Can inhibit cancer cell growth

>> good afternoon, everyone. welcome to all of you here in masur and all those watching over the web. today is 12/12/12. all of you numerologists must be having a special day. somebody did a little analysis here, maybe it was jackie, to

figure out just how often will that happen, where we have a date like that and it's also a wednesday afternoon lecture? well, of course we aren't going to have any more dates like that for the rest of this century, because once you get past 2012, well, you're into those dates

that don't have months that go that high, so this is the last of the nnn date, but actually we won't have another wednesday afternoon lecture that's like this until april 4th, 2204, because the others that happen before that don't happen to be wednesdays except for one that's

veterans' day and we probably won't have a lecture that day. so live it up, people! this is a special time. and i guess for all of us, have a little tinge of the idiot savant, this is a good time to think of those things. well, that was a silly comment.

a more serious comment, this is a special day and a special lecture, because today at the wednesday afternoon lecture, we have the annual margaret pittman lecture, which has been, as you can read in your program, something that's been in place for quite a few years with a

number of very distinguished presenters, including today. and established in 1994, to honor dr. margaret pittman, who was nih's first female lab chief, and who made significant contributions to bacteriology and vaccine -- something which now happily we have a vaccine

for. and a wide variety of other contributions that she made at rockefeller and then here in the area of vaccine development for such things as pertussis and tetanus and a long career here. so we honor margaret pittman today by remembering her in this

lecture, and we always seek to identify a lecturer who represents that tradition of excellence and who's also been particularly importantly involved in mentoring, mentoring especially women scientists who have contributed to our field. and jennifer grandis, our

speaker today, very much represents that tradition and we're delighted to have her with us. she did her undergraduate work at swarthmore college. i notice a joint major in biology and art history, so this is a renaissance person.

got her m.d. at the university ouniversityof pittsburgh schoolof medicine, and remained at the university of pittsburgh to this day in a remarkable career carrying her through to training in surgery and otolaryngology. from there, to assistant professor, associate professor

and by 2004, full professor and now distinguished professor in the department of otolaryngology as well as assistant vice chancellor for research, program integration and health sciences. she's received a number of distinguished awards. i have to mention the provost

award for excellence in mentoring, and she's a member of the institute of medicine. she has to her credit, and we should express our gratitude, been very gracious in extending service to nih. she has been on lots of study sections.

i think some three pages of her c.v. goes through this, and we do really appreciate that kind of hard work, including being a member at the moment on the board of scientific counselors for nidcr. and the recipient of numerous nih grants.

her work has been consistently focused especially on cancer of the head and neck, and any of you who are ib vofle involved inclinical care will know those are some of the most difficult and frustrating illnesses to wrestle with, because of the fact that these in general have lacked an

effective treatment, that they affect a part of the body often involving swallowing and speech that makes it particularly challenging for patient and physician to figure out the appropriate management. but she has contributed to this in very substantial ways, to

carry the understanding of these diseases into the molecular era, and she will be telling us today, i'm sure talking about egfr and a variety of other pathways. the title is targeting ongoncogenic pathways in head and neck cancer.

she is going to speak, there will be a q and a, there will be a reception in the library with refreshments, but i think i should get out of the way and ask all of you to help me welcome dr. jennifer grandis, the margaret pittman lecturer for today.

please welcome dr. grandis. >> well, thank you. it is so wonderful to be here. i'm honored to have been invited to deliver this lecture, and i'd like to start by really thanking the nih for being so supportive of me and my career. it's a really atypical woman who

chooses a career in surgery and then chooses a focus on science and most people feel that's a bit of an oxymoron, and i have to say that support from the intramural and extramural program through clacollaborations and indications to participate in the review process and to

help design programs has been invaluable to my career development, and i'm most grateful. thank you. i'd like to take you on a little bit of a journey today and really emphasize the buy directional dynamics of

transizational cancer research. as a person trained primarily as a clinician it's difficult to do anything but translational research although now it's a popular buzz word. for those who aren't intimately familiar with head neck cancer, i'd like to briefly describe my

chosen focus and then emphasize all of the principles that we'll discuss in more detail are really applicable to virtually every malignancy but potentially lots of other disease processes. until recently, we really believed that most of the cancers in head and neck were

