo therapy and five year survivalot bottom. of advanced stage disease, this is stage three and four only. at the time of the introduction of alcollating chemo therapy in 1960s there was basically no one

who lived beyond five years if they were diagnosed with stage three or four ovarian cancer. when platinum was introduced in the 1970s, that was the first sort of big revelation, big jump up to five% of people alive at five years in the 70s. so then in the 80s people

decided well, we should combine them with platinum and actually that was better. got up to 15%. five year survival in the 1980s and then in the 1990s was the introduction of the p a xatoxa taxle or dosattacksle, in the 1990s, another big job with the

p a clitaxel and carboplatin, up to 35% and then in the 2000s, it was actually not new chemo therapy agents but a new way of introducing them so instead of giving them ivs that was to give it interpero ton eel because that's where the bulk of the tumors are.

so when they did that, the survival went up to 40% in the 2000s. so this is survival outside the earth. but from a molecular standpoi, i think we still don't know very much about ovarian cancer it is a very heterogeanious disease

and it's heterogeneous at the point of looking at it. so if you think of ovarian cancer histology, ovariance cancer and made up of five distinct histologic subtypes. the most common of here is the theorist, so the high grade ovarian cancer, the most common

and also the worst prognosis, and that is a particular type of ovarian cancer which is shown in a and then have you a low grader here which is more pap illegalsar che is shown in b and then in c, you have endomeet rioid cancer. and d, have you a museinous

cancer and then in e, you have the clear cell cancers. so you have at least five so if they look this differentlyot microscope, you would think they are probably very distinct at least at a molecular level as well. right?

so we're treating them all the same way but maybe we could improve if we figure out what is actually different between them. the tissue of origin is also potentially different between these cancers, recent work has shown that perhaps serous cancers arise from the fallopian

tube. the endomede riidentity could give way to the subtype and the extra uterine epithelium might give rise to musnows or other types of ovarian cancer. so very heterogeneous to the point of a cell of origin. so the hypothesis is that if we

increase understanding about biological and biochemical events under lying ovariance canner progress, you can treat them, maybe we should give them all the same chemo but give them additional agent specific to their mutational status. can we use genomics to figure

this out. so i'm going to start with an overview of each type of cancer and show you where we are so far. i have--i can tell you the punch line is we don't have specific therapies as of yet but we're getting there.

so let's start with clear cell or endomeet rioid cancers that are related from the cell of origin standpoint. clear cell cancers are in fact 10% as i mentioned where serous is more common. 70%. they do have a worse response to

standard chemo therapy with the platinum and the taxain. and of course those are associated with endometriosis in 40% of cases so this study back if 2010, looked at what are the genomics of clear cell ovarian cancer so they started out by sequencing rna from 18 cases of

clear cell ovarian canckener one cell line in a discovery phase. and then they went on and they sequenced axons from 210 additional samples. 101 clear cells, 76 serous, negative comparison, and another clear cell line. and then they immunostained,

they did immunostaining for interesting things in another 455 samples. so a pretty big study. and what they found was that they found arid--they actually--stay span the entire gene so they're not in 1 recurrent location but it's the

gene itself anywhere in the gene is mutated so it's sort of--it's not particular kinase location for example, it's anywhere in the gene that can be mutated. so what is arid1 a and this gives us a clue as to why it's not in a specific location. so arid1a is in a chrome tin

complex, it's found in breast cancer and lung cancer previous to this publication in 2010. it occurs in chromosome 1 p 36 and it's deleted in all--in 6% of all cancers across the board. so this suggests a mutational spectrum across the gene suggests it's probably a tumor

suppressor gene, so it's not an onca gene, it's a loss of mutation, it's a loss of function of the arid1a gene. so they did confirm that arid 1a expression was lost here on the this it says the loss of 215a, and it's the loss of the protein.

so the protein was lost in 73% of clear cell cancers and with the arid1a mutation and it was lost in 50% of endomeet rioid ovarian cancer cases that were also arid mutations but 0 cancers. so this suggests, okay, it is something that's clear cell

endomeetriol subtypes in so, summary, to summarize, arid 1-a was mutated or lost in over 40% of cancers and these are ovarian cancers and less than 1% of serous cancers however since there's tumor suppressor, we're not sure of what to do with this iation we don't have a way

of which pathway is effected and there's therapic direction here with the loss of a tumor suppressor, it's difficult to target that in a way. but at least we are learning that there are specific mutations here. so we have potentially

diagnostic utility but i would argue that this was--this is an interesting finding but it didn't give us therapeutic direction because it was not identified with a funcl experiment so there was nothing where we put in the gene, put in nd an effect.as something from the genome and

no, we're just sequencing and we're making an observation at experiment so that we're left without a functional outcome. okay, so what about muse nows cancers, they have a worse over all survivals compared to nonmucinnous cancers and this is a survival plot since yours of

diagnosesi of muse nows verses nonmucinous ovarian cancers. and then in 2006 this, is an overview of gene expression event that looked at serous low grade, high grade and mucinous cancers either grade 1 or grade 2 ask they discovered that mucinous, were surprisingly or

not surprisingly were different. from a gene expression stand point, a early gene expression array, 10,000 genes on it, it was 1 of the early 1s in 2006, but still with limited information you could see that the mucinos cancer vs a completely different gene

expression profile than the cancers, so, okay, that is good. we found--we didn't find, they found that perhaps krase model was more common than in mutated cancers compared to serous again this is an extremely small today, and similarly in low grade serous cancers which are

distinct from high grade cancers again, krase model and braf were found to be mutated so braf was studied at codon 599 and codons in 12 or 13 which are the common sites of mutation, and was found interestingly that 68% in this small study of low grade cancers and 61 of serous border line

tumors which are kind of a precursor to low grade serous cancers were--had 1 of these 2 mutations but none of the high grade serous cancers that they studied have this mutation, so another clue, says pointing up towards musnows or lo grade cancers perhaps we can target

kras, o the rs pathway. this a graphical representation of what i just said, this is the low grade cancers were more commonly mutated than the high grade serous cancers endomeet rioid or clear cell. there you go. so how can we target ras

signaling. i think everybody heard over the ras pathway because we have the huge ras initiative at the nci. how do we target it? it's down stream at ras, that is at the point of--at the point of nyc. so ras, signals to raf, to mek,

and goes to erk and changes gene expression. so particle farley, et al, decided we should try this in women with low grade cancer so they treat the poo patients, 8 responses and 34 patients with stable disease greater than 4 months.

