Basal And Squamous Cell Skin Cancer

today we have one of the most exciting timely topics that one can think of in medicine and biology, and that's sort of where genomics enter -- interfaces with disease. we are fortunate to have two investigators from here at nih

whose work has been at the forefront of this. somebody asked me the other day why do you keep showing a picture of the brooklyn brid and i guess i could put a picture of my wife or grandchildren there but it's one of the objects that i find

myself in love with and symbolically it represents the purpose of this course. if you're ever in new york and want to do one of the most exciting things and it won't cost you any money just to take somebody you want to be with and walk across the brooklyn bridge.

there's a wonderful bar under it on the new york side and a wonderful restaurant on the brooklyn side and the view is unbelievable and the history is fantastic. but it's to connect two words and that's the purpose of all this.

so if you could, jeff, could you flip on the first slide. so what's so exciting about today. from my way of thinking, what sort of provoked this was reading a bunch of editorials and commentaries by learned people about the question of, is

there a new cancer paradigm. the old paradigm being based upon the fact that diagnose to a great extent is based on morph logic criteria. so if a cell is malignant and it's from a colon for example, it's a cancer of the colon. and it was based on these

criteria which were graded by severity. but now it is observed repeatedly that the clinical course, the response to therapy really didn't correlate all that well except maybe in a big statistical sense among groups of people on an individual

basis. there was a lot going on there that wasn't revealed by conventional or even specialized immunocytochemistry. and then when one enters the genomic era, this issue arises of whether this tumors which are expressing different genomic

patterns, not uniform. it may be a colon cancer in one patient expressing one pattern of gene expression but in another there were other genes, other mutations, a whole new level of complexity that's arisen. and out it has come the question

of whether genomic and other analysis of tumors actually provides more accurate diagnosis and permits individualized treatment and some prediction as to the clinical phenotype. what is this tumor going to behave like. that's an exciting frontier.

there are no answers at the moment, although our speakers are best suited to comment upon this, okay. so the subject, if you could show the next one, jeff, is a very extraordinary disease probably a fair amount about melanoma or malignant melanoma.

but it has some clinical mysteries connected with it. so here's a case report of a person very well-known to me, happens to be a cousin of mine who died a couple years ago, he was in extraordinary good health. he was an athlete, played tennis

up until the time he was 91, two or three times a week. he was in great shape and took very good care of himself in all ways, nutritionally, all sorts of good things. no bad habits. and at about age 74, it says myeloma, it should say melanoma.

he had a spot on his right shoulder. it wasn't very large, maybe one centimeter or something like that and it was a malignant mole, a malignant melanoma and the surgeon did what was to be done was to remove large margin around this.

and then subsequent analysis studies done over an annual basis by very competent centers. this was is he university of cincinnati had follow ups including scans of the abdomen, various blood tests, all sorts of things. and they were all negative.

and this was done routinely. at age 92, he observed that he was losing his memory. and everybody said you're 92 years old, you can't keep doing this forever. that went on for a little while but he kept getting a bit confused.

that wasn't quite like him and eventually he was worked up in general and nobody thought that this had anything to do with a melanoma. he had a convulsive seizure which led to a neurological work up and it found he had three roughly one centimeter tumor

masses in his brain mainly in the frontal and parietal lobe. this is biopsied and it was malignant melanoma and they had tissue from the original skin biopsy taken when he was 74 years old. and the pathologist and all at the university of cincinnati

went over it and said to the best of us there are no -- he had no response to radiation or medical therapy. he was kept comfortable, hospice, coma and he died. there were no reasons found clinically in his bone or his liver or any other tissues.

now, this is one of the great mysteries of this disease is that people can have diagnosis made and the subsequent course is highly varied. this is the most extreme i've ever heard. maybe you'll comment about this. but it seems that 18 years is a

long time, but nevertheless one can't escape the fact that this tumor was doing something for 18 years. it's highly unlikely that the primary melanoma was developed in the central nervous system. that's odd. [indiscernible]

[no audio] are there diagnostic genomic or other signatures which predict a clinical course highly variable in patients [indiscernible] this is one side of the bridge. the other side of the bridge is of course where the basic science advance is influences us

in terms of new ideas [no audio] after a post doctoral fellowship he came to the nih where he has been in the surgical branch of the national cancer institute. this branch is known as being at the forefront of the development of techniques based upon cytotoxic t cells to attack

malignant diseases. the case in point today is malignant melanoma, about which will be commented. he has been collaborating in the past year or two i guess or more with our second speaker who is yardena samuels who was in the genome institute, i guess still

is in the genome institute, but officially she is now an associate professor at whitesman institute in israel. so yardena's training was in cambridge. and then [indiscernible] at the hebrew university. she received her ph.d. from

imperial college in london following which she had a highly productive and very exciting post doctoral fellowship with the vogelstein group at johns hopkins. and she has been at nih as a tenure track investigator. as i mentioned to you, after six

years in that position, will be leaving us to return to israel. we trust we will see and hear more from her during the coming so it's a great pleasure to have you both with us and i'll be quiet. maybe you want to begin, paul. >> thank you for inviting me to

talk today. i encourage people to ask questions [no audio] during the seminar [indiscernible] first i'll talk about melanoma prevalence and etiology, a little bit of background on the disease. [indiscernible] non-synonymous

protein present in teurmtz. so, melanoma is a fairly prevalent cancer. this list here shows you the number of new cases of a variety of tumor types in males and females. melanoma of the skin here is number 5 for males and number 6

for females. there are about 74,000 new cases per year predicted for 2012. and this actually hasbeencreased so the numbers are going up and actually they've been going up steadily for a number of years as we'll see in the next slide. and over 9,000 deaths per year,

and that number is also increasing despite a lot of work by us as well as others to develop effective strategies for treating this disease. so as i said, the melanoma rate has been rising. this graph here shows the trend over more than 30 years for

melanoma as well as a variety of other cancers. there is a lot of variability in some cancer such as prostate but that may reflect the frequency of screening for prostate cancer which of course is often done by, which is done by screening psa.