associated with exposure to tobacco and alcohol, increasingly there's an epidemic of human papilloma virus associated head and neck cancer, primarily in north america but i'll tell you that's because we're testing for it here so i've recently been to india and

china in the last couple of years. they deny having any hpv head and neck cancer but if you ask them if they've looked for it the answer is no so it's hard to know what to make of that assessment. still today, the treatment is we

cut it out and deliver adjuvant and therapy like radiation or radiation plus chemotherapy or, when cutting it out is just too morbid to imagine in terms of loss of function, we deliver combined chemo radiotherapy, chemotherapy by itself is still not particularly active in head

neck cancer. even when we cure a patient of their initial cancer, there's a very high rate of recurrence and second primary tumor formation, so the statistic that really illustrates the morbidity is that if a patient survives their first head neck cancer, they

will usually succumb to a second cancer of the air or digestive tract. despite the tremendous work in the field, the only effective chemo prevention strategy thus far is to stop smoking, we'll introduce much more effective agents for head and neck cancer.

these are the anatomic sites, the sagittal mri. i'm primarily not going to focus on nasopharynx, but focus on the oral cavities pharynx and larynx. now if you lived in europe and you wanted to buy a carton of cigarettes, this is what they

would sell you. and in our good government's wisdom, this is not allowed because it infringes on the freedom of the tobacco companies, but this was thought to be a potential deterrent to smoking in europe. however, there's a cottage

industry, i understand, of very nice looking holders, so you can buy a package of cigarettes that looks like this, and you can put it in something very pretty so you don't have to remind yourself that this is what will happen to you if you continue to smoke.

so from my perspective, the gaps in knowledge that have really driven our program over the years is we still have absolutely no idea what to measure in a patient's tumor to tell us how best to treat the patient. so there's no predicted

biomarkers. we don't really understand treatment resistance, it's not very well-defined, therefore, we can't really elucidate the resistance mechanism, even though we understand that hpv is a very different cancer, and i'll show you some data about

that. we still don't have an hpv selective therapy, i suspect that one day if we would vaccinate all of our children, we might not have hpv associated head and neck cancer, but that is clearly decades from now. i'll tell you a little story

about how our genomics is revealing new targets, new pathways and new ideas about how we can really design innovative and more effective trials, but i think the real challenge is to elucidate these groups among the tremendous genetic heterogeneity with the understanding that many

of our subgroups are going to be small but quite meaningful. so i think before we undertook genomics, this was just a sample trial with a sample of some of the agents that are under clinical development, and you can see it's most of the usual suspects.

there are receptor kinases, non-receptor kinases, signaling pathways and many agents that have been developed to block different components of this pathway, but i think it's a little playing with the dark because essentially we developed hypotheses, interrogated cell

lines and human tissues but only in the context of really validating our selective hypotheses. so dr. collins is right, i'll start with the egfr receptor, what i started working on many years ago, and as most of you in the room know, it's a growth

factor receptor, it's overexpressed in many cancers, certainly in head and neck cancers. it's really not mutated in head and neck cancer. i'll tell you a little story of ovarian 3, and it's the only validated molecular target and

what i mean by that is fda approved -- for head and neck cancer patients in 2006, so now it's in our clinical arm but we have no idea who is best treated. there are no predictive this was a new england journal paper, it's been disappointing

and it's been disappointing because we really don't know precisely what it's doing to the tumor, we don't know who should get it, we don't have positive data with the kinase inhibitors, and disappointingly last year there was a report of the definitive trial of chemo

radiation plus -- versus radiation alone and the addition was not substantial in terms of improving survival. so we simply have lacked models to be able to test for -- resistance, so to that effect, really smart graduate students took a book from the herceptin

play book and we realized that virtually every cell line that we grew in animals and that made xenographs was sensitive so clearly did not mimic the diversity of head and neck so we grew cells over time, we selected out for two models that developed resistance.

