Cancer De Rinon

w. marston linehan:thank you very much. on behalf of paul spellman and myself, and the rest of the members ofthe papillary renal cell carcinoma working group -- analysis working group, we want toreally thank the nci and nhgri for putting this together. on a personal note, as a urologicsurgeon who’s worked on the genetic basis of kidney cancer for 31 years now, i wouldsay this project, and the previous project we worked on, the tcga clear cell, and thetcga chromophobe kidney cancer really are sort of a dream a lifetime. so what we’regoing to talk about is the comprehensive molecular characterization of papillary kidney cancer.just for background, kidney cancer is not kidney cancer. it is a number of differenttypes of cancer that happened to occur in

this organ; different histologies, differentclinical courses. we talk a little bit about the genetic basis of them obviously. what we’re talking about today is papillarykidney cancer which makes up 15 percent of kidney cancer. and what we knew when we startedthis project, what has been a given for a long time, about 25 years now, is that papillarykidney cancer roughly is divided into two types -- type 1 and type 2 papillary kidneycancer. type 1, from a clinical point of view, is a relatively homogeneous disease, appearsclinically to be more indolent then type 2 kidney cancer. at this point, in may of 2015,by the way, we have no effective forms of therapy for any type of papillary kidney cancer.the nccn guidelines -- if a patient comes

to someone like me with advanced papillarykidney cancer, is to put him on an experimental trial. we have seven drugs approved for kidneycancer primarily targeting to the vhl pathway. in clear cell kidney cancer, we have no drugsfor this. now, what we know, or the field has knownmost about the genetic basis of papillary kidney cancer really comes from a couple hereditarytypes of kidney cancer, the first one being this hereditary papillary renal carcinoma,hprc, which is the inherited form of type 1 papillary kidney cancer. the gene for thisof course is met. in all of those families we find met mutations in the -- activatingmet mutations in the tyrosine kinase domain, and a few met mutations had been found insporadic papillary kidney cancer prior to

the starting of this study. now, most of us who manage patients, you lookat type 2 papillary kidney cancer, you say, "my word, that is a very heterogeneous disease."at the nci, we even refer to them sometimes as non-type 1 to tell ourselves we reallydon’t understand these cancers very well. you could think of -- you could say clinicallyand phenotypically that these must be about three or four different types of cancer andwe’ll talk about that. again what’s been known a little bit has been found from studyinghereditary types of this, of type 2 papillary kidney cancer, a disorder called hereditaryleiomyomatosis renal cell cancer, very aggressive form of type 2 papillary kidney cancer, causedby a mutation of the krebs cycle enzyme fumarate

hydratase. so, in this analysis, of course, 161 samples,looked at copy numbers, somatic mutation analysis, methylation, mrna, microrna, and protein expressionanalysis. we will not talk about whole genome sequencing in this talk. so, initially, then,we had a lot of discussion about this -- i'm going to say a little controversy about it-- whether we should actually have our pathologists look at these and characterize them. we decidedas a group that it was the thing to do because that's what we do clinically, and so we thinkabout these, and we think they're very different. so, we put together a group of expert pathologists,really are expert pathologists? i would say the five best kidney cancer pathologists,i would say in the world, but we'll say in

the u.s. and they looked -- did an incrediblejob, looked at these, had a single slide to look at -- all were frozen, so it's not perfect-- but they did a great job -- characterized 75 of them as type 1 papillary kidney cancer,60 as type 2 papillary kidney cancer, and 26 they really couldn't call. they calledthem unclassified. and i think it was their sense and our sense that most likely mostof the unclassifieds were actually type 2. now, initial analysis, looking at copy numbers.so we looked at copy numbers. there were basically three different clusters. cluster one wasrelatively stable genomically, a few numbers of gains and losses. if you looked at clustertwo, very different. you most notably, big time, increase -- copy chromosome 7. alsosome 16 and 17. but the big thing here, in

cluster two, is increase copy number 7. forcluster three, a lot of deletions, a very genomically -- you might say -- unstable typeof kidney cancer. and most notably, we would say, would be deletions of chromosome 9, andwe'll come back to that. now, if you look at the clusters for survival,cluster one and cluster two, cluster for survival, pretty close. cluster one was more type 2papillaries [spelled phonetically]. cluster two was almost exclusively type 2, and clusterthree had the poorest survival, that was mostly type 2 papillary and high-stage, stage threeand four. now, when you looked at mutations, and this was done by looking at mutsig 2.0,and then also we evaluated genes that were identified in pancan21, and a number of notablegenes were identified -- met, of course. fat1,

setd2, which we'd seen, of course, in clearcell, nf2, kdm6a, bap1, pbrm1, smarcb1, nrf2 [spelled phonetically], stag2, and p53. thechromosome 3, chromatin modeling gene mutations, that we had seen in clear cell kidney cancer-- that being setd2, bap1, and pbrm1, chromosome 3, chromatic remodelers -- were mutated mostlyin type 2 papillary kidney cancer. we then looked at a number of pathway analysis,looked at mutations and pathways. found a number of pathways that were notable, we thought,one of course being met. other pathways such as hippo pathway and a number of chromatinmodifier pathways. the swi/snf complex was mutated in genes in that pathway in about20, 27 percent of type 1 and type 2, chromatin modifiers in 35 and 38 percent, both in 1and 2, and hippo pathway mutations were primarily

