Chances Of Surviving Prostate Cancer

>> good afternoon. welcome to all of you here in masur and those watching on the video. one of my pleasures as nih director is the opportunity to come on wednesday afternoons and be able to introduce the speaker.

there's no greater pleasure than when the person who's speaking is somebody you know well and in this case, somebody that i had a chance to be a partial mentor for back a few years ago, like 15. and it's always a joy to see how somebody that you've played some

small role in helping to train makes a wonderful success of their own career path, then you get to bring them back and embarrass them a little bit with an introduction. so that's what i'm here to do this afternoon for dr. john carpten.

john got his ph.d. at ohio state, and he came to be fairly flesh out of the blocks intramural program in the human genome institute back in 1994. when jeff trent had just been recruited as the scientific director, somebody that i brought with me from the

university of michigan, to establish the intramural program in genomics, it was probably taking a little bit of a chance to show up at that point and expect that things would actually go well, but they went extremely well, and no small part in the cancer genomics

arena because of john. john had great energy and does today, perhaps even more so, and had a hunger for trying to get answers about hereditary and somatic contributions to cancer that has driven his career forward over that time now over some 17 years to where he is

today as the deputy director of basic research and senior investigator at the translational genomics research institute in phoenix, otherwise known at tgen. over the course of that time, he has accomplished a number of very significant observations

and publications in the area of cancer genetics, beginning with when he was still here as a post doc coleading an effort that led to the first identification of prostate cancer susceptibility using linkage analysis on families. i had the great pleasure of

working with john in an effort to focus this effort especially on african-american families, which it otherwise not really had much attention paid to them, through that came about the african-american hereditary prostate cancer consortium which john had a major role in helping

to get started and which continues to this day as a very interesting and important enterprise to try to understand why it is that prostate cancer happens at a higher frequency, intensity, more severe with individuals with african-american background.

in his work at tgen, he has continued to make strides forward and understanding factors involved in cancer, including work that you might have seen in terms of identification of a hoc gene that caries substantial increased risk of prostate

cancer if a certain point mutation is present, as well as very interesting findings about akt and its role in lung cancer. so it's always a joy for me when i pick up publications or see what's going on in the world to see that john continues this kind of wonderful contribution.

i think he's going to give you a really interesting presentation about some of the things that are now going on at tgen with comprehensive molecular profiling, trying to connect what's possible now in the research arena with what's going on in the clinic without having

the barriers that sometimes get in the way prevent that from becoming possible in realtime for the benefit of patients. john is both a wonderful scientist, a wonderful human being. it's really a joy to see him back here today to give this

presentation. so please join me in giving a warm welcome to my friend, john [applause] >> well, thanks, francis, and to say that that was a humbling introduction is an understatement, and i've always appreciated your leadership and

your guidance and your mentoring through the years, along with those other individuals who have played a significant role in my career development. and so with that said, you know, i do have quite a few slides and i want to make sure i can get through as many of these as

possible, and again, thanks to those who were able to make it here and to those who are viewing remotely. today i'd like to talk to you about some of the work going on at tgen, a pipeline or platform that we've developed to help, as francis said, connect what we

can do as it relates to molecular profiling in cancer and comprehensive profiling, and to link that intimately with what's going on in the clinic, particularly around advanced cancer, to really try to form a true nexus between the bench and the bedside, and so i want to

bring to bear some of the technologies that were utilized and in the approaches that we've taken or developed and walk through a few vignettes on some of the -- you know, whether people want to call them anecdotal observations and results or some of the

significant observations that we've seen in some of the patients, and then finish up with some of the areas that we think are critically important in trying to build out these pipelines, these platforms, so that they can be more widely applicable.

so the basic premise is that many cancer therapeutics are currently aimed at specific targets that are in many instances associated with genomic alterations, and this has been somewhat of a paradigm shift in pharmaceutical development where in many cases,

drugs were identified more serendipitously where a drug has some effect on a cell line during a screen, and then that agent would be further developed and moved through the pipeline. currently a lot of drugs are being -- and compounds are being developed specifically to target

certain molecular perturbations that have occurred within a cell, one of the poster children, of course, being cml and the bsr effusion and response to brian dreuker's work, also in breast cancer, in response to drugs. in red i've highlighted a couple

of areas in which my laboratory has had the honor of working in in collaborations with the scientists it t. may it at themayo clinic in scottsdale, showing these alterations are in response to the proteasome inhibitor. in studying basal cell carcinoma, identifying

activation of hedge hock pathway in many cases through mutations in the patch gene and the use of of -- inhibitor. other interesting observations, of course, are seen around triple negative breast cancer or basal breast cancers, particularly those harboring

these germline mutations in brc1, brca 2, in response it a class of drugs called parp inhibitors, many of which are currently being developed in the clinic, and a new class called part trappers. there are a number of examples showing some first in clays

drugs and phase one studies that have showed outstanding efficacy in the clinic, the drug which works in lung cancer patients that have -- harboring the eml4 translocation, this is tumor burden on the y axis, these are the ones harboring the eml4 translocations.