so this is the model of acquired -- resistance, and you can see down here that these are the tumors that grew over time. we demonstrated we could take them out of the animal, we could propagate them and put them back in we found there was a fragment called c11cts that has been

reported in the context of herceptin resistant breast cancer, and this fragment appeared to be increased in the setting of resistant to se tucks mab. we treated them with dual egfr kinase inhibitor, irreversible small molecule, either alone or

in combination with cetuximab. the other altered form we noted was the variant 3. this was discovered in the context of glioblastoma by darrell bigner and others and clearly it lacks most of the binding domain of egfr receptor, so we reasoned that cells that

express variant 3 would be less likely to respond to cetuximab. so to that effect, we developed a model of head and neck cancer, we have several of these and i'm happy to share them, where we made tumors that express the dirty little se little secret,it will not grow as a tissue culture.

we're hoping to overcome that with heterotopic xenograph. one can appreciate when they treat these tumors with cetuximab, they're relatively resistant but this was an experimental -- the buy directional nature when a group from germany noted that in a

phase 2 trial, that expression of variant 3 and egfr -- were independently associated with resistance to this regiment, lending some sort of credence to the idea that if we could prospectively identify patients who have variant 3 in their tumors, these would be

individuals that would potentially not be responsive to cetuximab and could benefit from her 2 inhibitor. so how do you block variant 3? it's a little challenging. the pe38 linked to -- that blocks to variant 3 specifically, there's an

antibody that was developed by lloyd subsequently at the ludwig and now through abbott 806 and it blocks both wild type and variant 3 and this is in clinical development and for reasons that aren't very well understood, in lung cancer models it's shown these

irreversible egfr inhibitors tend to be effective against verity yant 3, so that's another potential place where the biomarker variant 3 can indicate likelihood of response to these agents. so as you noticed on the first slide of signaling pathways,

stat3 is a very common downstream pathway. and i'd like to tell the story of how we got stat3, rick clows ter when he was head of nci came to visit pittsburgh and we presented our data to him, we were talking about audokrin growth and subsequent activation

of stat3, he said wouldn't you just want to block stat3, go downstream? we're like, yeah, but that's kind of challenging, the transcription factor. what really persuaded us is when we put stat3 into head and neck cancers, they became resistant

to egfr targeting, so we reasoned this was an approach worth investigating. and then work by bernie viceman and others have found that -- in primarily tongue cancers were poor prognostic indicators. but let me tell you, this is an undruggable target, thinking,

and this is only to put up to emphasize how incredibly challenging it is to block a so i'll tell you a little bit of a story. we elected to follow the play book of victor zao and try nucleotides and -- so frankly, what victor had done in

cardiovascular disease was he had developed an e2f decoy using it in the clinic at the time we got into this to prevent -- hyperplasia, so cardiovascular by pass graphs, bathed in the solution, put back into the patient, the phase 2 results were really promising, and i

will tell you, cut to the chase, the phase 3 trial was negative, but i will tell you how we've done it with stat3. in addition, though, to trying this decoy approach and lots of natural products, which all kind of work a little bit but i think we can all agree that these

natural products have many other targets and they're not really selective for any particular molecule, including stat3. so in a collaboration with the nci through the next program, we've had a really exciting run for the last 18 months or so. we developed a high content

imaging assay in collaboration with paul johnston in the drug discough reinstitute in pittsburgh and the fundamental issue is that stat3 and stat 1 have a lot of hemology but stat 1 has more of a tumor suppressor function and we don't want to block stat 1, stat3 has

much more of an oncogenic function and virtually all of the drugs that are stat3 inhibitors including the jack inhibitors block stat 1 and stat3, so in an attempt to find a stat3 selective molecule, paul designed an imaging a assay shwe screened 100,000 compounds and

initially we have a four lead, we've gotten down to probably two of these compounds and all of these compounds potently inhibit stat3, they do not for the most part inhibit stat 1, and they inhibit head and neck cancer growth in vitro, four different cell lines.