psa levels. you'll notice that some diseases, the frequency of some diseases are actually going down. lung cancer and colorectal cancer is coming down presumely because again of better screening.

lung cancers have been going down actually for a number of years in men and they are leveling off for women which is probably a reflection of the fact that men had been smoking cigarettes for as long as women and women are just catching up and presumably or hopefully this

curve will be trending down very soon for women as well. melanoma actually has shown a fairly steady increase over the past 30 years, and in men as well as in women. so it's a serious disease, malignant melanoma, metastatic melanoma is a very serious

diagnosis. so we need to come up with better ways to treat it but we also need to come up with ways to prevent people from developing this disease. so there's a number of ways to accomplish this. one is go to see your

dermatologist. certainly stay out of the sun, certainly avoid tanning booths. it's very clear that these things contribute to melanoma, and that's evidence that the genomic level as yardena will discuss in more detail, the actual patterns of mutations

have a signature that is known to be caused by ultraviolet radiation. there's no question that uv rays contribute to melanoma. of course it's a complex disease like most diseases so the environmental architects as well as genetic factors influence the

development of melanie ma. but nevertheless that's certainly a contributor. now there's a number of factors that dermatologists take into account when trying to decide whether or not a mole which is shown here on the left side is truly a mole or is a malignant

one is asymmetry and this is a fairly round shape here. the shape is very irregular. the border, this one has a very clear border where this one does not. the color, the color of a mole is generally union form where a malignant tumor is not.

again, notice just simply color alone, a darker color does not necessarily, is not necessarily diagnostic for melanoma. in fact some people have very dark moles as well. so a number of criteria really has to be used to distinguish melanoma from a benign mole.

certainly the diameter is diagnostic but one of the most important characteristics that's used in diagnosing melanoma is evolution. in other words, as these spots change in appearance or size, if that happens fairly rapidly, then that's a sign that that

could be melanoma. now it's not actually all that straightforward. there certainly are what are called benign nevi growth of legal no ma cytes which produce pigment in the skin ask give rise to melanoma. they can certainly grow.

the ultimate diagnose really has to be made by a pathologist. these are some signs that can be used to distinguish melanoma, a malignant melanoma from a benign mole. but the bottom line is go see your dermatologist. that i think is the key thing.

some patients actually have large numbers of these fairly dark moles and basically what's done is patient's bodies are photographed. you'll take a photograph and then come back several months later and take another photograph to track the progress

of these moles. if any of the moles are increasingly changing in size and shape, then that's used as a diagnostic tool for a melanoma. there's also a number of melanoma risk factors, of course. certainly prior melanoma is one.

if you had a malignant melanoma removed, then you're at an increasing chance of developing if you have atypical mole, certainly you are at greater risk and you have some exposure, extreme sun exposure as well as fair skin with an inability to tan.

that's allege so associated with your ability to tan is affected by your genes as well because the genes that are involved in pigment formation some of these have mutations that affect your ability to actually melonnize. so that can be influenced by normal genes.

genes that are expressed in melanocytes and associated with the function of the melanocyte with the ability of that melanocyte to produce melanin. but there certainly is a strong association is germline mutations. now up here i say they are

estimated to play a role in about 1 to 2% of all melanomas, but of course that number could be significantly higher as we discover more and more mutations that are associated with melanoma and yardena will be talking about this some as well. that number may of course go up.

but the genes that have been strongly associated with melanoma include a gene called cdkn2a. a mutation in this gene predisposes individuals to develop melanoma at a very early age. so people that have these

mutations in this gene, can develop malignant lesions in their 20's or 30's. as opposed to what happened in general with sporadic melanomas which generally develop much later in life. patients typically that we see in our clinic are in their late

40's or 50's. and individuals that have mutations in this gene have a very very high likelihood, a lifetime risk of about 80% of developing melanoma. so certainly family history is important, and that's why genetic testing is so critical

for diseases like this. because if you know that this is a mutation that's present in affected family members then you know your likelihood of developing melanoma is much much higher. now there's other genes that are associated with melanoma, cdk4

is also is another gene which is a kinase gene, a gene that phosphorylates proteins. as well as mutations in a number of genes that are involved with pigmentation. two of these are mc1r as well as -- which is the gene that plays a critical role in

synthesizing melanin. if you have a mutation in that gene, that can lead to albinoism. so people who have no pigment in this has that gene. other genes involved with pigmentation have been associated with increased risk

of melanoma. not as high as mutations in these two genes but nevertheless there is an association. i think this slide really sums up the importance of screening and getting screening and scratching the disease at a very early stage, because melanoma

can be staged into four broad categories. stage one, where the melanoma is very thin, less than a millimeter in thickness and hasn't spread to any sites. patients with primary melanomas that are diagnosed at this stage can do very well.