so that sounds promising but if you look, then at the mutational status of these women the punch line here is that there were responses in 5 patients who had no zraf or kras mutations and only 2 responses in mutations so actually the responses were nor common in people with the

mutation than without the mutation. so that didn't quite go as planned. so where do we go from here, so as i showed you before the ras pathway was neatly described as going from raf, to ras and beck to erk, however that is not the

case this cancers or other cells as well. it probably doesn't care to much about mek because it as a lot of ways of getting around that so that's probably why it didn't work so it's not a driver. so again we found this mutation by sequencing, so we did not

find the mutation by viewing a functional experiment. we just said, wow, there's a k-ras mutation but there were no functional experiments to say that the ras was actually a driver. and even if it was a driver, mek is not the driver.

as i mentioned 70% of cases are diagnosed as high guide serous. so back in 2011 published as part of the first time genomeat las pintas, it was the second or third cancer that was precepted in the cancer gerontologysts genome atlas, and identified molecular abnormalities that

influenced, well, the goal was to identify molecular abnormality pathophysiology and effect a potential therapeutic target. so what they did was another--what i call observational study, was 489 hybrid cancers looked at

microrna expression, dna copy number, promoter methylation and they did whole exome sequencing on 316 of those samples. so sample encliewgz of course, they had the newly diagnosed patients, no prior treatment and companion normal tissue, adjac ept normal tissue, or peripheral

lymphocytes or previously extracted germ line dna. so that was this experiment which was the copy number analysis. so what this is showing you is on the y-axis or chromosomes, so chromosomes 1-22, and on the x axis is individual cases.

so these are 489 cases, and they're showing you here, hybrid theories of ovarian cancer compared to the previously studied tcga samples, and that's what gbm is. so what can you see here, well, and then the scale is that red is amplified and blue is

deleted. so what can you see here is that in glial blastoma there was a relatively quiet genome and there were distinct areas of either amplification or deletion. but in hybrid cancer, the amplifications and deletions are

kind of all over the place, so you see distinct areas of amplification or deregion, and all the chromosomes, there are some distinct 1s but it's really not pointing there. so the conclusion from this is that there was a high amount of genomic disarray or chaos in

ovarian cancer that is not seen in other cancers. so is this a problem with dna repair? that's leading us towards something interesting? significantly mutated genes, p53 of course was the most commonly mutated.

brca 1 or 2 was the next most common, brca was on the top and brchot bottom, so you combine them together in the same pathway was the next most common pathway. and then a number of other 1s but as you can see, nothing really sort of jumps out at you.

as anything besides [indiscernible]. i didn't press the button first. okay, so, if you focus there on this idea of genomic disawry registered investment adviser and homologous recombination with khi is a way of fixing genomes, we than brca is alter

indeed 33% of cases, either by deletion, mutation or methylation. so if you look at the survival, the brca mutated cases actually do better, interestingly. so the brccase mutated here on the left is in the red, and the methalated and wild-type is in

the black. so, the thought was that with all this genomic disarray, and the recurrent brca mutations, maybe there is a common theme in ovarian cancer as defects in dna repair. so they looked at where else do we have museumitations that

we've seen, and where else are alterations and we can say that maybe in 51% of cases not just the 1s with the brca mutations in 51 overall we may be confined of specific defect in homologous recombination or dna repair. then this is another repair, dna cases have dna repair or cell

cycle problems in fox m-1 so we have dna repair and cell cycle, maybe that accounts for ovarian cancer. who knows? so what next. what do we do with this information. it was--now the functional

experiment but maybe we came up with potentially functional therapeutic targets with respect to homologous recombination defects and 1 of those drugs is inhibitors and commonly deregulated pathways poetic sentially was p a rp inincrease in bodyitors, for the first time

in 1990 by hall and king et al, who discovered that there were--there was a particular part of the genome that was altered in a families where there was this recurring breast cancer throughout. so if you look at this pedigree, you can see that the solids are

females and the squares are males and the solids are effected with the cancer. so can you see there's a recurring hereditary breast cancer in each of these family which is effects about half of the women. so these people identified

eventually the brca gene as the ocuesal locust at that point. the brca gene that was mutate indeed these families. in hybrid series cancer, the braca like i mentioned was about 10 or 15%, the braca in the germ line, but then also braca can be mutate indeed the cancer itself.

and not mutated in the in the peripheral dna in the genome. it can also be methylated but then other things in the dna repair pathway, such as the emcy amplification or other homologous recombination defect and this is where we think that in about half of the cases, we

can find genetic evidence for defects in homologous recombination. interestingly instead of just looking at copy alterations numbers and found that there were 230 common genes that might present a signature of a small recombination deficiency, and

this is actually looking in breast cancer patients with homologous recombination deficient again had a better survival homologous recombin intact. and if you look in ovarian cancer cell lines, even the case of 2 different cell lines 1 with

homologous intact signature and 1 with homologous deficient, the deficient 1s is more sensitive to the perp inhibitors. so what our p a rp inhibitors? they are inhibitors of single strand dna repair. and this is a specific cartoon of how we think it works in

brcmutated cancer. so in a normal situation, you have the dna suffered single strand breaks by various processes such as cellular metabolism, exposure to the environment, et cetera. so what happen system that the p a rp comes in and marks these

are repair, so if you block the p a rp, in replicating the cells the single strand gets convert wide with the dna, so gets converted to a double strand break which is dependent on homologous recombination or repair that uses the brca 1 or brca 2 gene in this pathway.