so we've got some really exciting preliminary evidence, it's probably a little bit too early to present, about the mechanism of action of a couple of these compounds. this one and this one in particular, that might lead us to select them, but we're

working with teams of chemists and teams here at the nci to really try to develop a small molecule that is not a kinase inhibitor, these are not working by inhibiting jax to humans, so give us 50 about a year. this is our decoy. this double stranded piece of

dna was dose-dependent, incorporated into the cells without -- it works in the cytoplasm, also works in the nucleus. so potentially what we could do first in animals is easy, and that is we can inject the decoy into the tumor, and i think this

is an experiment that you did, used in my lab maybe 10 years ago, did these experiments and published this paper. when they injected them with the stat3 decoy, they were growth inhibited, differs by a single base pair, doesn't bind stat3, had virtually no effect, and it

was very reassuring when john at anderson took the same decoy in his mouse model with skin cancer and could show the skin cancer was prevented largely by injection of the lesions that developed with this decoy. so we thought there was some hope for it, and again, the only

perk of being a surgeon and a scientist is you can go right to the clinic. so this is what we did around this time, fda developed the concept of a phase zero trial. they thought this would be a suitable scenario, this was the first in human stat3 inhibitor

study, so we took patients who were having resection of their head and neck cancer and we took them to the operating room, we biopsied them. then we injected the decoy, and we did the operation and then we took out the tumor, and we harvested a specimen in the

region of injection after the surgery. this was about four hours, and you can see that most of the patients demonstrated down regulation of stat3 targeting expression, in the post treatment specimen compared to the pre-treatment specimen.

so we were pretty excited about this. then a very wise grant reviewer said what about hypoxia, what about surgery, how on earth do you know that this is specific to the decoy? so we then -- we needed a control group in a phase zero

trial, and we treated half the patients with saline and half the patients with the decoy and found that in general, the decoy-treated group had greater down regulation of targeting expression compared to the saline group, and this was about 15 patients in each group and

this was great except we had to inject this into the tumor. and we knew from working with richard at vanderbilt that the half-life of this decoy was only about an hour. it was totally impractical for systemic administration. so we engaged a very talented

chemist at carnegie mellon university down the block, and with this chemist, we experimented with lots of modifications and the goal was to stabilize the molecule without losing biological potency, so we found that if we made it cyclic, and this is just

two carbon spaces on each end retaining the same sequence in the middle, we could retain the ability through fpr, we would show this modified decoy was still able to bind very acidly to stat3, so it could retain its biological potency, but we had a dramatic increase in half-life

so we're now up to 12 hours. now it becomes more representative of a drug, and it also has a much higher melting temperature of 80 degrees, so it's really unlikely to fall apart when injected into the bloodstream. and then we did the key

experiment, which is we put tumors into mice and treated them with intravenous injection of either the cyclic stat3 decoy or the cyclic mutant control, and the decoy-treated tumors failed to grow, the controlled treated tumors grew and there was significant down regulation

of target genes in the cyclic treated tumors, again, this was iv treatment so no direct injection. we published a paper just a little while ago, a couple months ago, this is what the decoy looks like, and where we're taking this is we're

finishing pharm tox studies in animal models. looks like the only thing the did to the mice was make them a little hyperexcitable, but that was temporary, and in addition, we're collaborating with a cardiologist, we've put the decoy into a micro bubble and

the idea is that we'll be able to deliver this micro bubble encapsulated decoy to the head and neck region, th viaultrasound application. so far it's working in the animal model, and lastly, we're collaborating with an engineer, who has engineered a sustained

released micro particle. our idea is that in the surgical setting, post resection, that we irrigate the wound with the decoy in these micro particles, which can be engineered so that they have sustained -- kinetics over the period of a year or two, and that might be a

reasonable way to prevent recurrence, so we're exploring different ideas. but we still have absolutely no idea how stat3 is activated in human cancers. we still remain really challenging because it's undruggable and we don't really

know which patients would benefit from a stat3 inhibitor. so just when we thought we kind of understood head neck cancer biology, we were able to engage several investigators, the team was led by eric lander, to sequence head neck tumors so at the end of the day, we sequenced

74 head neck tumors for whole -- analysis and we published this, and here is the finding that this was a typical head neck cohort that is mostly men, mean age was 58, 75% men, 80% recurrent or former smokers, and about 13% had hpv positive tumors.

we looked at hpv by several methods and i'm happy to describe that in more detail. the challenge in all surgical cohorts, and you require surgical cohorts because you need enough material for sequencing is that most hpv tumors are in the oropharynx and

the treatment of most oropharynx cancers is non-operative therapy, so even with the tcga, we have a relative underrepresentation of hpv positive head and neck cancer but it's a different disease. these are not the same tumors. in general, when we looked at

the entire cohort, the thing that jumped out at everybody and was the title of the companion paper by the hopkins and anderson and baylor group was there appeared to be notch alterations, and this was unexpected in squamous cell carcinoma of the head and neck.