this is a 15-year time scale here. so a few of those patients will go on to develop malignant melanoma but most of them will remain disease free. if the tumor is, if the lesion is thicker, up to two millimeters, but still hasn't

spread obviously to lymph nodes or distant cital there's an increased risk of developing melanoma and actually dying from the disease. and then if it spread to nearby lymph nodes but not distant sites increasing number of patients will die from this

disease and the final stage where it spreads to more distant sites is uniformly lethal. basically everybody has develop metastatic melanoma to any distant sites will die of this so what can we do about that. all right. so first i want to discuss some

of the current strategies that we have that are available for now there are a number of these. these are listed as strategies for treating patients that have stage 4 disease or unstage disease which means they have cancer, melanoma that spreads to lymph nodes but for one reason

or another it can't be removed or resected. so if patients have stage 4 disease, they can receive surgery. in some cases this is beneficial. if it's only spread to one site, it's possible that this may cure

the patient. although in most cases, if it's spread to a distant site, there are micro metastasis that may not be evident on x-ray scans to other sites. but it may in some cases benefit patients. now high dose io2 or tc growth

factor actually was approved in um of years ago by the fda. and this leads to complete responses in about 5 to 10% of the patients. merely administering a t cell growth factor can lead to a complete remission of tumor. and i'll talk more about use of

t cells in the following sections of my talk. the use of t cells for treatment. and antibody called -- which is directed against inhibitory molecule ctla4 also expressed on t cells can also be effective. although there are few complete

responders to this -- inhibits a kinase called braf. this was recently approved. this causes dramatic progression of -- melanoma but essentially no complete cures. so patients have remissions, they develop, you know, complete, a nearly complete

regression of all their melanoma but then subsequently the melanomas recur. so by inhibiting this kinase which is crucial for the growth of that melanoma, that melanoma will at least temporarily regress. but additional mutations then

can occur in the tumors and then they can, then they recur and essentially all patients who receiver this. now there's a lot of work on combining inhibition of this kinase with other molecules, other inhibitors of the signaling matt way.

and there are some encouraging results. but nevertheless, we don't have many complete cures. we don't see many complete cures in patients that just received this inhibitor. we're actually exploring the use of this inhibitor along with t

cell therapies in the surgery branch and don't have anything to report yet. but that's another possible approach that may lead to better therapy. chemotherapy really is basically ineffective. it was used for a number of

years but we know that that has essentially no effect combining chemotherapy drugs with antity feron has no benefit. basically the homes for curing metastatic melanoma certainly reside in the current and ongoing clinical trials. now there's a number of these

trials some of them involve vaccination of patients with short amino acid sequences, whole cell immunization as well as antibodies against another inhibitory ligand pd1 or pd1 ligand, these antibodies interrupt the signaling pathway that actually inhibits responses

of t cells to antigens. or adoptive cell therapy. that's what i'm going to primarily focus on because that's what we do in the surgery branch. this slide here basically gives you a summary of the response rate.

complete responses are defined as the difficult appearance of all radiologic disease for at least a period of two months. partial response is shrinkage of at least 30%. and this is the objective response rate which is the sum of the complete and partial

response rate. so you see in response to chemotherapy, you have very few complete responders. there's a few complete responders in the io2 group very few responders. there actually is a fairly high response rate as i said that to

the braf inhibitor. so nearly 80% of patients can respond to this inhibitor but those regressions are temporary. generally between six months and a year after the patient receives therapy, those tumors recur. so we need better treatments.

we need other approaches. cancer vaccines are relatively so immunization with the proteins or whole cells really lead to no complete response. but t cell transfer. the transfer of autologous t cells from, that recognize these tumors actually does have a

significantly higher complete so these responses in contrast with the responses with the braf inhibitors are very durable. and can be very durable. so i'm going to talk about the adaptive immune response to cancer. so this is the adaptive immune

system which are the t cells and b cells that can recognize northern entities as well as molecules that are expressed on tumor cells. and this is based on an observation that was made a number of years ago initially by steve rosenberg, but it's

subsequently been analyzed by a large number of groups that t lymphocytes are present within growing malignancies but only rarely trigger spontaneous regression. that's the very rare event, perhaps one in a thousand patients do we see any kind of

evidence for spontaneous regression. [no audio] and use these to genetically engineer t cells to confer the specificity for tumor antigens on t cells from that patient. and this will allow us to broaden the treatment to a wider

patient population. it can only be isolated from about 50 to 75% of melanoma patients and can rarely be isolated from patients with so on this summary slide i'll show you the approach that we've taken, which is to isolate tumor cells from a patient, culture

those cells in vitro and this is generally carried out in the presence of interleukin 2 which costs the outgrowth of lymphocytes from the patient. and then those cells after in vitro expansion are infused back into the patient along with io2. now the alternative approach is

to remove the lymphocytes from a patient and genetically engineer those cells to express a t cell receptor that will now allow that t cell to target the tumor cells. and then transfer that back into now, one of the thing that we found that seems to be very

helpful is that the preconditioned patients, we give them chemotherapy either in the presence or absence of total body irradiation. what we found is that basically provides space for these transferred cells to proliferate, to persist, to a

much greater extent in a patient who hasn't been ablated. now this treatment which consists of treatment with sigh toxin and -- which by themselves have no effect on the tumors, the tumor regression. either with or without total body or radiation.

basically that creates a lymphopenic environment where the number of lymphocytes in the peripheral blood drops from around a thousand to 1500, which is the number present in a normal individual close to zero. and then the cells that have been expanded in vitro are

infused back into the patient. io2 is generally also administered at this point to help support the lymphocytes that are infused. you can see here that actually the lymphocytes don't appear to recover immediately at least not in the peripheral blood but they

nevertheless do appear to undergo significant expansion. we can actually in some cases see lymphocytosis. we see higher lymphocyte counts than normal in these patients after they received the transferred cells. and at this point in time, a

relatively high proportion of the cells in the patient's peripheral blood actually consist of the cells that we've administered because we've ablated this patient. we haven't totally removed the the patient still has some endogenous lymphocytes that do

eventually recover. a high proportion of the cells in the peripheral blood in many patients can consist of the we think this expansion provides an advantage in that we have a lot higher number of cells to cause tumor regression. now, this slide summarizes our

recent three trials that have been doing using cell transfer therapy where we transferred from the patient in the absence of total body irradiation or 120 centigrade total body irrad information. these numbers here are the number of patients responding

and then listed below are the duration of those responses in months. so a number of patients have partial responses. in other words some of their tumors regress. they regress to a significant extents.