in metastatic mel low na patients harboring these mutations and this in patients with basal cell carcinoma with hedgehog activation. the context is generally not at straightforward as looking at one particular alteration. with inhibitors in patients that

have these alterations treated with the inhibitors tend to do well, however, when these mutations are present or wild type in the context of patients that have mutations in kras, these patients actually do worse. so understanding the genomic

context of tumors can help in potentially determining the best course of treatment for patients. so the question we've asked at tgen early on was is molecular profiling a rational approach to increase the options available to oncologists for treating

these patients, particularly those patients who have failed standard of care who have advanced chemo resistant cancer. so u one of the first studies to show the benefits of molecular profiling in the treatment of cancer is a study at tgen that was primarily funded by one of

our good friends jerry. van hos, performed a study looking at about 106 tumors, performed gene expression profiling and treated patients based on molecular alterations and then measured progression-free survival after first-line therapy and was able

to show that they could actually see a significant increase in progression-free survival when they treated patients based on the molecular profiles. this study, although landmark, was faced with several levels of criticism, one being the limited technologies that were used,

another one being ha they focused on a number of tumor types instead of focusing on one, and then finally that it wasn't a randomized study. so again thinking about the different molecular measurements and different types of mechanisms of alterations, we

see that many of the -- and this is just a small list of known cancer genes, but what we know is that many of these genes can be altered through various mechanisms, her2 can be amplified and overexpressed at the protein level, egfr can be amplified, mutated, overes

prexed, kras as well, that can be deleted or mutated, tumor suppressors like rb and p16, and then we also know that some genes are activated through structural events such as translocations and fusion transcript. previously we've had to rely on

a series of different technologies, methods and core capabilities that were required to be present to assess these different types of alterations. we want to look at point mutations, in many cases, we'd use a sequencing or genotype technology to identify copy

number changes like amplifications, we'd use cgh micro array technology for rna expression probably everyone in the room has run a -- array before to look for gene expression differences and to identify structural event like eml4elk, we still use figures

microscopy, but what we see is the development of amazing technologies over the last five to seven years, particularly next generation sequencing technologies, and these technologies are amazing because of course they've increased our ability to get to a human genome

by order of magnitude in terms of time and price, but for cancers, they're amazing because they allow us to make all the necessary measurements by identifying the types of alterations that one might look for in a cancer.

point mutations, we can i'd pie across the genome, copy changes, we can assess deletions, translocations, break points, inversions can be identified using these technologies. if we have rna available, we can also perform very detailed transcriptional profiling

measuring expression at the base level, identify particular isoforms and also look for fusion events. so these provide the platform for very comprehensive genomic interrogation in a single assay. so utilizing these technologies, we've now built out a platform

that is currently under fda based analytical validation, where this pipeline typically starts with the patient mountain doc and consent, through biopsy, sample and -- deep profiling, and utilizing a proprietary analysis tool that will take a somatic compendium report that's

a report that will contain all of the somatic events within a given tumor and match the events against a drug gene relationship database to generate a report that can then be discussed at a molecular tumor board, and then we can provide those recommendations back to the

treating physician. we currently have this pipeline, again, going at around 21 days and we're pretty sure that we can get it down to 14 days with improvements in our analytical pipeline. so the question now becomes can we use deep profiling using next

generation sequence approach to to oncologists. so instead of performing standard whole genome sequencing, we actually use a different approach which is a genome-wide approach, and we typically want to sequence both the tumor and the normal,

particularly at the dna level, and we do do a whole genome but it's a long insert whole genome methodology here instead of sequencing frackment fragments,we increase our ability to identify structural events so although we have low base coverage, we have very high clonal coverage across

break point, so by sequencing only to 50x coverage, we have -- i mean 5x coverage, we have 50x physical coverage across the genome, which is very powerful for determining copy number changes, identifying break points, translocations and inversions.

we couple that with at the dna level with -- sequencing, have roughly 100x coverage now, technology has allowed us to go to 250 to 300x using our current approach, it gives us a great sensitivity to identify point mutations down to about 10% allyl frequency, we believe we

can dw down to 5%. again having rna available, we perform rna sequencing of the tumor, and a reference -- there's no such thing as a perfect reference in differential expression so we use a different series of methods for gene expression

profile and identify overexpressed or underexpressed genes, but also a powerful way to detect mutations, particularly in tumors that have low tumor normal ratios, identifying the a mutation at low frequency in the tumor dna, sometimes we'll see that

mutation in the rna and it lends credence to the likely credibility of that particular mutation. we're leveraging the 2500 system. we were one of the first four labs in the world to have one of these set up and running, and

they're now commercially available, which will get us to normal tumor whole gene analysis within a 27-hour time period. we've built out a very comprehensive somatic bioinformatic pipeline that goes from standard data processing and management through typical

secondary analysis alignment and variant detection, all the way through identifying the genomic and clinically relevant events down to a report. part of the reason we've been able to improve this pipeline, and now our current research pipeline is about 12 hours, is

because of a collaboration we have with dell computing where we've been able to get access to some of their more innovative sandy bridge processing technology. again we've built out this platform, we initially called it jedi, some people loved it, some

people hate it. jeff actually went to high school with steven spielberg, so while we worried about running into problems using the word gedi. but here we have a somatic compendium report from a particular patient's tumor,

which will have all the mutations and translations and differentially expressed genes. we've also developed an in-house database which contains evidence and information that relates different genes and different perturbations to known therapeutics.