18 months later, i still can't tell you what the function of these alterations are, so i'm not sure that they're druggable. but what i can tell you is that the most commonly mutated on cogene was pic3ca and otherwise it was a complex story of a lot of tumor suppressor jea genesand

here was the cure yait raited list at the time. we were interested in trying to see if there was a genetic signature that linked stat3 hyperacactivation to any kind of genetic alteration. so back to basic biology. we sequenced a lot of tumors

over the years and there are no activno -- the general thoughthas been that it's a compilation of activation of upstream receptors such as the egf receptor, pgdf receptor, but we were wondering if there were really meaningful negative regulatory proteins that by virtue of mutation were

inactivated. so we knew about this family of protein receptors about five years ago where he reported that there was an unexpectedly high frequency of mutations of these receptors in colon and in lung as people further investigated, pptrt -- stat3 was shown to be a

substrate for these two receptor cairo seen phosphatases. so specifically these phosphatases activate stat3. the role of the other ptprs has been largely unknown, there's a trickle of papers over time and almost every paper at least in the context of cancer

dem sphraits thademonstratesthese are tumor suppressor genes and there's occasional indication this they mediate stat3 -- when we look back at our cohort, i don't intend for you to look at the details here, but in our 74 patients, we found ptpr -- now with 374 tumors to look at that,

is our cohort plus -- it's a little over 40% of human head and neck cancers that appear to have loss of function mutations in the ptpr family, so this may represent an unexpected and common pathway that leads to stat3 hyperactivation. so this is just a list of the

mutations and the two most commonly -- this is where they're located and i don't mean to spend more time on it except to say that we engaged a structural biologist at the university of pittsburgh and it's been really exciting to work with them.

the crystal structure of stat3 has not been reported. we just have stat3 beta. but based on what we know, what joseph was able to do is look at all of -- this is the region that we expect binds to stat3. and this is where the mutations are, and i think you can

appreciate that the fos foe tyrosine residue of stat3 is in very close proximity to most of these mutations, so it seems to be structurally plausible that an inactivating mutation in the region that -- would lead to failure to dephosphor late stat3 and hence hyperactivation.

so this has been our central hypothesis, that stat3 hyperactivation at least in part results from loss of function mutations in this family, and we can prospectively identify patients who have those mutations and prioritize them for treatment with stat3

inhibitors as they are developed. so we made the mutations and we put them into head and neck cancer cells. i want to emphasize that in conjunction with this, what we did is we created essentially an il3 dependent analog in head and

so the model has been terrific to screen for oncogenic mutations, you take away the factor il3 and if you put in a mutation and it's an on cogene, the cells survive. but there was no such thing for head neck cancer. so we took an hp positive and

negative cell line and made it serum dependent. so i think you can appreciate when we put these mutations in in the absence of serum, we get tremendous increases in survival, so this is a dose-dependent increase in survival, when we express this

mutant form of ptprt in a head and neck cancer cell. there's hyperphosphorylation of stat3 with this -- and here is a preclinical jax stat inhibitor and there appears to be enhanced responses to at least this molecule. this molecule is off target

effects and it's not perfect. then in an effort to find more relevant models, we were able over the course of several years to learn from people and other cancers, and we're taking tumors from patients and we're growing them out at heterotopic tumor grafts in mice, so these are

tumors from patients whose tumors have mutations in these ptprs, they have hyperactivation of stat3, and you can see that they respond to treatment in this case by azd1480, a jax stat inhibitor that's in clinical development, so we are encouraged that this

might represent a plausible model for testing drugs to translate to humans. so this is our idea about a clinical trial in a window setting. we would screen patients for mutations, we would enroll them, we would treat them with this

jacks stat inhibitor, we would -- before surgery, this is a few weeks, they would perform operation. again like the decoy trials, we have the biomarkers, the mutation status at a base like, so we'd know whether or not it was a predictive biomarker.