if you have cells that are deficient in brca, they cannot repair this and they get sort of overwhelmed because they don't have a homologous recombination pathway in fact, they have huge amounts of double strand breaks because of the p a rp inhibitor and they would selectively die.

that is the hypothesis. so a p a rp inhibitor that we've done trials with is azd, 2281. this is a orally available p a rp inhibitor and has been shown in the synthetic leathallity in the cell lines and it's been approved for third line treatment of women with brca

mutation and recurrent ovarian cancer. our study was to take it further so going from single agent to combination with chemo therapy and the idea here is that the carboplatin will induce a lot more breaks than just environmental exposures and

cellular metabolism so we will give people carboplatein so we can cause single strand breaks and we will give them the ibt--integrate hickittor and if they don't have a homologous combination because of a brca mutation or because of 1 of the other 35% of women who don't

have brca mutation that have evidence of defects is homologous recombination so cohort 1 was brca mutant breast or ovarian cancer and that was publish indeed jmci 2014 and cohort 2 was triple negative breast cancer in recent tcga analysis have similarities in

the deficiency, similar to what's found in ovarian cancer, nonbrca mean. and this cohort 3 and serous ovarian cancer again with the thought that even with a normal brca, there's still a 35% chance that you might have a defect in homologous recombination and

these were both presented by laboratory 2 in our group, at azr 2014 and then also recently in asco this past year. so we were able to dose escalate and we found a safe dose of ola p a rib, in the carrier which is is comparable to what we found in the nonmutation carriers

either breast or ovarian cancer. so as i mentioned phase 1 study was recently publish indeed 2014, and we enrolled 45 patients, 35 ovarian and only in the brca mutation cohort. those dose escalation and dose escalation, and phase 1 b expansion tolerated doze in

additional 15 patients. and happily woo found we had hit on something quite interesting. so this is the colors are the dose levels but you can see at all dose levels we have people responding to this drug, even though they had previous not responded to platnim alone, so

with addition, to the platnim we were getting good responses, deep responses. we had 3 patients that have been on study for more than 5 years that have on this particular cohort. so this was very extremely promising.

this is just to summarize that we had--this [indiscernible] but since then we had the compleat responses. we had in the ovarian cancer cohort, a pate response, pate response rate of 44% and stable dedisease of 38% so overall 82% with chemo refractory disease,

who have been given the carboplatin and functional that we can work with. so the conclusions were from that trial that it's about tolerated in combination of carboplatin and highly act and i have advance refractory brc efficient cancers more activity

of the higher dose, and a positive proof of concept that we have activity and tolerability in this negatively defined targeted with the brca mutated cancer. we did present the results of the high grade cancer at the asco 2015 meeting and don't have

the results here but those were--those were very promising so that brings us to exploration of new targets. so how can we now go from using the observational sort of gene expression mutational analysis, to going toward the more functional experiment, how can

we sort of recapitulate what was done back in what did i say 1981 with the her2 new? let's just step back and say there are a lot of tests that are being marketed out there for these actionable mutations. but when you hear this, this does not things that had been

functionally validated. these are things that might potentially be targets of of therapies that exist, and you see them being sold by--from places like [indiscernible] or foundation 1 and you get back a report that have you this krase model mutation for example, that

is possibly sensitive to drug xyz, or you have something that is unlikely to benefit from drug abc. so you're getting these reports that are more observational than functional. what we have ongoing right now in the nci and also around the

country are basket clinical trials to test these hypothesis that if have you this mutation, are you actually sensitive to drug x, y, or z. so what these people are doing is they're going to c-cleanse these and these are not tested and they're not definitive going

to respond. typically as i menkzed there's no function algorithms link other than we introduce this mutation to a cell line and half of them had a response. so what are actionable mutations? this is an editorial,

interesting, by redig and janne, which is quite apropoe. so they say actionable mutations depends in large part on the strength of the data linking the targets and the targeted therapy. thr this trial designed to work 2 key conditions must be met.

number 1 the target must dependot target pathway and number 2, the target therapy must reliaisonnably inhibit the do you see we have a lot of information missing. we do need to do these childs to figure out what works in what situation.

but a lot of the information is still missing, which i think these 2 key concepts, sort of a hurd toll what needs to be address indeed in whole quest for personalized medicine. achieving both goals can be a matter of some complexity in these.

so how do we get actionable targets? from a experimental standpoint i think we need a functional experiment. so how do we use functional so 1 way that can be done is 2 this happened and 6 by dr. lou stout at this information is a

functional approach using an rna library, asterisks shrna library, so what they did was they had a library containing shrnas between 6 and 12, shrnas tarpgetting 2500 different human genes. they said that i have 3 contacts per gene but since then i know

they've updateditous 6-12 per gene so there's 15a thousand constructs in in library, all sequence verified and they contain a bar code so they can identify which target sequences have been lost from the pool. the library contains protein 3 kinases, d-ubiquitinning

pathways and other genes of any, so what does library screen does is it gets infected into the cancer cells and it's inducible so can you induce it in half your population and can you leave it off in the other half and then you look for genes that are or bar codes that are lost

in your induced population and presentient in your noninduced. so you're really looking for things that are specific to this cancer cell growth and survival in culture. so we did this library in 4 ovarian cancer cell lines. we looked at 2 serous and 2

nonserous ovarian cancers and we found there were a number of genes, 55 genes that were common to at least 3, 3 or 4 of these, of these cell lines. could these be drivers? i don't know. we have a list of genes that are common to serous cancers we also

have a list of genes that are common to nonserous cancers and we are following up on each of these to see if we have actually identified a functional target. like i said we're following up in 6 additional cell lines tworks different rnai constructs and we're selecting

the targets that do have drugs designed for these genes. in addition we're doing functional screens and we focus on the nf-kappab signaling pathway as what might be a targetable taught way. so if you go back to the cancer denome atlas, 1 of the other

things that they looked at was by gene expression kwe identify subgroups of ovarian cancer that syrian cancer that have been common. and 1 of the things that we became interested in was an immuno reactive subgroup so it looked like there were a lot of