these two databases are merged through an algorithm that we call turbine to generate these clinical gedi reports. these will contain the gene of interest, the type of biomarker, the specific aberration, the evidence, whether it's direct or inferred, the particular drug or

class of agents that might be relevant to that alteration, and the indication whether that drug should be given, given that context. these tables have evolved over time and now we even provide information such as the frequency of that particular

mutation and the sequencing pileup and a series of other ancillary pieces of information that provide more information as the tumor board meets to make appropriate recommendations. so next i'd like to walk through a series of vignettes where we've applied this type of

approach to real patient sample samples, and looking at some of the outcomes of these studies. the first study is a project we did where we were focusing on patients with metastatic triple negative breast cancer. it was an amazing study. the clinical lead was joyce owe

shaughnessy, who's at baylor salmons in dallas, texas. she's also part of u.s. oncology and tex texas oncology and life technologies where we're able to apply this genomics knowledge towards treating patient with metastatic triple negative breast cancer, using the genome

and transcription approach to develop an integrative analysis pipeline as well as a genomics medicine knowledge engine that we've been able to modify and improve through time. and to assess the outcomes in the context of genomic and transcriptional profiles in

these patients, and i'd be remiss not to mention david craig, who is sort of my partner in crime in developing this pipeline, who is our deputy director for information sciences at tgen. the initial study was 14 patients, again, with clinical

diagnosis and metastatic triple i think it was really amazing that unlike some studies, we had a very nice diversity among our patient population where among the 14, eight were classified as white and six were classified and self identified as black or african-american.

you look at the ages, we had a few really young women in the study. we were able to assess tumor cellularity so we had very high quality specimens that led to really nice analyte. like the tcga, we're able to really understand the mutational

landscape of triple negative breast cancer, maybe not with the numbers, but it was interesting even with 14 patients, we had very similar numbers. the most frequently mutated gene in triple negative is p53, which again was shown by the cancer

genome atlas breast cancer group rb mutations, ptens, the alpha catenins, which i'll provide a few slides, because there were some interesting observations around alpha catenin, and erbb4 alterations. again we are rna sequencing data so we were able to look at

differential expression and able to compare the rna seq result from our patient versus a set of normal controls that were ethnicity and age matched, and one of the things that fell out among the vast majority of patients was we always tended to see this g2m checkpoint cassette

cassette, where foxm1 was always one of the most highly overexpressed gene. plk, ttk, which is msp1 and we're actually working on developing some new compounds around targeting ttk for a specifically in triple -- breast cancer, nicole lavender is

heading up those studies. we were also able to compare the gene expression pr profilesagainst the layman program, which was the group at vanderbilt that developed an algorithm to determine triple negative breast cancer subtypes, and the vast majority of our patients fell

into the basal-like classification, but we also had a series of patients that were deemed immunomodularity -- we also had one patient that had the luminal ar subtype. one of the other interesting observations was around p53 mutations, and this is something

that didn't fall out in the tcga data, was that we tended to see significant allyl-specific expression of the mutated allyl at the rna level, so when you're just looking at the dna level, many of the me mutations fellout to where they looked to be heterozygous where you could see

about a 5050 ratio of mutant to normal. but what we would almost always see was very high expression of the mutated allyl. this was interesting because none of these patients had loh at the 17q region encompassing the p53 locus, but what was also

interesting was we were seeing this type of expression of allyl even in the context of normal contaminating stroma. so you'd think there would be some normal p53 expression in the stroma, but we were seeing these real high levels of p53 expression of the mutant allyl,

even in the context of contaminating stroma, and we were able to search through and find a couple papers that it showed that there was this interesting phenomenon where mutant p53 may actually be turning off the expression of wild type p53 in the surrounding

stromal tissue. it's a phenomenon that hasn't been deeply studied but other people have seen this with a paper about a year or two ago looking at glioblastoma in cancer research that saw a somewhat similar observation. it's an interesting observation

that warrants further analysis. another interesting observation were these a alpha catenin homozygous deletions. this is just the re-pileups of one of the patients. here's the alpha catenin locus and flanking gene fil1. you can see in this patient

tumor this huge deletion where there's very few sequencing reads across the -- encompassing the alpha catenin locus and here we see a second patient. what's interesting about this is that elaine martis' group back in 2010 sequenced a triple negative african-american