but i also want to remind you in addition to stat3, we have this usual suspects. so what does this mean in head neck cancer? i think it's really important because what we found is when we looked back, and this is about 165 tumors we've done in

collaboration with gordon mills and n.d. anderson, so we looked at mutation and gene amply faition and we looked at -- for expression of proteins that we might hypothesize would be correlated with -- amplification. and what we found is that pic --

pi3 kinase alpha subunit and -- forms of akt were highly correlated, the levels were highly coordinated within the tumors that had mutation and the emerging story is that the frequency of this mutation with hpv negative head and neck cancers is about 15% mutated,

another 5% amplified, if it's hpv positive, it's about 30% mutated, and another 15% amplified. so it looks like right now, again, there's only about 50 hpv positive tumors in the tcga pittsburgh cohort, university of chicago has also looked at hpv

positive tumors and found this very high frequency picc3h alteration in head neck cancer, so this could be the predictive biomarker for hpv disease because quite frankly, in the world of clinical medicine, all we're doing today is we're dialing down the intensity of

treatment for hpv-positive disease, we're giving them less chemotherapy or less radiation, but we're not giving them anything that would specifically harness the pathways that are turned on in their tumor, and if one could really cure hpv-positive cancer with a

pi3 kinase inhibitor or -- or something like that, that would be really helpful for this population. so our overall approach has been to create this model that i've showned you with respect to the ptprt. we have the genomic screening

platform. we're now making all of these mutations and putting them into this platform, and then we're looking at drug screening efforts. so here's what we've done with the pi. c3ca.

this happens to be an hpv positive model. you see either amplycation or mutation is driving growth and it's driving signaling and this is also true in hpv-negative disease. and when we treat with either compound, now bez235 is an m --

pathway inhibitor that was developed by novartis. it's not being further developed because of toxicity. px866 is a pan pi3 kinase inhibitor developed at our institution, now being developed in head and neck and lung and we're doing a phase 2 trial --

and we have very similar and compelling data with other pi3 kinase pathway inhibitors, and what i want to emphasize is these are now endogenous mutations. so these cell lines have wild type and this has mutant. wild type and mutant.

so growth inhibition is dramatically enhanced in cell lines that have naturally occurring mutations of pic3ca. so here's the bottom line numbers, the numbers of amplification, positive versus negative, and then another 10%, 15% has amplification.

so this is where all the mutations are, this is interesting if you look into the tcga cohort. this looks like breast cancer. it's the only other cancer that has mutations throughout the gene. we're not sure what all these

mean, so we've made all of these and we're systematically testing them. these are previously reported hot spots, they account for about half of the mutations in head and neck cancer. in addition, though, to pic3ca, i want to emphasize that there

are mutations in other genes in the pi3 kinase pathway, a handful of the other pi3 kinases as well as m-tor. so these are the agents, a couple of the agents that we're looking at. all of them are either approved or in clinical development and

we're trying to develop will targeting different nodes in the pathway is more or less effective in conjunction with a specific mutation. perhaps most exciting for us this is now our heterotopic tumor m model. this is a patient whose tumor

had a pik3ca mutation and they're hpv positive. when we treated mice with this tumor, with the bez compound, they went away. the tumors disappeared. and this tumor was pik3ca mutant and hpv-positive. in contrast, a tumor that had

low -- akt this, is the patient's tumor, pik3ca wild type didn't have very much of a response to the pi3 kinase pathway inhibitor. this patient's tumor was hpv-negative but had the high phosphorylated -- and a mutation and it also had a dramatic

response. so what we think -- what we believe is if we can assay patients either for the mutation or even you can see just look at phosphorylated akt, it's a very nice surrogate marker for activation of the pathway, this would simply be used to enrich

clinical trial populations so that we're giving patients that have tumors like this drugs that are affected and we don't force patients that have tumors like this to suffer the toxicity and delay effective treatment because they're probably not going to respond.