but then they later on recur. so if there's a plus sign here, that's indicating, that indicates that the response is still ongoing. so these patients have all eventually recurred. some of them actually have had very long lived responses before

the tumors have started to grow. so this patient here went for seven years before the tumors started to grow. but a until of patients in this trial have had now complete responses. and you know, these responses have a range generally between

five and nine years. these are very long term responses and these are complete there's no tumor detectable in these patients at this time. so these treatments can be very effective. as you can see here, 40% of the patients that were treated in

this arm of the protocol have had complete responses in all but one of them are ongoing for five to seven years. so these are very long term responses that we see in these just to give you an example, some of the characteristics of the melanomas as well as the

response, here's a large lesion on the shoulder of this patient here prior to therapy. 12 days after treatment, that tumor is gone. that tumor has now regressed probably completely here. there's no real evidence of tumor there.

we can also see complete regression of extensive visceral lesions. these are, this is a patient who had multiple metastatic lesions in the liver and you can see it's studded with lesions. this is just one slice of the liver.

if you look throughout the entire liver, there will be many more. more of these metastasis. a month after the treatment, the patient's liver appears to be free of december. and this is now more than two years after treatment, and this

patient is still disease free. now seven years after treatment. so the treatments are effective. they are long lasting. they can be effective for visceral as well as a subcutaneous disease. and even very large lesions such as this lesion, the scalp of

this patient. this patient actually received a treatment in 2005 and you can see here now the patient is disease free. it actually regressed much more quickly than that but i'm just showing you this slide here to show you the durability of the

response. the patient is still disease free. there's no cancer in this patient. so relatively small numbers of cells can treat large, very large lesions on these patients. so, what else can we do.

how else can we treat these cancers. well another approach that we've taken is based on isolating the t cell receptors from reactive cells in a patient. and then transferring those to another patient's t cells. so t cells recognize their

targets primarily through the t cell receptor, and this confers the specificity of recognition of the antigen which here is presented as this mhc molecule. it's basically like a lock and key. this receptor has a great deal of specificity for the specific

peptide which is used between 9 and 10 amino acids that's bound tight mhc molecules of that individual. and so that allows us to create a cell that is a high degree of specificity for the cancer. it will not recognize any other it will only recognize that

particular molecule. now, this approach initially was carried out by targeting -- expressed in normal melanocytes. this is involved in forming pigment. so there's a number of genes, many of these are enzymes that play a role in synthesizing

melanin. variations in these genes effect skin color and hair color. so mutations in a certain enzyme in the pathway will lead to red hair or fair skin. so they play a very important role in determining your susceptibility to cancer.

but they also, these very molecules can serve as targets for the immune system. and we initially focused on these melanocyte differentiation antigens as we knew that many recognize -- derived from these proteins. this is actually a sort of

graphic illustration of how these melanocytes can be targeted. this is a patient who received therapy with hill and this is his face shown under a woods lamp which is basically black light. uv lamp and the areas of the

skin that are bright that appear fluoresced, the area has been lost from the skin. it's extensive. it's not complete and it's not something we understand. we don't really understand why there can be such apparently random pattern to it but

nevertheless this can be one of the consequences of treatment with our t cells. in fact there's an association between the pigmentation, between viglio and therapy. we targeted some of the antigen that we knew were predominantly recognized by identifying a

tumor reactive t cell receptors, cloning them into viral expression vectors producing retro viral super deans to transduce a patient's peripheral blood cells. and then those cells are expanded and infused into the however one of the things that

we saw in this trial was toxicity. this is a patient who richard t cells that express a very high receptor and we saw the severe rashes on a patient's arm as well as on this patient's trunk. now that did resolve, eventually the skin came back and had a

completely normal appearance. however, this is obviously a is he severe reaction. patients have also have eye and ear toxicity. some patients have actually had to take steroid eye drops for years after the cells. the cells remain in the

patient's body. they remain active and can attack tissues in the body. that have pigment such as the eyes and the ears. and skin. so we change our strategy now and targeted other types of antigens.

there are cancer germline antigens which are more attractive targets because these are really limited expressions to germ cells. they're expressed in the testes and not any normal adult tissues but they are expressed in a fairly high proportion of

multiple cancer types. some of the amount judges that we've targeted are nyeo1 this is a relatively small family that includes two genes. this is a larger family of genes. as i said they're expressed in a wide variety of tumors.

here you see the percentage of tumors that express these ant judges in a variety of tumor types. so bladder, non-small cell lung cancer -- or squamous cell carcinoma. so many of these tumors express a fairly high levels of these

so one of the clinical trials that i've been involved with over the past several years is a trial targeting the antigens. this is a member of the highly restricted family of antigens that's expressed in melanoma as well as a number of other tissue as you can see here, we have an

objective response rate of over 50%. and we have three complete responders. out of the 17 patients that we've treated. actually four, this patient subsequently recurred with another tumor.

a fairly high response rate and some very good long term and these are responders that have had disease in lymph nodes as well as visceral tissues. here's an example of a large lymph node. you can see this is very large lymph node that has now

regressed after nine months and this patient is now disease free after more than four years. a patient here had a large liver metastasis that was basically gone or nearly gone at 16 months and this also is a long term responder. now we've also treated other

cancer types and so this has helped us to move outside of melanoma which has been the primary cancer that we've treated. we've treated a number of patients with synovial sarcoma. we don't have any complete responders but we have one

patient who is now has some residual disease but is doing very well, more than two years after treatment. generally patients with this disease have long as well as lesions in the bone. and these can be very large as shown in the next slide.

this is a very large bony lesion. this is present, this is a pelvis of this patient. this patient had a lot of difficulty walking because the tumor invaded her pelvis. it was painful and difficult to walk.