genes that were potentially evolved in immune inflammation or immune regulation. ask when you put that in the analysis, you come up with the complex regulating those genes that's why we focus on nfcap b so what is nf-kappab, and it's a signaling pathway that

promotes survival and proliferation of lymphoid cells and cancer cells this, is an example of 1 signaling pathway of nf-kappab, that's in the nucleus that promode sural and proliferation genes but they're held cytoplasm in the inactive state by the ikks--sorry the

icappa b, it's bind and retain the transscripgz factor in the nucleus upon stimulation of a molecule such as tmk is you get phosphorylation of the icappa bs in the proteosome and then so what we did is we actually looked at ikk beta and epsilon as regutors of this pathway to

see if we can come up with a functional experiment to target that area in ovarian cancer. so we use the library in combination of either i kkhi, or epsilon low and we looked at ovarian cancer cell lines and we looked for things that were specific with the epsilon was

low. interestingly we found check 1 wch was highly sinner gestic with the loss of check 1 and vice versa expression of check 1 was very sinner gestic in promoting survival of igf 1 cancer high, interestingly check 1 peaked our interest

specifically in ovarian cancer because we found that it was overexpressed in cancer genome atlas in almost a hundred% of ovarian cancers. so perhap its has something to do with the pathway of genomic dsarray. in response to dna damage.

so dna damage it activates in once sense atm which activates check 1 or 2, to eventually halt the cell cycle at the g2 m check point. if you don't halt it there, it goes through it's cell cycle and then the dec kind of amplifies and you get additional dna

damage and the cell will die. so can you imagine that cells with a very messed up respond will have elevated check 1 or 2 to block that response and allow the cell to sort of imperfectly repair the d na enough so they go through the cell cycle without going to apoptosis.

so it makes sense in this type of cancer. and representing other histology, and we actually do have a clinical trial ongoing with and interestingly we are seeing higher responses in people without a brca mutation suggest thanksgiving check 1

might be more important in women without brc mutation. this was highlighted as i mentioned here by functional approach instead of just an observational screen. so in summary this is what i went through quickly this, is divided into epithelial and

nonepithellial, epithelial are high grade, low grade, clear cell, serous and we have each of these on the homologous combination defects in the high degree of ovarian cancer with some moderate success. so, just to tie it all up, 1981, the discovery of her2 was a

functional genomic approach by throwing genes into mouse fibroblasts and seeing what worked. we've done shrna library screen and thrown that in there and see what works when you knock it out. what kills the cell when you

knock it out. robber so functional denational library of medicinic screen and then it is clinical markers and we've take 10 to clinical trials with that success. with that i will stop and take any questions. [ applause ]

>> [inaudible question from audience ] >> so the question was why did loss of brca function relate to ovarian and breast cancer and not other cancer since all cell types need homologous recombination and that is a fascinating question and i've

always--i've thought about that for the last 10 years. because, you would think that you would have any type of cancer, kind of like taking from any patient who is have multiple different types of cancers. so, i mean, 1 thought is that the--the breast and ovarian

cells because of the way they're cycling with the hormonal sort of influence, they might depend more on the brca related homologous combination as oppose to other doubts. this is all hypothesis. the other thought is when leukemia--sorry in blood cell

fist you lose both brca genes it's lethal so it will loose the function of brca, so it cannot, it just dies, like the cells in the blood and the bone marrow might be getting, you know they're losing both copies of brca, they can't form a cancer because they're going to just go

away. so they might have the ability to survive, not that they're effected by the brca but they're the only 1s that can survive without it because they have other ways to operate and other dna repair mechanisms like nonhommologious enjoin tag can

allow them to grow even though they have this loss of homologous recombination. yes? >> so the question is why is ovarian cancer diagnosed in late stages and what can we do? that has been a dell impedimentsa for many, many

why it's diagnosed so late is because it does not present with obvious symptoms. and it's really hard to examine that area, so if you think of breast cancer, the women can do a self-breast exam and notice they have a mass in the breast. with ovarian cancer, the

symptoms, you can't really feel the mass until it's enormous, these symptoms are big, they're sort of gi type symptoms, nausea, bloating, diarrhea, constipation, the ovarian goes for a long because there are nonspecific symptoms. people look for markers but the

markers are pero ton eel irritation, not specific to any--any and not because of cancer itself. so the marker s&p cac125 it can be elevated for conditions like lung disease or you know pancreatic disease or any irtags of the bowel.

whereas so it's not specific but then also not very sensitive because with 50% of cancers that don't have allocated at 125, so we're looking for marker bus it is difficult. the big question now is just awareness that pelvic and gi symptoms look for a cause.

the other reason is because the biopsy is the kind of invasive so it's hard to [indiscernible] but ovarian biopsies will be more of an open surgical procedures. >> yes? >> yeah, so, that's another interesting point they was going

to talk b. i know we talked a lot that it spreads--sorry, the question was what about the rapid progress of the disease. can that--does that contribute to the late stage? >> and i think it's not necessarily the rapid progression of the disease but

the way it spreads. so a lot of cancers would spread through the blood and the lymphnodes, ovarian cancer will--before it get interests the blood stream will shut, so it just sheds from the source of the ovary so it's easy for it to go and landot liver and go and

land on the diaphragm. so it's very easy for it to float around and take up a nigh hole. so i think that has to do with it. >> yeah, so quiet is any radiologic modalities play a role in screening.

there was a recent study, which looked at imaging studies either ct, ultrasound or transvaginal ultrasound and in combination of ca-125, so with a serum marker and that did show--did show some promise in screening for ovarian cancer bbut did not have a huge impact as 1 would hope.