so in collaboration with the university of chicago, we're doing a phase 2 clinical trial. in this case, we're using a novartis compound, this is recurrent metastatic setting, and the key thing is biopsy. so baseline biopsy, biopsy after single agent, biopsy after the

addition of ctuximab. the reason, it's real world clinical medicine and this is f. da-approved in this setting and it would be really difficult to get this through ir bs without the opportunity to retreat with cetuximab. we have modest evidence this is

far more effective than pathway targeting alone. but we at least hope in the context of sequencing and fish on all of these specimens that we can begin to test these hypotheses to really broadly apply more commonly. so in summary, ur r i think i

want to emphasize that understanding both the signaling mechanisms of any model disease, whether it be head neck cancer or otherwise, but also look at the basic biology. it's daunting, it's overwhelming, it's confusing. most of these mutations are

passengers, they're not drivers, but if we begin to put our understanding of biology and the enormous data resources that are being delivered by the tcga, we may be able to quickly identify mechanisms of resistance as well as new targets, patient specimens and relevant

preclinical models are really critical in this effort. the theme so far is -- receptors and -- downstream signaling pathways may be helpful in the setting of cetuximab resistance. there's many therapies in the early stages of development, we can't possibly look at them all

in a systematic way, we'll run out of time, so i think the challenge is really to begin to highypothesize what are the features in the patient's tumor that would predict which of these therapies they should receive. i'll stop by saying i don't get

to do anything except come here and have a wonderful time. there's a group in pittsburgh, clearly the tcga, and my colleagues at other institutions as well with my collaborators at the broad and lots of resources from the nci and the nih. [applause]

>> thank you, dr. grandis, for a wonderful tour through a lot of interesting pathways and potential actions. so we have time for questions. >> how are you? >> good to see you. >> you're not playing rock music.

>> francis didn't ask me. >> there you go. next time. i have heard it. >> but i will at any time. so in our business, we've debated back and forth the relative merits of blocking at high points in the pathway with

jax or downstream and with a stat, and there's reasonable arguments for both. >> yes. >> this is sort after two hadtwo-part question. in patients who don't have the pi kc3 mutation do you see activation of akt, suggesting

that there's auto krin production of cytokines that could be activating -- >> absolutely. >> and then even though maybe it doesn't make any sense to target both jax and stat, have you ever dumped both of those things into an assay to see do you get

synergy in any way? >> that's a really good you know, john, all the jax inhibitors block stats, right? >> but they also block -- >> lots of other things. so if we block stat downstream with jax plus another mechanism, is there any benefit to that?

the answer is we haven't tried, the challenge is using molecular strategy which is not particularly effective -- it's great as a tool but we haven't tried the decoy in conjunction with the jax inhibitor and that probably makes sense to do. >> the hpv positive patients

have a much better prognosis, right? they are much easier to treat, almost 80%, don't you think the big -- don't you think -- >> it's really interesting, i've now heard this in many settings, most recently in a summary statement.

why bother studying hpv positive head and neck cancer? it's all cured. i can tell you as a clinician, and i think the clinicians in the room can tell you, there's not a week that goes by in tumor board that we don't see an hpv positive recurrent tumor.

so in general, these patients have a better prognosis than hpv negative disease, but there are still patients with hpv positive disease that recur, and the question is, are these the patients -- is this the 30 to 40% with pik3ca mutations and the answer is we don't know yet

because they haven't been studied. -- will have impact in both of space of hpv positive head and neck cancer and hpv negative i was meeting with pat this week to really try to understand, i'm not a virologist, what is it about human papilloma virus

infection that would essentially predispose the selection of alterations in this gene, whether it be amplification or mutation, and the disappointment, of course, is it's not only an hpv positive cancer, it's a negative cancer too, so i think it's going to be

a complicated story, but worth considering in both spaces. >> so do the hpv positive patients have -- 53 mutations? >> never. >> virtually never. at least in the tcga, we haven't found in 370 some cases, we haven't found a single case of

p53 mutant and hpv positive. now again, that's 30 some cases there, then we have another 12 -- it's a relatively small number, but i think to date, there's probably maybe 108 -- positive tumors that have been sequenced and none of them to my knowledge have a p53 mutation.