you can see after 18 months this tumor has shrunk down. you can see regrowth of the bone here in parts of the pelvis. so the body actually can regenerate some damaged tissue. and this patient is now walking. and doesn't have pain, doesn't have the problems that she had

before treatment. and hopefully that will continue to shrink. and we'll see complete response eventually. now we can also target cell surface antigens using antibodies ant and the waive we do that, these are

called receptors that identify an antibody that recognizes a sell surface molecule and actually engineer the molecule. so antibody that recognizes the antigen is engineered into a construct that contains signaling domains for the t cell.

and so now that cell can be redirected against a cell surface molecule, expressed on the tumor cell. this is one marker cd19 that's expressed in these malignancies. we have several long term responders now in an ongoing clinical trial.

we are targeting egf receptor which is a mutated gene product blastoma as well as these mez -- we think we can direct it against a mutated antigen because that will be specifically expressed in that patient's tumor. that will be a foreign antigen

basically. and so the way we're going about this in a project that we started with yardena samuels is to -- and then to predict which peptides or which genes we might be able to use to identify either nationally occurring cells or to actually raise cells

from patients or from mice that recognize those antigens. and then those t cells can then be used or receptors from those t cells can then be used to develop personalized treatments for patients. now, one of the things that we as well as others have seen is

that melanomas which have a much higher mutation rate than other tumors themselves can have they highly variable mutation. these are non-synonymous coating regions. they range here from almost 2000 to about 30. so there's a lot of

heterogeneity going on. these are all melanomas and that's i think going to be a big challenge to understand what are the key genes that play a role in developing tumor types and yardena will certainly tell us something about that in a few minutes.

but we've also used this approach to screen naturally occurring pill for their ability to recognize these mutated episodes. we've identified 11 different mute eight epitods recognized by patients till. we've identified them from seven

of the eight patients where we carried out this screening. we don't know if this is associated in think way with a we think these are important for the response but at this point we've analyzed too few patients. a couple were non-responder, one however did respond to at least

one mutated epttaupe. we're still screening these patients to see if we can identify more mutations because you can see some of those tumors have hundreds and even thousands of mutations. it's a challenge for us to identify all of the potential

mutated products that could be recognized by that patient's but that's basically where we are. so just to sum up, i think we can say that adoptive therapies with hill with genetically engineered cells are affected to mediating responses.

mutated antigens can be better targets for therapy than self antigen as you think they probably trigger more effective responses and development of methods to target mutations involved with maintaining the tumor phenotype or drive the mutations may lead to even more

effective therapies. and that is what yardena will be talking about next. [appuse] >> [indiscernible] >> t cells can recognize [indiscernible] peptides that are endogenously cleaved within the cell.

>> not a hundred percent. there are other genetic factors that are going to determine that. people with disease tend to stay out of the sun. they protect themselves from the damaging effects of uv i would presume that's part

ofate. it's a more complicated thing than just that. >> do you have any comments about this australian controversy of whether sun screen actually does anything? >> i don't really know. i think the best advice is

probably just to stay out of the sun. i don't think anything can be completely protected. >> thank you very much for inviting me today. we are focusing on genetics trying to search for [indiscernible] so i'm not going

to go over melanoma as a disease because paul went over that. i'll be emphasizing a more genetics. really the paradigm for the genetic development in cancer comes from colorectal cancer. you can see here where you have a colorectal cancer developing

stage wise histologically, you can see how it is linked to various molecular changes at each one of these stages. and so really this has become a paradigm for solid cancers where each stage has additional molecular changes. and so what we know about it is

we've learned at we're trying to identify much more. and so that's really our aim. we have good proof it's worth looking at these genetic changes because we have examples that have been identified and here's braf that paul was talking about that is in various cancer types

especially mel -- melanoma. now we know there's -- which is fda approved which is specific for the bras mutation. the kinase also found to be highly mute eight kong genes, human cancer. again this is a target that many are trying to inhibit.

so this proves it's worth looking for these schematic mutations in order to identify new targets to therapy. so in order to do this kind of analysis looking for mutations, we first establish a tumor bank. i'll give you here a few slides extending what i mean by tumor

banks. so here we are see this is for breast cancer but of course it's for various cancer types. in this case you have a primary breast tumor which in some cases you can turn into a cell line or you can show it as a xenograph this is expanding the cells to

make a genomic dna. you're cleaning it up from the normal cells that are surrounding like the stroma and fibroblasts that you get the population of cells which is much more homogeneous. and then you can actually extract the genomic dna and

analyze it for sequencing. this is just showing that you can take early lesions and late lesions and use those in your tumor bank as well. now a tumor bank can we made from different sources of dna. you can see here you can make, you can take a frozen tumor,

otherwise known sometimes as an oct block. you can use imbedded tissue. most of the tissues are -- imbedded especially as pathologists look at this. then you can make a cell line not every cancer cell you can turn into a cell line but with

melanoma it's very high number of tumors hat can be turned into cell lines, 50 to 70% success like i said this is a xenograph from a mouse. the advantages and challenges using each one of these sources. so of course if you're using the first tumor it's going to be

reliable data because it's right out of the patient and it hadn't been processed for example for the cell line. of course you're going to have limited amount of dna from the fresh tumor. it's going to be heterogeneous -- macro

dissection micro dissection and even later capture and the pathologist involved. again pair fib imbedded tissue is great for your data but you get limited amount of dna, labor intensive. the dna quality sometimes is not very good.

it's hard to apply to various sequencing methods. the cell lines you can get lots of dna, the extraction is very straightforward and you can use it in functional studies which is a very powerful follow up when you identify the somatic but you do need to go back to

the first tumor to validate that it would be identified in the cell line is actually here in the fresh tumor as well. make sure everything else following up on any artifacts. xenograph kind of advantages to the cell line but it's very expensive.