the reason for that again is a lot of false-positives with the screening, detecting a mouse, that's not necessarily cancer and you have a lot of intervention going on for something that's not cancer. the early studies of that actually showed an increase in

diagnosis but it was in the diagnosis of late stage cancer not early stage cancer so thatta why they moved to not looking at a particular cut off at 125 but looking at the rate of rise of the ca-125 marker but there's lots of ideas out there, so the take home point is that the

imaging has a lot of false-positives and so it's kind of hard to use as a grade. okay, you're welcome. >> okay, i have 1 more announcement and that being, i'll be sending all the registrants an e-mail this week about the tumor boards and the

core facilities. so this is all voluntary if you would like to learn more about what goes on at nci, we have 3 tumor boards, medical oncology, ero logic oncology or pediatric oncology tumor boards and then we'll also have a tour of the pathology facility and the small

molecule core where they use robotics to develop new small molecules for cancer. so our next speaker is stephanie goff, she did a surgical residency at columbia, went to the brigham women's hospital at mass general she's now in the surgery branch.

she will talk about a hot new area, chick point modulation. >> yeah, i am, thank you. so before i get started how many of you are immunologist or in the immunology field? you might know more about t-cell activation than i do but we'll get started.

so to understand check point modulation you have to understand how the t-cell works and why we think cancer immunotherapy might may be a good idea. so the objectives i will go over today is talking about the mechanism of action of blockade,

and i will talk about the early clinical experiences i had here i was able to partic 8 in as a fellow when i was here before and how we discovered what immune related adverse events were and how it changed, how we had to think about clinical trials using immune mediated

medicine. there are now 3 fda approved check point modulators and i go thrgh the clinical data that led to those approvals as well as check point modulators as well as development of a fourth that's not on the list. so to know where i got started

had i made the slide 5 years ago, it would have had 3 pillars and those would have been surgery which date back to the egyptians, they found hirough atom grief showing them taking a breast cancer out of a patient. next would be radiation which got start indeed the late 1800s

and 1900s, incidentally, and then chemo therapy and activities and projects che got started at the dana-farber and was more fine tuned here in the nci in the early 70s. and chemo therapy goes back to 1940. but in the last few years, particularly in 2012 and since

then, they've been how to treat patients with cancer. we will go through that today. so there are 3 main ways cancer immuno therapy and activities and projects kework, the first is nonspecific stimulating the'mune reaction of the body. we can either simulate effector

cells, we've done this in the past using high doses of the hormone interleukin 2, we can inhibit the regulatory factors by turn being off check point block aid. we can actively immunize patients to get a more specific antitumor reaction and that's a

vast majority to go out and read what's happening at places other than here and we can passively transfer focused on antitumor activity, and which s&p really heraldad here by dr. rosenberg and continues to move forward as kind of a salvage therapy. but we focus on today is erne

hickitting regulatory factors or check point blockade. so if you know how a t-cell gets activated. it's a 3 step process, requires first signal 1 which provides the t-cell with the specificity. when have you a protein whether it is self-or nonself, it's

eaten up by the proteosome on the processed, placed on a major histocompatibility complex chrks is a pocket and makes its way out to the e. r. and back out on to the surfac of the cell. so any peptide that you might think could appear in the cytosol of the cell, comes out

as the small 9-10 peptide residues, sit nothing the pocket of an mhc. our bodies have the ability to manipulate our genes to create a vast array of different t-cell joining regions and we join with specific t-cells that can connect to this peptide in the

pocket of this mhc. that's what's called signal 1 or t-cell activation. it can happen with endogenous protein, it can happen with endosignifyitosed protein which is gentlemenly get present indeed the context of mhc 2. all nucle88ed cells have mhc 1

and they have the ability to reach in and pull out and put a 9 or 10 in the t-cell to find. dendritic cells have this ability to deal with peptides to get endoiitosed whether these are cancer, viruses, the flu you got last week, it bring its out and give its to a cd4 or

traditional limited partnerrer cell which then gives the specific t-cell activation. that's only the first step of the process, this t-cell won't go and do anything until you get further. solet next step in t-cell activation is that there has to

be a co ligand. so when the tcr next to the cd3 which is the marker of the imp o site gets activate today requires 1 of these second signals. some of these second signals are positive that lead to down stream activation but some of

these as you see pictured here in red are negative signals. ctla 4 which is 1 form of be is 1 if that gets activated that will shut down a cell, pd1 will shut down a cell. this determines if it's active or amergey, which does nothing until eventually apoptosis.

the third signal which not awell elaborated as the other 2 is the presence of cytokines in the area after t-cell has been activate today can polarize to 1 type of a helper cell or a t-effector cell and that is heavily dependentot micro environment in which it ashes

rifes. signal 3 is something that's being worked out these are things that would be more of the truggable target small molecule, we will talk about today is what happens at signal 2. what is the role of the signal 2, why do we have them, why are

they important. it seems like all they're doing is telling us about cancer, it prevents us from having autoimmune ria actions from the t-cells that find normal pieces of our body. so, initial response to an angioma will occur in the organ,

now those can be the lymphnode which is we have all over our body, tonsils spleen, and the gut, mucoseis associated lymphoid tissue right under the surface of the stomach and they allows us to recognize what is our cells so we don't start to destroy it.

the immune check points also limit collateral damage, and it's something that's activated and turned on, and then it goes to the side of say the splinter you just got in your hand from running your hand down the stair well, if you left that t-cell response going on and on you

could destroy your entire finger because once it turns out tdoesn't know how to turn off. so these immune check points stop the collateral damage by shutting down the response after it's gotten there is engaged. so in the context of cancer, we have 2 opportunities to break

the tolerance to self-antigen. because for the most part cancers were the something we get. it's not a virus, it's not something that our body can say, hey that's not your dna, let's attack it. it is your dna, it is part of

you and have to trick your immune system into doing it and that therapist is 1 of the ways to do it. this is the first check point blockade that was investigated by dr. allison who won the lasker award, i done finish you're aware of that. being

prominent in the immunotherapy community. the reason is once this t-cells gets engaged with signal 1, the it, cr engaged with the peptide and the context of mhc, gets turn office of diversity by activating cd-28. now depending on the affinity of

this complex, if this is a high binder it will increase the ctla-4 that goesot surface and once it goes to the surface, it actually bind this is ligand with more affinity that cd28 so if you did something that turns on, the off-signal comes on and it's stronger and it pull its

and it turns the t-cell off. so this is happening in the periphery when a t-cell first encounters an antigen. it helps shut it down. it has a dampening effect so that once they get started they don't get out of control. in mice that are double knock

out for this, they don't die in utero but they die quickly of lim foe proliferative autoimmune disorder within a few weeks. td1 is another ligand, excuse me, and this is out once the cell has been activated and make its was toy where it wants to act.