patient, they had the primarily, the met and xenograft of the metastatic tumor, and they also identified a homozygous deletion in that patient that encompassed the alpha catenin locus. it's really interesting in that three of 16 triple negative -- both of our patients were

so that would be three out of seven triple negative patients that are african-american that have homozygous deletions across we actually were trying to filter through the cancer genome at las data, they had about 100 triple negatives in that study but very few were

african-americans so we would really like to look into the real frequency of alpha catenin -- we know it's involved in cadherin and cell invasion and metastasis, so is there a context around perturbations on loss of alpha catenin related to the more invasive phenotype that

we see among african-american women with triple negative breast cancer. so again, another study that we're following up on. so these are circle plots, many of you have seen this before. pardon the resolution. which will show the genes that

are perturbed around the outside, cot pi number changes, and the magenta lines represent translocations and structural events, and these are just six different circles from the different patients. i wanted to focus on patient two here, you can see this is a

60-year-old caucasian patient with recurrent disease after seven months o on -- so this patient actually had primary disease that came back, she then went on standard of care for metastatic triple negative, which is -- and she had also been randomized in a -- study

and we sequenced the tumor, and you can clearly see a series of copy number changes, some amplifications here, she had a series of deletions, but also a series of translocations and break points. if you look closely, you can see the sort of explosion of events

out of chromosome 7. so further analysis of this patient's sequencing data show that this actually was a series of events occurring across three different chromosomes that formed what's called a double minute. so you had breaks here, here,

here, here and here that formed one small circle, and these extra chromosomal circles are autonomous replicating. so you had this little circular dna containing different bits and pieces of different chromosomes that contain the braf locus.

so through the replication of this double minute, we were able to get a back clone though encompassed the braf locus and performed fish and you could see validation of the confirmation of these double my nowt minutesin her cancer. this patient, we went on to

assess clinical utility of the genomic data, the braf amplification and double minutes, there was also high level gain across the hraf locus and inpp4 down regulation, shown to be a tumor suppres suppressorin breast cancer that acts through the -- pathway, so the molecular

mechanism here, we believe, driving this cancer was -- but also with concomitant activation of the akt pathway, so the thought was it would be great if we could identify a combination trial where this patient could be treated with an inhibitor that would inhibit -- as well

as --, it was important to -- validate these events before we went forward and in this case, we were able to validate in downregulation and braf upregulation, we were able to work with the group at start, tony and the group in san antonio and the group at md

anderson and others who have a clinical trial with gsk where they actually were looking at combination therapies to hit -- and expanded -- i believe it was expanded phase 1. there weren't a lot of patients benefiting from that study. we talked to them about the

genomic analysis that we had for this patient, worked with gsk, were able to get this patient on that combination therapy and lo and behold, she had what joyce called an other worldly response, greater than 85% decrease in tumor burden with only two cycles on

combination -- which is actually fda approved now and at that time, the gsk214akt inhibitor. one of the common themes we saw in this particular setting were a convergence of mutations within these two pathways, and although this wasn't something necessarily new, we actually

were able to act on it therapeutically and show benefit for some of the patients where we saw a number of events affecting raf raf -- again in the same patient, and we had raf amply figures and overexpression of patient with an nf1 homozygous deletion that

also had a p10 homozygous deletion. we had a patient with the b raf and -- down regulation, we had a patient with fdxw7 homozygous deletion and activation and amplification of fgf -- one so we saw this sort of exciting dual pathway activation and were

able to act on it. so the basic outcomes to this study were activation of these two pathways in about 40 if not 50% of patients, some of the events that i just mentioned, we were among the first to show possible efficacy of the -- combination and there are

currently several trials ongoing currently looking at combination -- and triple joyce owe shaughnessy is running one and working with pat larueso to hopefully get another one off the ground. the results of the study were published in molecular cancer

research earlier this year highlighted as a significant the next application of this -- the application of this technology and case study is around metastatic -- bile duct cancer, can be associated with -- this is work we did with our partners at the mayo clinic.

so this particular patient, 50-year-old male, was superclavicular lymph node mix, collage yal carcinoma. the patient had been previously freeted with -- we sequenced this tumor and here is the -- you can see a series of mutations, a series of trabs

location events and intrachromosomal rearrangements one of the interesting things was that the path reports showed about 25 to 75% tumor sell lairt across a series of biopsies, but the sample that we sequenced actually looking at the allyl fractions in the sequencing

pileups suggested we were only dealing with about 10% if not 15 -- 15 to maybe 10% tumor content. we identified 35 somatic coding mutations, two genes why -- none were in any of the known commercial panels that are available.