and you may have mouse dna contamination so you have to be careful about that. okay. so of course if you're looking for somatic mutations you're comparing the tumor to the normal cells in the body. usually you do that using blood

so that's not always available. so you may take -- tissue but the problem with -- tissue is that with sensitivity of current sequencing technology, you may be picking up contaminating you have to try to gather as much clinical information as possible from the patient date

of birth, date of death, metastasis tumors and also theorems that that patient underwent. of course if that patient underwent radiation chemotherapy those have induced mutations in their own right. so you have the -- that's just

standard. so when i establish my lab six years ago and i decide to start working on this we were very fortunate to be able to work with the rosenberg group and paul robbins to set up an excellent tumor bank. after that we expanded forward

additional groups in order to have samples derived from different clinicals. this is a summary of the tumor bank where we have metastatic tumor dna. we have late stage metastasis. we have the normal dna that's blood.

we have the original oct blocks for all of these examples so we can go back to them if we ever want to validate any of our data. for the rosenberg group we have matched cell lines so we can derive from those rna, match protein -- and pernormal

functional studies. for all of these we have clinical information. emphasizing again it's important when you bring these kind of studies to look at additional groups because it's good to validate whatever data you get in one clinical center.

in another clinical center, there may be some biases because of patient recruitments for example. so once we establish our teen we perform several quality controls to make sure the data is reliable. the tumor and the normal match,

mean the tumor and the blood came from the exact same patients using -- is analysis because there are errors sometimes and we don't want to follow up on a polymorphism thinking this is a somatic mutation. we make sure the cells we're

looking at are mainly melanoma derived. we look at antigens of the specific melanoma and make sure that 75% of the cells are expressing them. this allows us to be able to look at the loss of hereto zygosity.

it's similar to what's known in the literature. once we did all this we're happy with the tumor bank to go ahead and start sequencing novel so just to recapitulate what is somatic mutation is. you have the patient, the tumor in the normal tissue from which

you could make a cell line, extract is dna, sequence the gene of interest and you do the same for your normal tissue and you compare these sequences. only these somatic mutations are interesting to us and to follow up on. to begin with you're going to

look at exxonic sequences so the one that alters the proteins and amino acids now we look at those mutations as well. so we started off by sequencing single genes -- moved on into sequencing families of genes. and thousand we're doing whole exome, whole genome analysis

which is much less biased. so this is i showed you at the beginning the progression of colorectal cancer. in melanoma you have such a progression where you have melanocytes within the ecto dermis. this was developed into -- and

may develop into these additional stages where you have radial growth phase -- at that point it may invade the dermis and then the metastatic also here i'm showing you some a molecular changes that are well-known. of course by now we know many

more of these changes. but our aim is to identify -- like i said we focus on this stage, the metastatic stage. so like i said we're looking for these mutations. we don't stop at the genetics. we really believe it's important to understand what these

mutations do, the function of the protein. and we do try to be relevant for the clinic. so if we have a lot of choices in terms of somatic mutations they will be the ones applied back to the clinic. i will give you two examples of

the studies that we performed. one is looking at the kinases, the tyrosine kinases which -- here you're looking at the human -- we decide to look at this family because we had different cancer types and they are known to be mainly oncogenes and there are pharmacological in

life and like i said we wanted to be clinically relevant. so this is just a screen that we did. we always do our screen in a two-stage manner. we start with the discovery phase and then we move on to validation phase.

discovery phase is done with a smaller number of examples. in this waste we used 29. we looked at 386 genes. we had a mutation and one would expect it to be there. any gene that had one somatic mutation or more moved on to the next phase, the validation

stage. there are 19 such genes and we sequence them in additional samples, 18 total. what we found was one gene that was highly mutated and that was ob4. this is just a scheme of the protein and the various

mutations that were identified. so this is an example of the sequencing method that we did in 2009. and so we moved on to understand what these mutations do. they were so highly mute eight. so what you learn before we know etcetera part of the erbb family

which kisses of the egf receptor which are both known to be good targets in the clinic. ob3 and ob4. these are -- on the cell membrane and upon ligand binding they will correlate forming phosphorylation sites, docking sites for various signaling

molecules. so in my lab cloned 7 of these mutations based on analyses that we could go into later. -- -- to the kinase activity -- we compared it to wild type. you can see here it doesn't increase signal for over seven years since they are more active

compared to the wild type. even though they are similarly expressed. so we did multiple studies to look at these mutations and touching upon very few of these. so this one for example, if you are sure we could use the health no ma cell lines in which the

mutations were found and modulate those cell lines. in this case we used shrna to knock down the endogenous erbb4 where you see it's lost. we make these stable cells examine we checked how does the knock down affect the growth of the cells.

we did this in cells of either wild type of erbb4 or mutant. when we did this for wild type we didn't see any effects on the growth. we see similar to the control. when we did this in service for mutants for erbb4 they did not grow so cell.

so they suggested to us this could be an interesting target because if you're knocking it down specifically in mutated cells they stopped growing. and this is called oncogene addiction. so the next step was to see if we could use a small molecule

inhibitor that's currently being used in the clinic. so we use -- which is fda approved and it's being used for breast cancer measures and we exposed multiple cell lines. knees -- these are only a fewof them of the wild type and you can see here the various

concentrations. when we plotted the survival we could see the ones that had -- more sensitive to the exposure of lapatinib compared to the wild type cells. based on this study, we had hypothesized that 20% of melanoma measures are going to

harsh -- harbor the mutations-- performed here at the nci in the surgery branch in collaboration with -- so patients were being screened for erbb4 mutations and anyone who consent and had the mutation got their lapatinib. what we do have from the studies if these mutations were the ones

i showed you to begin with. these are the ones that were identified in the clinical trial. and what we see here is the frequency similar to what you identified to begin with. and these are additional whole exome genome samples we have

sequenced since our first paper and you see we're still finding and this particular study was by nick hayward and you see additional mutations in erbb4 and finally this is another paper publishing erbb4 we can see not only the frequency -- we're starting to

see recurrence mutations in particular locations. some of these were functionally evaluated. the ones with the stars. so we know they're activating. what this means is we still don't know but it does seem to be hiding and does seem to be

playing a role in our normal progression. so then we moved on with technology and we started using second generation sequencing approaches. and so i'll give you one example of those studies. and this is a whole exome study

where we captured and we were the first to publish melanoma whole exomes. this was in 2011. this was very expensive so we did only 14 are samples. we had 14 tumors and 14 matched normals that we captured -- and we sequenced using the lumina

flat form and then use eland to look at the data. any of the validatations was done using sanger sequencing. we use different methods to validate the date. this is just an example of how the capture works. you can see here the -- are

capturing the braf here and you can see the sequencing for the various exomes right here. you see there's a variation. we need to go through it in order to look at the alterations in each one of these. so you'll see we go through over 100x coverage.