a follow up on john's question. in several examples now of molecular targeting, you do well for a while and then you escape. >> and so why not already go to a strategy in which you actually look at the network connectivity here and see where the likely bypass pathways are and do --

treatment right up front rather than going down, letting everybody escape and try and catch up with it. >> i think there's two -- number one is to have models that are relevant to human beings with cancer to understand the escape mechanism.

i think our cell lines are in imperfect and we have to -- getting them from both primaries and recurrences ideally the same patients and growing -- we can understand the resistance mechanisms for targeting. the second challenge which we have been able to overcome but

it's not trivial is that sometimes your ideal cocktail is manufactured by different drug companies. and we all recognize this and we all try to deal with it but it's sometimes proved really challenging to get everybody in the same room, particularly at

the end of the day when we don't necessarily have an economic argument for why they should do this if it's a very small population of a relatively uncommon cancer, it doesn't get their attention. i recognize that as a challenge but it does mean our tool kits

for our plea pree clinical models sometimes have to be modified when we go to the clinic because engaging the people who have the drugs we can give to people is so difficult on levels that are not scientific. >> francis is going to solve

that, right? >> well, it is certainly not just unique to this situation but in so many other circumstances, so probably cancer particularly, where you'd like to do combination trials because you know from everything we've learned in how we've

succeeded in hearing things like leukemia, lymphoma, a single agent isn't going to do it. why would we think that it would be true in some of these very difficult to treat cancers, yet there are all these barriers, some of which relate to commercial situations, some of

which are regulatory. it is a major topic of discussion in the national center for advancing translational sciences but i would not tell you we have the answer quite yet. >> so jennifer, i was wondering whether you could give me a feel

for what percentage of the tumors have both activated stat and activated pi3 kinase atp pathway? >> it depends on how you measure it. i think that the best way of figuring that out right now is through the rtpa data of pcga,

and i don't believe that all that data has been uploaded and publicly available. we have a couple of heat maps that are coming out of the paper right now, it's being written, and the answer is going to be not everybody and not necessarily the same tumors.

it does look at least preliminarily that the activated akt and the activated p stat3 are not the same subgroup, but that's not always true. so i will -- when that data is locked in and finalized, i will let you know. >> somewhat more general

question, in your last slide where you talked about the designing of your clinical trial, you referred to serial biopsy, and it sort of reminds me of serial imaging. how important is that tool in general going to be to -- as a strategy to kind of dealing with

these diseases? >> that's a great question. there was a really lovely editorial i think within clinical cancer research about -- somebody did a study that was really illuminating about the number of studies that were reported at the acr or asco

that reported to have -- biomarkers, then when you looked at the publication, own 10% of them ever reported -- biomarkers which meant that people were collecting material and either doing nothing with it or more likely all the data were negative so they weren't putting

it into the publication. so i think that it reminds me that you have to have a question that you're asking in order to interrogate the biospecimen. so i think from my perspective, the most meaningful thing is what happened in the context of treatment resistance.

so if you biopsy a patient at baseline, then you give them a drug, if nothing happens, nothing happens, then that tumor grows, it's really important to understand what's happening in that tumor, and that's been really elegantly shown by jeff engelman and others at mass

general, really ee lewis naiding novel mechanisms of treatment resistance. so i think that if you have the context, it's really important to do this. and the second thing is a high quality tissue is really challenging to get, particularly

if you're talking about a fine needle aspiration or an interoffice setting, and it requires a team and a commitment, and frankly, i think it also really requires that we educate the patient and engage them as our partners, because it's not clear to them that it's

going to benefit them. and it's usually inconvenient and sometimes it's uncomfortable, and depending on the context of the study, it may or may not be compensated, especially if we're talking about going back to the operating room, so i think we've

paid an enormous amount of lip service to serial biopsies, and i think that they're critically important, but you have to be asking an important question that when you answer it with that biopsy, it will change what you do next. and i don't think we always

think that through. >> these have been wonderful questions and answers, and we will have a reception now in the library with some refreshments, if you have further discussion you'd like to have with our speaker, please come, or please come anyway.

and let us one more time thank our pittman lecturer, dr. grandis.