one was flagged for their pew it tick coul context, i'm going tofocus on that. one copy number event, nothing flagged -- didn't flag anything, structural events didn't flag anything. you can see the -- plot, it's a pretty stable genome, nothing

sort of jumping up and down. we identified this one mutation in a gene called errfi1. pardon the resolution here. i probably should have zoomed in on the mutation, but this is the patient's normal genome, this is the tumor -- i'm sorry, normal exome, tumor exome, and rna

sequencing. and interestingly, we only saw the mutation in the exome at about 11% frequency, but 80% in the rna. again, another example of an important tumor suppressor like p53, where we see preferential expression of the mutated allyl,

again, no deletion across the locus, but preferential expression of the mutated allyl. this is a very, very important gene, again it's not in any of the cancer panels but why, i don't know. because erfi1 is actually an endogenous inhibitor.

biochemical and cell-based analyses confirm that the interaction contributes to egfr interactions, so errfi1 sits in the -- pocket of -- receptors and prevent them from dimerizing, thus preventing them from becoming active. this study actually went often

to show that if you put in wild type eerf, you can significantly decrease egfr phosphorylation, however, series of mutant constructs were not able to abrogate reduction in phosphorylation. our good friends actually show mig 6 -- they actually developed

a transgenic model of mig 6 by knocking out mig 6 and lo and behold, the knockout niceknockout mice developed carcinoma in other organs including the bile duct, very compelling evidence about the importance of re -- relating the importance of mig 6 to bile duct cancer.

you can see there's really nothing here, however, when we stain -- we see lots of staining, importantly membrane staining which was indicative of egfr activation, again, just a model of where it sits in this dimerization pocket and promotes -- of receptors.

through an inactivated mutation in preference with expression of the mutated allyl and potential sensitively to egfr inhibitors. and so we highlighted the -- and we also know and i showed earlier on that kraff mutations in the context of egfr mutation can prevent the efficacy of egfr

inhibitors, and having the exome data, we were able to show that there were no activating mutations or oncogenic -- so an egfr inhibitor was recommended after -- validation of the event. this is tumor before, and this is tumor after.

so about 40 or 50% tumor reduction and this is the metabolic activity of the tumor essentially when completely away after a few cycles on -- in the patient has a real nice stable response to egfr inhibitor. so we sequenced six metastatic cholangios, identified the --

context but the other incredible observation were fusion events and translocations including the fgfr2 oncogene, and i won't run through this other than to show this little line of red sequencing reads that are indicative of a translocation event encompassing the

fgfr2 jane, in this case it's partner bic1. now -- group in michigan also found these as well, and in clan jeel carcinomas and was able to do some really elegant functional analysis to show that these fusion events are actually ononcogenic, we validated theseby

finger, we have three different partner, all -- to the last -- of fgfr2 so essentially you get all of the kinase domain binding to these different genes, fusing to these different genes. a rule was able to go on to show that interestingly the regions that are retained within these

genes are oligomerization domain genes, so you're getting lots of -- heterodimerrizing together, and we were able to show significant overexpression of fgfr. trfthere was another paper that showed that delete dleeting the -- of fgfr2 is actually

oncogenic, so -- fusing it to -- domain activating fgfr2, so in these particular patients, fgfr2 activation sensitivity to fgfr inhibitors was the therapeutic context highlighted, and at the time, there were a series of agents being developed that target in many cases fgfr

and -- receptor, however, some are far more specific to the fgfr receptors, some of those being --. at the time, panatanib was being touted as probably the best fgfr inhibitor but it was sort of in the last phases of being approved by the fda and they

wouldn't allow us access to the drug because he it didn't want to threaten their fda approval, so the first patient actually went off -- and here's one of the tumors before and this is four months later, and this patient had a relatively stable and significant response over

about seven or eight months. later on that, patient actually was put on penatanib, this is looking at tumor density this, is tumor before, lots of cancer, and you can see the tumor is almost completely melted away after about six weeks on panatanib.

and here's more tumor, same patient, tumor. so this has led us to think about perhaps a bucket style trial for specifically around cholangial carcinoma where we take a series of patients and actually place them in a bucket based on molecular profiling so

patients within raf might go -- might go on fgfr inhibitors and patients with egfr activating events like erf -- 1 inactivation might go -- or some of the others. another way to look at this would be to do a randomized study where you place patients

who are in an adaptive study where you place patients in the appropriate bucket and your control arm is standard of care chemo. so we're trying to figure out what the best design might be for a study, but we have a number of clinicians who --

gastrointestinal clinicians and oncologists really excited about these discan cover res wanting to do clinical trials. so again, i hope what i've been able to show is the power of using a more comprehensive approach, use the long inserts to identify -- rna to identify

fusions and to provide competence for mutation calls especially in the context of low tumor cellularity, and -- sequencing for identifying mutations that might be in a lower content within a specimen. and just to further show the importance of rna, it can also

detect critically important therapeutically relevant isoforms, so this is a patient with pediatric gbm, the patient harbors a germline p53 variant. that's an acmg variant of unknown significance, but predicted to be deleterious, and we now know that the mutation is