so this is the performance of the whole exome you get 12 gigabytes of sequencing. for example like i said there were mean department was 180x greater so we have high coverage, very low negative rates and our sensitivity is 81%.

these are the steps that we took for this study. like i said, we looked at 14 untreated samples. we got a number of potential somatic mutations. this was prior to filtering. so we awe seam -- assemble the data and filter the somatic

and for this particular study was developed in new algorithm called mpg to cover the ratio. the mgp was already published prior to this study which was the genotype. we took into account the coverage of each one of these examples and so we moved 18%.

and this is the way the mutation's divided. we're always looking for non-image mutations as well, not only the one that affect the coating but also synonymous because we want to look at the ratio between the non-synonymous and synonymous mutations.

if it's not significantly above what's expected by chance, these are probably neutral effects. they are not really effect in the tumor. you see the ratio is 2-1 which is not significantly higher to what's expected by chance. so these alterations are

passenger and they don't have a role to play in melanoma. that's really the challenge. as paul was talking some of these melanomas have an extremely large number of it's really hard to find out which ones really play a role in the disease and which are just

neutral. just because they were exposed to uv and they just have these alterations. that's what we're working on now to apply various methods to know what are the drivers in the and so the statistics we looked for you hot spots meaning the

same exact mutation again and again and again in the same position such as braf. we also looked for the frequencies that are highly mute tated being selected for. i told you about the synonymous and non-synonymous. so when we looked for hot spot

mutations we found nine such that was surprising because we knew braf. we can't expect any additional recurrent mutations. so again we validate it. we look at additional sample set to see whether this produces and scales up.

and it did for one, at least for one of these positions which is called trap. this is a gene called trap and you see the same exact alteration occurred in six different patients. and this was also seen in other studies as we see here.

and so we're now working on this particular protein. we know it does affect the cell survival in the cell in which it's mutated. genomic data you can find these novel recurrent mutations in genes that we had no idea that had the role to play in the

and this is the other side of the coin looking for highly mutated genes. and so we found 16 highly mutated genes and again with a validated them in more samples. and this is the condensed slide with a lot of information but the take home message is when we

do this statistical test, we do find braf to be at bthe top of the list which is to be expected with the lowest p value, the most highly mutated. but you find other genes in this list as well. then when you validate in more samples, the frequency remains

similar. so it's important suggesting that we're doing something right so this was the second most highly mutated gene and we found it fascinating because it was never linked to melanoma before. and so if you look at -- this is these are the mutations that we

identified. and the box ones are nonsense these mutations truncate the protein. so this is present in about 15% of all the mutations which is definition of a tumor suppressor gene. something is going on here.

this is probably the tumor suppressor gene. this is a -- some of these mutations are already confirmed in other studies. an additional one since as well. so in addition to looking at genes, single genes since we are looking at the whole genome we

can start looking at pathways as well. there are many different ways to do that. the first person who did this was -- and his group where they looked at all the exomes in pancreatic cancer and found 12 whole pathways.

even though two different patients had completely different mutated genes you could start noticing thesegenes actual fall in the pathway. you can't have theorem for each one of these mutations but you can try to develop different therapies for different

pathways. this is the optimistic side of looking at all of these so when we had our exomes we did a similar analysis and we found one pathway to be significantly mutated and the value is here. that's the glutamate signaling pathway.

i told you about -- which is a glutamate receptor and binds glutamate and you can see these various proteins that are highlighted. these are the ones that we found mutated. before i told you about, there's actually a paper in medicine

showing that -- ob4 interacts. and ob4 phosphorylates and we have recapitulated the melanoma and then we have other gene here and we don't have the time to go into them. but this is the pathway that is fascinating to us. so we're going to continue

sequencing more melanoma in that's because if you look at the mutation frequency in melanoma compared to other solid cancers you see it's extremely high. i would like to emphasize. this is an average of many melanomas.

so of course you see a variation between them. it's still extremely high. it's about 16 to 20 mutations. so we're saying we need to analyze many more of these to find out which are the drivers. and so that's what we've been doing.

we've also been merging a lot of data from other sources because this has been a very fruitful time in the melanoma field. you see there are many studies published in the last two years, genomes, exomes and we are merging this together with our data and other unpublished data

from other groups. at this point, we have very large number of melanoma cases that we have merged. and we can see that there are they keep extending and you see that how this control comes up from the braf, nraf -- these are known hot spots but we have more

to look at. so that was really our aim. so we're looking at these and we're looking at a total of 600 zap pulls. we have some having targets already that we're validating at this point. in addition to all of these

groups with the sequencing exomes and genomes the tumor cancer genome at -- atlas that was launched and is ongoing. we actually now have over 200 melanoma exomes. so we're going to have a very large amount of data to look at very very soon.

this is just to show what tcga does and in fact they are indeed looking at dna sequence as well as many other aspects of these different cancers. our challenge now is still looking at the drivers and the passengers. we want to keep analyzing and

how to interpret the data. we do the functional analysis but how do we do it in a high super fashion and how do we apply this to the clinic. so now we're mainly looking at disease in the clinical and -- aspect and this is really going back to how we started the whole

afternoon. we now hope to be applying more mutation analysis in order to find the various melanoma subclasses of this disease in order to further develop a more targeted therapy that may be more specific to the ones that we see here in the slide.