so have you a t-cell, engaged peptide in the context of mhc. it's been turn office of diversity by cd28 or itos, 1 of the other positive stimulators, it gets to where it wants to be, it starts the inflammatory process, this process makes the target cell, express pd-1 which

then shuts off the t-cells. and the site where the t-cells activating to shut it down. so any inflammation can upupregulate cd1 and it limits collateral damage when you get an infection or a splinter, it stops from going too far. so how does the pd1, pdl1 system

work in cancer. there's some cancers glee o blastoma is 1 example r the pathway within the cell upregulates cdl-1 on its own, it doesn't require a signal from the outside to turn on. there's been a mutation that make its upregulate cdl-1.

but another way we found is that when the system works, when we start feel killing the cancer it releases interferon-gamma and that interferon-gamma upregulates pdl 1 and starts to shut it off again. so we get some success, the body says, we are going to shut it

down before it gets out of control. unfortunately we want to get out of control because it needs more to kill the tumor so, we have to find a way to block this response to interferon-gamma. so with check point where should we start.

where did we decide we need to we start wide other tumors that were known to respond to immunotherapy so the first is melanoma. the patients that need help on the metastatic disease, they get into trouble, meaning it's gone elsewhere in your body, have you

a 16% survival rate. the doctor across the street in building 10, can provide a curable cure, this isn't something that we say in cancer very often, 4% durable cure rate in patient who is had a wide spread metastatic disease. the other cancer that's known to

work with immune modulation is kidney cancer. there's an estimated 15,000 deaths per year, again once you develop that wide spread metastatic disease, the your vuciphal is 5% and again these patients at building 10, durable you're in 7%.

so this is what we started with, this was the check point blockade, again in dr. rosenberg's grants. antictla 4, monoclonal antibody that blocks that signal 2, it's called ipilimumab, it's monoclonal antibody plus peptide against gp100 which is a marker

of mel an ohm a. only 14 patients enrolled. two complete responders and 1 partial response. however the accrual got for toxicity and we start to see these, derma titis where they would fluff their skin. colitis, bloody diarrhea,

sometimes operations, hypothicitis which is problems with the pitue tary and the patients would lack adrenal function and lack testosterone and be on steroid replacement. so while it had great effects you see this is a lesion in the lung here and can you see it's

gone here. but what we were seeing this, is skin and these dark blue dots those are all nutrifills infiltrating the skin. this is biopsy of colon, all these small round cells that should not be there. lymphocytes that should not be

there and when you stain for cd3, that's what all these brown dots are and this is the patient with hepatitis and it's difficult to see in this mag field functionsfication buoy see this destroying the tribe and destroying the duct and the billiary system and you stop it

and had to figure out what was going on. however, this happened. after we stopped the trial, we kept watching these patients, 2 of these patient his every single bet of disease they had disappear. so we started to investigate,

why is this happening. so we cautiously, started the trial again. we drone the dose, 1 milligram per kilogram. what we found was if you had 1 of these autoimmune events you had a higher likelihood of developing a response, in

patients with melanoma if they had a greater than grade 3 adverse event, you had a 36% chance of developing a response whereas if you did not, can you a 5% chance of developing a response. in renal cell cancer, the same thing, no patient that did not

have a grade 3 ace, developed a so then we decided we would lower the dose, let's raise the dose. we think that causing these immune reaction fist we can get the patient through them which we could. all of this could be reversed by

steroids fwe can get them through the toxicity. if it cures them are they willing to put up with that and the answer of course is they're williaming to put up with toxicity if we can rescue them from it. and so then we started the dose

escalation trial. what we found that was the colit isotope again with bloody diarrhea sometimes requiredda an operation was the most common grade 3-4 immune related adverse event and we did give steroids to manage these and while we worried because the steroids

shut down the system temporarily, they did not stop the patient from having a so if you didn't have any immune adverse event at all there was 1 patient that did get a response. but far greater numbers if you developed immune related adverse event.

so what did we do here at the nci for that. we got to the point where now it needed to go out in the world and have a multiinstitutional trial to get prepared and once that happened we kind of let it go and let them do the trial, let pharmacy do the trial.

when we followed up our patients, recently, we had an overall response rate of 13-20%. and a 5 year survival of close to 23%. and more importantly we developed alegorithmses that went out into the world. so what are any of you familiar

with survival curves? have you had enough clinical lect ours at this point. so this tells you at beginning of the trial all patients were alive. at 12 months, 1 year, about half the patients were alive, depending on what trial they

were on, now taken--they we're out here, what you have to keep in mind though, we're at 7 years and there's still patients these are patients with wide spread metastatic disease that had 1 or more other treatment before they got to us. so they already used up that

once have you metastatic disease, have you this long to live. they got that, and we've got them now living 7, 8, 9, 10 so that went out into trial and it had a big national trial, published in the new england journal and 11% response rate.

complete responses in 3 patients and you can see the survival curve here again, they didn't have as running to follow up but at 4 years they have 10-20% of patients alive and so this was approved for metastatic melanoma, the check point modulator ever to be approved

for anything in march of 2011. nivolumack, was a phase 1 dose escalation and in addition to the tumors that we knew could respond, melanoma and kidney cancer this was a trial at johns hopkins and they--a number of different treatments, mel an ohm anonsmall lung cancer, renal

cell cancer, castration resistant prostate cancer and colorectal cancer. this is called a spider plot and this is for each and every patientot trial, it marks the length of the sum of all their tumors that were measuring. and it does it as a percentage

so at baseline there's been no change. in patient who is go negative, that means the tumor has shrunk during the course of treatment and the patient who is go positive means the tumor goes up and what they're attempting to show with this fire blot is that

tumors don't go completely away, they stay away for a while. now keep in mind this, is in weeks and so this is only about about 10 years here this, is a patient with small lung cancer. something we found out is when they rush in, they actually make the cancer swell a bit.