transmitted in the family. interestingly the copy number analysis also show a significant high level amplification of egfr, which still does occur in gbm. but it was interesting that, you know, we were able to provide this information to the

oncologist who then went on to ask us, owe can kay, cgfr amplification but is it the v3 variant of egfr, which is really a highly oncogenic variant of egfr, but if all we had were exome data, we might be able to identify the amplification if we had cga --

whole genome, but if we didn't have the rna, we would not be able to know whether the three variants were being expressed because this particular mutation is known to occur in pediatric high grade gliomas and is relevant with response to targeted therapies in this

particular tumor type and we were able to show that rna sequencing data, these are sequencing reads, here you can see reads mapping to x on 1 of egfr and spliced over to x on 2, 3, 4, 5, 6, 7, and 8. so these are your standard egfr -- transcript, however, you

can see a series of reads that map from x on 1 and skip all the way over to x on 8, and this is the indicative profile for the egfrv3 variant, so by having the rna data, we're able to show with relatively high sensitivity and specificity that not only did the patient have high level

amplification -- but was expressed in the v3 mutation as well. so through our comprehensive approach, we've identified these events including in triple negative breast cancer, in looking at some of these other studies, rearrangements and

fusions, erfi1 mutations, and sensitivity to egfr inhibitors, again, this is very, very therapeutically relevant gene is not available in any of the known cancer panels, we're hoping it eventually makes its way, we know there's a group in spain that found a --

inactivated mutation in a patient with gbm. rna data also provides us additional power for identifying mutations and looking at allyl specific mutation, again the isoform, so we really believe in the prehen sieve approach is most suitable for capturing the

exed yum of somatic events that might occur with any given tumor to reduce the potential for false negatives when applying this technology at the clinic. lastly, i just want to talk, one of the major issues we know in cancer is heterogeneity, we know that there was a really

impressive paper that came out in the new england journal about a year and a half ago showing the complexity of any given tumor specimen, and i think it goes beyond a tumor, i think there's also heterogeneity within the context of the patient's cancer.

there's a tumor, then there's a and so we've been stating heterogeneity from a series of different approaches, and one i wanted to talk about was -- which we describe as clonal tides, we were able to show this whole genome sequencing in patients with multiple mile know

ma where we actually had a single patient, and we followed the patient longitudinally from diagnosis all the way through plasma cell leukemia, which ended up killing that particular cancer patient, unfortunately, so the patient was initially treated with --, and you could

see a significant reduction in tumor burden and then a relapse here. patient then went on -- a protease inhibitor, you could see a nice response, then another relapse. the patient went on another -- plus response, then -- plus a

series of other genes -- other therapies, plasma cell -- and again it was an amazing study because we were able to capture the primary tumor pre-treatment, the tumor at first relapse, the tumor at second relapse, at third relapse, as well as the plasma cell leukemia.

and sequence all four of those tumor specimens. what we were able to show was that there was -- i'll just sort of walk through some of this, that there were two clones at diagnosis. one made up about 95% of the tumor, the other made up about

5% of the tumor. and so what happened was the primary tumor was very responsive to -- but the minor clone was resistant. so what happened was there was an initial response and a first relapse. this was a tumor that came back.

this tumor was treated with proteasome inhibitors and responded. then there was a relapse. guess what happened? the first tumor came back. right? now, if the doc had known that there was this clonal tide, then

perhaps the patient could have went back on dex plus -- and maybe there would have been a response, but it's very, very rare that a doctor will go back to the same treatment twice, because we tend to think of resistance as a linear sort of evolution of tumors that have

become resistant to the first line of therapy, so we were able to show this clonal tied that you went from diagnosis to first relapse, back to a tumor that was almost the same, and then back to another tumor that also was similar to the first which was just an amazing study

that just goes to show that heterogeneity is not as simple as taking a tumor, breaking it into pieces, and saying that the pieces are different, but the patient's full context of cancer should be taken into account. so in summary, we're building out and refining this pipeline

so we can apply these technologies to provide options, again, we're not making treatment decisions, we're providing options that the doc can use to make a recommendation. we're working on a current randomized clinical trial

through our stand up to cancer study, which is specifically looking at melanoma patients with metastatic disease who are non- -- carriers, we're working with the fda very intimately and we're submitting an umbrella imd as well as an ibe type of submission to be able to bring

to bear the genomic technology on these patients to select a therapy that we believe is most relevant given the patient's genomic context or molecular some of these include knocking down our rule selection for matching genes to drugs, and again, these documents actually

are going to be submitted later on this weekend. we're hoping for the best. so we're refining our pipeline and we're applying it to a series of studies. again, i talked about the study, the cholangial cars momah study, the randomized stage it