and of course the therapy that we're talking about is extremely exciting and hopefully we'll leave this to better results in the future. there are many people that have contributed to the work. thank you. [applause]

>> thank you very much. >> speak loudly. are there any questions? >> i wonder if there is something unusual that you have from your exome sequencing from -- melanomas as compared to the more common pigmented. >> we haven't categorized them

based on the melanoma content. so i can't answer that question, i'm sorry. >> in the absence of all the data, it's hard to, you can't draw conclusions. you can certainly get impressions that influence the way you think about it.

so this question may be a little bit strange. when you look at the histologic pattern of changing let's say of melanoma from an initial mole eventually to an invasive thing. classally that's what is described at specific stages. is it your kind of feeling that

in any given patient there are likely to be common drivers for those stages or is it more complex in that there are many drivers interacting with many different kinds of machinery? >> that's a great question. so because i don't think we have enough data to completely answer

that question. this is going to merge hopefully at least 600 exomes and genomes that include clinical but again these are all metastatic. we don't have the primary or earlier stages. i hope you will be able to

divide those into subclasses based on mutations, right. and link that to at least some of the clinical and pathological information that we v we're hoping in the future technology -- you can start doing sequencing with less dna and analyze it even though it's

heterogeneous, we'll be able to pick the most interesting targets. and sequence those in earlier lesions, okay. which is difficult to come by. but you know, we have been able to get some of those. and then be able to see which

ones of all the compiled list of mutations occur earlier and which ones occur later. do they have similar -- if you look at braf for example, you find it in -- already and it's sequencing -- and that's an early stage and an early denominator.

that's an example where you have a driver that does occur in the melanoma population. but whether there will be more like those is still a question. do you have anything to add to that? >> how do you control, i'm sorry.

how do you control for the fact that most patients, if there are found to have a melanoma are treated in one way or another. on the outside they may get chemotherapy, i'm not sure. but how do you control for the fact that when you have the tissue and you do all the

genomic analysis, how can you discriminate between what is the response of the tumor to the therapy that was given. both beneficial and maybe resistance. and the enact that the -- the fact that the patients are all different in that regard.

how do you control for something like that. >> at the moment not that many targets, not many inhibitors . so there's only like one correlation i think. if you take braf as an example and you give them -- you're just following up on that particular

i think am i understanding your question correctly. like paul was saying, they are doing now some very, they are merging different drugs together -- and that is increasing survival by three but that is really our dream to be able to have off the shelf

inhibitors, have the sequencing information and based on that, be able to mix and match and see whether the tumor is more responsive. >> you started originally talking about radiation, sun raped information as one of the drivers.

you talked about some of the mutations perhaps not being the driver within melanoma. do you get a sense that all the drivers that lead to melanoma origin of some systemic toxins that have caused mutations or is there some combinations. >> there is a signature which is

a uv signature and even more on -- signature. so if you start looking at these mutations and just look at those signatures, you'll filed that signifily moves alterations. not all of those are drivers. obviously you have all of those by standards just -- on the

other hand braf which is the main driver in melanoma does not have a -- mutation. so based on that people start to say okay if you look at the mutation it's probably a passenger it's not going to be a driver because of braf. recently there have been a cell

paper by -- showing that actually droifers do also have a difference between signatures so you cannot exclude them as playing a role in the disease. now in addition to uv, you have other toxins you were saying could be inducing other kinds of toxicities and i think david

fisher recently had a major paper dealing with that. does that answer your question? >> it does. somebody grew up in cave and never got exposed to someone doesn't have a uv driver at all, is there still going to be melanomas and if so what are the

drivers behind them. >> that's a question good. that's -- even in the absence of uv. we were thinking okay no uv than what was going to be protected if you use sun screen. unfortunately that is not necessarily the case.

>> some of the melanoma samples we looked at don't have p redominant cdt. on that slide i showed with the variable frequency of these mutations, a couple of the tumors that have fairly low numbers of mutations don't have a predominance of the mutation.

that's where i think the clinical data is also important. that tumor i believe arose on foot or heel. so it may even be the location of the tumor plays the role. sun exposed part of the body then it's going to be more exposed to ultraviolet

that introduces more complexity into the situation. so that driver driving the tumor from a patient's foot primary are the same drivers driving the tumor that developed in a patient's scalp. but that's why studying these large panels of patients and

correlating that with clinical characteristics i think will be fascinating because it's only been are we going to be able to really make some sense of it. but that's why that study's so important. >> is there a initial difference in the epidemiological

occurrence of melanoma as having mel nottic stem protect in any way. >> frequency is much lower. go ahead. >> the frequency is much lower in african americans, and of course there are, it's affected by the color of the skin and the

color of the skin of course is affected by mutation also in the pigma genes that give rise to the melanoma so that as populations presumably he looks occurred by migration of populations from africa or parts of the earth that were warm, and as those populations migrated

north, there's more a selective pressure to maintain darker pigment skin because there was less uv radiation. so it had less of a disadvantage. so genes could mutate. they created saw sentability presumably to cancer if patients

were exposed to ultraviolet so there was perhaps some selective pressure but nevertheless those mutations still presumably pre dismost posed to mutations. to sort it out you need large studies. so these large studies like

yardena was talking about will help to sort it out. we'll have the characteristics, you'll have the clinical history and then you can sort out the type of mutations that will occur much patients that have fair skin, red hair. you can perhaps start to parse

that out. you need a big study to do that. you can't just study the small number of patients who won't have enough data. >> are there any other questions? i want to thank you on behalf of all of us.

we're really very excited [indiscernible] the way things