so if this were a normal chemo therapy when we went from this x-ray to this exray on a normal chemo therapy wooled have said that patient progressed, time to twitch therapies because we knew this might happen we waited we waited and now these are gone. so the pseudoprogression happens

as the immune cells intrucks into the tumor and make the tumor go away. this was followed up with a bigger phase 3 randomized control trial where they tested--these are all patients because nivolumab was an approved drug, and it had to

already failed the control disease. so these are patient highly refractory to treatment. they got either nivolumab, verses chemotherapy of choice. this is called a water fall plot, very similar to a spider plot, except showing the course

of the entire treatment, it just shows maximum response. so any patient here that goes down to 100, those are complete responders. any patient that gets below this dotted line that's a pate partial response is a 30% decrease in the sum of the

longest disciplinary amterof the tumor and then people are declared progressive disease if they go board 20%. ot basis of this, nivolumab was approved for refractory melanoma in december of 2014. just last year. they then said, well we use

today in patients who used nivolumab, so these are patients who hadn't gotten treatment yet for metastatic disease, the reason they had to have wild-type braf, had been approved for melanoma because it works on braf mutated tumors. in this trial, the patients got

the nivolumab or dacarbazine. so you see the yellow curve are the patients that got nivolumab, and the blue curve is the patients that got chemo therapy. it's a wide split, unexpected we would see this kind of survival benefit and at 1 year almost 60% of the patients were alive

verses 20% in this arm. what's interesting though is that this bottom graph which is actually progression shows that even if you progress because this curve cums down a lot quicker than this 1 does, have you a survival advantage because it's done something to keep your

tumors under control. it hasn't mate them go away completeee but it's keeping things you should control. pembrolizumab was just a different company with control, and it was test indeed patients to bebin with, this was a phase 1 dose comparison.

you know phase 1, phase 2, phase 3 ff they discuss that. so phase 1 is looking for safety, phase 2 is is looking for efficacy and phase 3 is to show that it's currently bet they're an what's availae. so this is the same drug at 2 different levels and it shows

approval between the 2 but motier importantly did shows 50% patients alive at a year so that's pretty good. so they said you know what in this looks good, keep going. do a bigger trial. we're going to go ahead and approve you for refractory

so they did the bigger trial. again dose comparison but also versus chemo. but the curves aren't as impressive as they were in the beginning but still better than chemo therapy. so then they try it again much like the other 1, well, what

happen fist they give it first. so they tested pembrolizab, verses the anticdla 4, at 2 different dosing schedules and you can see there's a difference in the response rate. in a large number of patients again the survival curve you can see here, again though, you saw

and it was higher in the pembrolizumabaise arm. so to approve antibodies interfere in 2 different ways so antictla 4 blocks this interaction, we have tumor on the side, t-cell on this side. antit-cell works here and it works here and here.

why don't we use them together, so 2 different mechanisms. so they did that. so previously untreated patients which is now phase 3 randomized control trial 1 verses the other, vers the combination, what they did do for the prial, however and you had to have pdl

1ot surface of the tumor so that's selecting for a slately bter response rate than you might expect. it might be hard to see this but the response rate in the volume alone was 44%, the response rate with ipilmazab,--however the adverse events went way up in

the confrirmation so you went from 16, 20, which is tolerable to over half the patients in the combination arm getting a serious side effect. this is just looking at those patients in a different trial who were defined as pdl 1 positive, you can see those

patients seem to have a slightly higher response rate, but it's not a perfect biomarker, so even those patients that are pdl negative enjoe a high response rate. so why melanoma, this is something we've been trying to figure out for a while.

if you look at suffer from stuff coming out of the ccga, you can see melanoma has a high level of nonsinon mouse mutations so that's when you look at the tumor gna, tumor dna and you look at normal dna, and you compare the 2 looking for a change in the dna that leads to

a change in the protein that's coded so nonsinon mouse mutation this, is not a logarithmic scale and i tell you from the work we've done across the street, we get 800-900 patients permelanoma that resilience sectional so quey figured because we're breaking cell tolerance our

immune systems know how to find these mutations, they come up in the patient's tumor, maybe that's response to the t-cell maybe that's what's doing? so what are highly mutated tumors, so can we test those in the environment? so nonsmall lung cancer, can you

see here we're no longer talking about rare diseases, 160,000 deaths a year from nonsmall cell lung cancer. worst prognosis for patients with wide spread disease, only 2% alive at 5 year survival and a high correlation between smoking is a number of

mutations. another group of tumors that might be worth while looking at, tumors with mismatched repair deficiency so these are tumors that either have a germ line mutation in 1 of these 4 genes, plus or mine us a couple others, that make it impossible for them

to repair their mismatch during the combination, there's an additional group of patients that develop this sporodorsically, and it's a familial syndrome that passes down. and a third type of bladder cancer, only about 16,000 death

per year but it's highly lethal once it's metastatic. bladder cancer because every noxious thing we eat or drink gets filtered through our kidney and ends up in our bladder so that's potential causing mutation specialization of specific endothelial we decided

to start testing tease drugs in the diseases, so for nonsmall lung cancer this, is nivolumack, this is through 2 other treatments this, is third line salvage kind of chemo therapy. they treeded 117 patients and were able to get you know a descent 14-15% response rate and

we can see the water fall plot here substantial shrinkages and the patients in the blue are all they did a abdomen order of micronsized control trial with the same antibody and it's not as impressive as the other 1 but the curve definitely split here at 1 year, have you a 42%

survival verses 25% survival. this was approved for refractory nonsmall cell lung cancer in march of this year. pembrolivumab, 495 patients south america of these patients had never received patients before. the patients that were smokers

as you might imagine at mutation rate his a higher response rate, 22%. those that never smoked or at least reportedly never smoked had a 10% response rate. --stopping the check point blockade o these 3 parallel cohorts, these 2 are colorectal

cancers, those in aa patient whose mismatch repair genes are proficient, those who are de-ficient whether it's germ line or inherited and those other cancers ovarian and so focus on the colors for a second, proficient is in the red, deficient is in the blue

and black. can you see almost all the patients with the proficient accident don't respond to check point blockade. they don't have the same nur of mutations. whereas all those this the blue and the black seem to be

experiencing a decrease in their tumors. again, more fuel to the hypothesis, this was all based on antigen, but again we're building a lot of circumstantial evidence, and no 1's been able to elucidate the mechanism and why this happened.