2 clinical style, another study in glioblastoma, and then we have a longitudinal study to look at the natural history and heterogeneity specifically in multiple myeloma, so this one i'm really excited about, it's being led by really fantastic junior investigator, it we'll be

a thousand patients at diagnosis followed through, we're collecting tumor pre-treatment, and then we're going to collect tumor at relapse, and we're performing deep genomic profiling, whole genome exome rna on the primary as well as the relapse tumor so that we can

look at the natural history of this disease during the course of different treatments and clinical trials. of course there are a number of considerations and aspects that require further discussion and refinement. again sample qc tumor

cellularity and heterogeneity. we have to understand ha we're only treating the tumor sample that we biopsy. trying to make any other assumptions is moot. can we deal with needle biopsies versus full resections? we're currently able to perform

our deep molecular analysis using 18 gauge frozen needle biopsies. study design, strength and weakness of using panels, they're usually fast, they're very specific and sensitive because you can sequence deep, but there's a specific false

negative rate because you're going to miss all of the things that you didn't assess. the strengths and weaknesses of a comprehensive approach, the defects known to provide higher sensitivity, rna seq direct targets, and recording integrated information, but one

of the weaknesses is that it's a little more lengthy and a little more costly. clinical trial design, we've seen adaptive trials like the battle trial in lung cancer, we actually want to work more towards randomized -- to determine whether or not

molecular profiling is as good as, worse or better than the current standard of care in the advanced setting. one of the challenges is always access to pharmaceutical drug pipelines and portfolios. we're lucky to work with an amazing clinical collaborators.

one of the other issues is combinations, we're seeing multiple events occurring in patients and we need to combine some of these treatments which you just can't do on the fly because you've got to have at least minimally dosing studies, pk-pd, to know whether or not

you can combine two agents. the other important thing that we're doing is we're generating tumor grafts where we can, so we're taking teut more out of patients and planting them in animals, and we can actually test the efficacy of these combinations at least in a

preclinical setting to determine whether or not they're going to be effective or not. technologies, we've seen incredible improvements that provide benefits for sensitivity and specificity. we need standards specifically for look at soma tics, and we're

working with justin and mark at who've developed genome in a bottle, for clia but we need a reference for cancer, and we're working with some of our collaborators at one of the technology companies as well as mark and justin to develop a cancer standard for clia

and again, it's important to get all of this into clia, which we're currently doing. and speed. some people, i remember francis, the genome project was 15 years to get to one genome, right? and you know, to get this done in the clinic, the oncologists

need to get this data within two to three weeks, and that's what we keep hearing. and we believe we're there, but you know, i'll be honest, there was a big push towards a thousand dollar genome. i could care less at this point, give me a 24-hour genome.

it costs $10,000 to give you 24 hours, i'd rather have that than a thousand genome -- a thousand dollar genome that takes a month to get to. so at the end of the day, the promise is to deliver the most appropriate drug to each patient's tumor based on a

combination of molecular and evidence-based medicine. the challenge, of course, is does this approach lead to improved outcomes and effectiveness without sacrificing efficiencies to the healthcare system. so i'll stop there and entertain

any questions. thanks. >> thanks, john. we have the lights up and if people have questions, please approach the microphones in the aisles because that way, the people who are watching on the web can hear as well.

>> have you been able to see any therapy-induced escape mutation? >> oh, absolutely. absolutely. again, the question is, do those mutations arrive or is an evolution, meaning did the tumor -- did the bulk tumor or was there one clone that

acquired a new meuation based on the therapy? or wa that mutation already there in a minor clone that you just couldn't detect because it only represented one to 2% of the tumor mass? or you didn't sample that particular part of the tumor, or

maybe you didn't sample that tumor at all, right? you take a biopsy from the liver, the patient also has a lung net, and all of biopsy is the liver. does that mean that you've actually also represented the lung net as well?

so i think the challenge is to really understand, are we looking at tumor acquired mutations or are we looking at clonal evolution, meaning there was a clone that had already acquired that mutation. there's been some amazing work by a number of investigators,

whether it's dan haber, circulating tumor cells with that lung and the egfr t2 90m, i think that it was there in the circulating tumor cells. but when you look at the tumor mass, it was very, very low percentages, and matt larson's work almost 10 years ago now,

where he took a plural effusion from a patient with lung cancer and showed that the t -- i can't remember -- was there but it was only in about 2% of the tumor. so initial assessment of the tumor -- of the patient's bulk tumor didn't show the mutation. but when you look pleural

effusion and sequenced -- i think at the time it was -- they saw the mutation, so it was there in a minor clone, so again the question is, are there mutations that are arising specifically from from the tumor being bathed in chemicals? or are those mutations already

there, and would cause the mutation to occur in the first place? great question, though. >> john w respect to the three out of seven african-american breast cancer and homozygous alpha ka tee neen mutation, do you know whether there was a

germline mutation associated with that or was that an entirely -- if it's the latter, how do you account for that? >> they are somatic. they were somatic events, meaning the germline was essentially wild type across the region, and the tumor deleted

both. >> so why would that happen preferentially in an african-american woman? what's the concept? >> the concept, or my hypothesis, is that there could be a -- type that is associated with breakage of that renal.