in type 2 papillary kidney cancer. now themet mutations that were found, 14 met mutations were somatic, three were germline. most ofthem were in the tyrosine kinases domain. had a couple outside the tyrosine kinase domain,one [unintelligible] domain and one in the sema domain, and they were all except one,type 1 papillary -- well, either type 1 papillary, or unclassified. sorry. also found a specific met splice variantin eight samples resulting in loss of the first two exons and the gain of a novel exonof met. not unexpectedly, but when you look at met mrna, significantly higher in type1 versus type 2. same thing for phospho-met. so we find amplification of 7, mutation, splicing,and increase copy number of chromosome 7,

primarily in type 1 papillary kidney cancer;we’ll come back to that. also [unintelligible] analysis revealed a specific region deletedon chromosome 9 containing the cdkn2a p16 gene. cdkn2a hypermethylation was identifiedin 10 tumors; each of these both in mutation and methylation correlated with low expressionof p16. and when you looked at the 21 altered cases of either silencing, loss or mutation,they were almost all either type 2 papillary, 71 percent, or unclassified. when we looked at survival of all the tumors-- of all the type 2s, the ones with cdkn2a mutation had a significantly decreased survivalversus those that were wild type; excuse me, this was all of the tumors, all of the papillarytumors. we also found tfe3 and tfeb fusions

in 12 percent of type 2 papillary renal carcinoma,which was surprising. this is a surprisingly high number of tfe3 and tfeb fusions. traditionallywe would think of it as being more like one percent, including patients in their seventhand eighth decade, which is also surprising. usually we think -- this was originally tfe3we originally described in kidney cancer in 1996, and we traditionally think of as tumorin children and young adults. but it turns out that’s not the case. the tfe3 fusionsalso included four known fusion partners as well as two novel fusion partners, and bothof the tfeb fusion partners that we identified were also novel. now, when we did methylation analysis, wesaw a number of things. we saw three different

clusters. one of them, which demonstrateda cpgi with methylator phenotype, a cimp phenotype, in eight of the nine cimp phenotype papillaryrenal carcinomas, were type 2 papillary renal carcinoma, and the cimp phenotypes stronglyassociated with somatic or germline fumarate hydratase mutation, or decreased levels offh. that’s the krebs cycle enzyme. again, when we looked at the cimp papillary renalcell carcinoma phenotype, we found that it characterized early onset papillary renalcell carcinoma and low survival. so some very aggressive early onset phenotype. now, when we looked at a number of metabolicgenes, what we found was we think were pretty remarkable. when we looked at glycolysis,fatty acid synthesis, anpk and tca cycle,

what we found was in the cimp phenotype -- sohere we have the cimp phenotype, here we have type 2, and here we have type 1 papillarykidney cancer -- in the cimp phenotype this was very, very characteristic, increased glycolysis,increased pentos phosphate shunt, decreased tca cycle, okay. this is like your classicwarburg metabolic shift, increased glycolysis, decreased tca cycle, all right. but, if you look then at the type 2 papillarykidney cancers, what you see is a shift toward increased glycolysis and pentos phosphateshunt, but decreased, but normal oxphos activity, looking at gene expression patterns. so, thiswould be consistent with what we see clinically, i.e. the patients with papillary kidney cancer,very high pet scans. so what this would suggest,

then, would be that those tumors are doing-- have shifted to -- or made a significant shift to aerobic glycolysis. now when youthink about that, then you could also imagine that those tumors also would be characterizedby increased oxidative stress, increased reactive oxygen. so we’re going to come back to thatin just in a minute. so think about these type 2 papillary kidney cancers, the posterchild being the cimp tumors, but the other papillaries as being very metabolically active. so how do you put this together? this wasthe model, very similar model, that we saw in clear cell kidney cancer, in the kirk paper,in the clear cell psga paper [spelled phonetically], in which we saw, with high-grade, high-stage,low survival aggressive clear cell kidney

cancers, we saw a metabolic shift toward aerobicglycolysis and decrease oxphos. we’re seeing the same thing here, exact same thing here,in the cimp phenotype. increased glycolysis, increased pentos phosphate shunt, and decreasedtca cycle activity. now, when we did -- looked at cluster analysis,the coke [spelled phonetically] analysis, we saw four different clusters basically. this waschad creighton and katy hundley [spelled phonetically], did a beautiful job putting all this together.and the first cluster, c1, you can see here, is primarily type 1 papillary renal carcinoma.and the cluster 2a, 2b and 2c, are type 2 papillary. and the cluster 2c is -- went completelywith the cimp phenotype. cluster 2b, which i’ll show you is the very aggressive typeof type 2 papillary, was also characterized