interestingly though when you look at where the tumor and the stroma interface so this is traditional staining of a colorectal cancer and tumor is on the top left--my left, normal tissue in the bottom corner, this is staining you can see the tumorerous glands, these are

ordered out of control and this is nice, normal smooth this, is sustaining for pdl 1. and almost all the reackivity is right on the edge of where the tumor hits the normal. this staining cd8, again, allot edge of where the tumor hits the normal.

can you see here in--mismatched proficient colorectal cancer, almost no pdl expression at all. what we do eye lot and take out these tumors and we look for the lymphocytes that grow in them. you can see the mismatched sufficient colorectal cancer who is would have more mutations,

have's attracted more lymphocytes where the 1s that are proefficient don't have any patients, don't attract the lymphocytes have a smaller volume. so bladder cancer, again,--very small numbers only 29 patients, but at 12 months, 40% alive.

fairly remarkable. this is not fully mature data this, is present indeed abstract from asco. tremelimumab is another form, and it failed its clinical trial. it did not get approved for melanoma, it did not get

published, although you won't find it from the a lot, from the survival damage. on a new front we're starting to look at breast cancer issues previously thought not to respond to immunotherapy. but triple negative breast cancer has a fairly high amount

of pdl-1. 59 of the samplele studied in this particular test had pdl 1, they treated these patients, they had a response rate of about 18%. there's a lot of controversy though on how frequently you should expect to find pdl

positive 1 tumors in breast cancer because another group coming out of m. d. anderson and they're only about 20% of triple negative breast cancers. but you'll see this repeated over and over again and about different histology someone between 15 and 20% of patients

will response to therapy which is remarkable. gastric cancer again a difficult disease to treat, almost uniformly fatal. again 40% of them were positive for pdl 1. they had a response rate ever about 20%.

and again, 25%, 20%. no interestingly again with the theory that mutations that cause this, if you have hpv as the dominant drive for why you have the squamous cell cancer you don't need a lot of cancers to do and if you don't have hpv, can you see the response rate is

higher in the patients that are hpv negative than hpv positive and again circumstantial, but more evidence to put on the side that mutations are what we're reacting to. sethis is the pipeline for check points. these are all--well, not all,

all the 1s we know of interacting between a t-cell and an antigen presenting cell or a tumor, here's our main signal 1, pc r. with an mhc class 1 or 2. , all these have yet to be investigated. tc la is here, it's here,

interestingly, tdl1 has another receptor that we haven't identified yet. cd80 which normally acts as a korescepter for cd8, actually acts as a ligand, so there are drugs coming out of the pipeline, try to break each and every 1 of these interactions.

so antipdl1 we block today on 1 side and what happen fist we block it on the other. again about 40%. they got a response rate of 36%. bladder cancer, again the water fall plot we discussed before, the patients in green have high levels of pdl-on their surface

and they're showing the best so these are the drugs currently blocking the pdl, pdl1 pathway, and melanoma 32%. nonsmall lung cancer 17 to 30% none of these have been approved currently. fda approved, this 1 will be decided in a few weeks.

the current state of where we are with check point blockade in the clinic. so, again, check point mottulation, these are our targets and this is where the next generation of immuno oncologists will be working. there's a whole the information

i just through at you. but i'd be happy to take any questions. >> [indiscernible] >> so our theory at the current time is that when in the course of your life, you have encountered this antigen in the past, it is likely just been

shut off, so it's become anergic and that's through negative selection in the thymus, through a number of mechanisms. but it's still out there, it's floating around. if you look at the genes that help develop the tcr, there's a joining region and cd-3 region

and tcr alpha and beta and within those you have multiple different ways, you could literally with the content of worn person's dna, makeot order of 10 to the 11th different ccrs. and so, we have the capability of creating all those ccrs if

they don't encounter an antigen, that cell dies. so that t-cell creates a ccr, never encounters antigen, it dies. the dna gets read eye different way tdoesn't encounter antigen, it dies. and they keep creating cells

until nay find 1 they can engage with and it gets activated or becomes anergic. it's stops that initial anergic reaction so it may lead something to apopitose that shouldn't have where cdl 1 gets shot down at the second step, does that make sense.

>> okay. >> right. >> [ inaudible question from audience ] >> it is being regenerate so anytime a lymphocyte can regenerate and make pc r it walks.

so it doesn't seem like it would connect more frequently than it does but the way epigenetic modulation gets in 1 family and 1 certain spot, it all depends on what your mhc is as to what you can recognize in the pocket of an mhc molecule and so there's accord nation between

what ccr alpha and beta genes get turned on depending on what h, la you have, because if you have hlaac1, you won't make pc r that don't recognize hlaac1, so we don't fully understand all the mechanisms but there's enough there to cross talk. any other questions?

inaudible question from so a lot of that work has been pushed forward by tom's group in chicago drying to create a tumor that as a more inflamed look. so they joke about trying to make all the tumors look like melanoma so what can do you to lead to and increase and

infiltrate the tumor because if did you that, you can use the immune modulating things again and there's john hopkins where a create a more inflammed phenotype and if they can do that, they want to try to use the immune mote modulators against by creating an inflamed

tumor then they can turn down inflammation and kill the tumor at the same time. so converting a inflammed tumor to a inflame 1. >> this biopsy was done with pemeberlmab, this was done prior to apr, i think. in their supplemental figurures

you can see pdl 1 upregulationed and this is a pree treatment biopsy. [ applause ]