by a significant increase in the number ofset v2 mutations. so then if you look at survival, if you justlook at type 1 versus type 2, type 2 papillary kidney cancer has decreased survival overtype 1, which is what our clinical experience would fit with. then when you look at thecluster analysis and look at survivals, the c1, which was type 1 papillary predominantly,and c2a, which was one of the clusters in the type 2 papillary kidney cancer cluster,they did relatively well to each other relative to survival. the cluster 2b, t2b, is significantlyworse survival than c1 and c2 and the cimp, of course, is the -- the 2c is the worst survival. now finally, we looked at a number of pathways,including this one, which were our most prominent

pathways when you did pathway analysis. andwe talked a minute ago about the type 2 papillary kidney cancers being characterized, potentially,by increased reactive oxygen, increased metabolic stress, and a more metabolic shift aerobicglycolysis. now a very pretty study by van tay [spelled phonetically] had shown, lookingat array patterns in type 2 papillary kidney cancers, that this fit with this pathway,which is the nrf2 pathway, and nrf2 is the -- turns on [unintelligible] 1 and the antioxidantresponse pathway, which is essentially a defense against reactive oxidative stress. this isalso something that many of us are working very, very hard to target, and we are veryhopeful that this will be a targetable pathway. so because we’re so desperate to find therapyfor this, we’re hoping this may lead us

down the right path. but when you look at the nrf2 pathway, andthe surrogate here is nqo1, which is kind of a perfect readout for a nrf2, as you seewith the c1, which is predominantly papillary renal carcinoma, and then the 2a, 2b and cimp,you see that the nrf pathway is increasing in those versus the type 1 papillaries. andyou can see that also correlates with mutations in this pathway which is nrf2, cul3, and keap1.and when you look at survival we see significant decrease in survival in the tumors that haveincreased expression of nqo1. so, what we show in this study is that type1 and type 2 papillary renal carcinoma are genetic -- genomically distinctly differenttumors with different clinical outcomes. the

type 1 papillary renal carcinoma is associatedwith met mutations, met splice variants, and gain of chromosome 7; that type 2 papillaryrenal carcinoma is made up of at least three distinct subtypes with differing survival;that cdkn2a alterations are associated with type 2 papillary renal carcinoma, and withpoor survival; that tfe3 and tfeb gene fusions are found in 12 percent of type 2 papillaryrenal cell carcinoma, and are found also in older patients versus just in younger patients;that the cimp type 2 prcc phenotype tumors are early onset, poor survival tumors characterizedby metabolic shift to aerobic glycolysis and decrease oxidative phosphorylation; and finally,that the nrf2 pathway is up regulated in type 2 prcc, and is associated with high-grade,low survival papillary renal carcinoma. thank

you very much. [applause] katherine hoadley:we have time for one or two questions, if you want to come to the microphone. male speaker:thank you for your presentation. could you give more details on what you saw about the[unintelligible] subset, with respect to the genomic alterations you saw [inaudible] talkingabout the fact that it was an unknown substance. male speaker:do you understand that? katherine hoadley:can you --

male speaker:you had a non-type 1, non-type 2, a third other category, where do they fit in? that’sthe question. katherine hoadley:if they’re not type 1, they’re type 2. male speaker:you're unclassified. katherine hoadley:you're unclassified, where do they fall? w. marston linehan:oh i’m sorry, sorry, sorry, sorry. the unclassifieds, i’m sorry. the sorry -- great question.the unclassified really -- okay, so the unclassified, that was by pathology. and that was by looking,all we could do was take the frozen, make an [unintelligible], dcj did a great job onthis getting us those slides, but really hard

for pathologists to make a call on that. anyway,so the 26 un-classifieds we had, most of those fit with type 2. okay, so our presumptionis those were really type 2 papillaries. male speaker:yeah but i was wondering [inaudible] one of the subsets of type 2 or not. w. marston linehan:you know that’s a very good question. yeah, i don’t know the answer to that. it is avery good question. it was where would they fit in with the tfe3 or the others. that’sa good question, and i don’t have that in my head right now. i will have to look thatup and i’ll get back with you on that. that’s a good question. we didn’t analyze thatspecifically, although we should have.

male speaker:so i wanted to ask about the non-met mutated type 1 tumors. any ideas as to what the driversare in those -- w. marston linehan:boy, that -- male speaker:-- that sort of make up the met. w. marston linehan:that is a very good question. and that is something we are desperate to know. and, ihope you recognize that we said chromosome 7 increase copy number. we didn’t say thatmet was the target there, because there is a lot of stuff. it is incredibly characteristicof type 1 papillary. but what the driver genes are, we don’t know, that’s a very goodquestion. so we would assume -- i mean, we've

got the whole of chromosome 7 to think about,including egf, hgf, all sorts of things. we have increased copy number. we have increasedphospho-met, we have increased met, but is that the target? that remains to be determined.that’s a very good question. we also have 16, 17, but we don’t know. katherine hoadley:okay, thank you so much. w. marston linehan:sure, thank you. katherine hoadley:our next talk is dr. angela brooks, who will be telling us about the high through-put somaticvariant impact phenotyping using gene expression signatures.

[end of transcription]