and in a particular context, there's hyperbreakage at that region. if you happen to be the unlucky individual homozygous for that -- type, you get both copies. what you have to understand is that the deletions are

different, right? there's two deletions that sort of overlap and then you get the region of homozygous deletion, so the two different allyls, right, two chromosomes, in some cases, the whole chromosome -- one copy of one chromosome has been completely deleted, then

you have the deletion that encompasses the alpha catenin locus. so we see this type of ethnic enrichment in other diseases, i think, pml in latinos, and the break point that occur in the retinal -- translocations and fusions in pml, there are very

specific break points that you see almost exclusively in mexican-americans. so the question is, is there a germline am ploa type that is related to genomic instability at that region and preferential breaks occurring there. so to tease that out, i don't

know how we get there. we need to look at the frequency of the breaks, and then we have to take a careful assessment of the germline and compare that to the germline of patients who don't have the break point. >> i guess it's also possible if you don't see anything -- that

could be a transmutation? >> right. and the question is, if there is a -- type, is it in sis or is it in trans. these are the types of things that we need to -- you know, it's tough to get specimens. >> wondering when you're giving

recommendations for treatment for combination therapy whether you can get to the point where you're suggesting relative amounts of one particular -- instead of just a yes or no, give that drug or not. i also have one more question after that.

>> that's a great question. you know, dosing, drug sequencing, do you give one therapy first, and pulse, and then give the second therapy, and if so, in what order? i think there was a study out of yale that -- this amazing study, i think it was triple negative

breast cancer, look at doxycycline, showing with dox first, then hit with egfr, you see massive response, but if you give them together or in the other order, you don't see any response. and so that is an active area of research, and i guess the simple

answer to your question is no, we have not done ha yet. doing combinations at all is a challenge in and of itself, and i think -- but we're going to have to battle and tackle combinations because the vast majority of these tumors have concomitant act the vaiting

mechanism and events that we're going to have to target multiple things to see the best response. >> my final question is, when you have a mutated gene that -- overexpressed versus -- could it be that the rna that's mutated is do you have any -- wild type allyl could be meth lated, there

could be -- meaning normal control some and duplicated mutated chromosome. so it looks like there's no real loh. but -- does occur. or it could be an issue of rna stability, absolutely. >> hi.

so i work on sickle cell disease in africa, and one of the problems that we've found is that when we try to replicate some of the -- that have been found in african-american -- hasn't been able to find either number of frequencies or prevalence in the population and

it's very difficult to validate the results of new -- that we've identified in africa, in african-american population. so can you comment on your experience whether you've done any studies in african population of breast cancer and what would be the strategy for

fast tracking that kind of research, and what would be the chances of making sure that african populations get access to the drugs that are available in the country? >> wow, that's a mouthful. a lot to address. so i'm not surprised, i don't

think. we've been studying the genetics of populations of recent african-american ancestry for years. i've had some amazing collaborations with rick kiddles currently, one of the pioneers with ancestry, particularly at

recent populations of -- populations of recent west african descent. so i'm not surprised there's a -- amount of heterogeneity just within africa it self, much less between africa -- west africans and african-americans. and i think we have to look at

populations, i think we have to -- it would be great if we could first determine ancestry by looking at ancestry informative markers rather than saying here's a class of african-americans, here's a group of west afer be cans and asking the question can we

validate. is there a way for us to look at the african-americans and ask for a group or cohort am them that have ancestral allyls or patterns that look more west african. and then foe ks on that group and ask whether or not we can

validate in that group or not. but i'm not surprised that we don't see, and we've been lucky, one of the studies i'm really excited about, we have a komen programs grant at the university of michigan looking specifically at breast cancer stem cells, triple negative breast cancer,

and within that study is one of the aims is to look specifically at women from ghana, and that's part of the work being led by lisa newman and one of her clinical fellows who were collecting specimens triple negative breast cancer specimens for women in ghana, we're

implanting those tumors in mice so we'll have tumor graft models and we're performing deep sequencing, and hopefully evelyn is going to present some of it at this meeting in atlanta, but you're right, because one of the things we want to do is we've got a small number of

african-american triple negative sequence now, so the question is, how similar or different will the tumors from ghana look compared to african-american women and the tumors that we've looked at here? then finally, i think what you're talking about is policy,

you know, making these approaches and these technologies applicable for everyone. i'm at nih but i don't work for the government anymore so i can say it. you know, perhaps the affordable care act will help us get some

of that. i'm hoping it does. will it be the only thing to help us get there? no. i think we're talking about culture, we're talking about the way we practice medicine today, i think all of these are factors

that we're going to have to deal with. it's great to see john rustin's group, i had an amazing conversation with him, hoping that -- it's part of his vision to help us understand how to get there, understanding disparity, understanding the challenges we

have in dealing with disparities, and trying to make these technologies available to our first study with triple negative breast cancer was almost 50% african-american. if nothing else in the study, we felt that in itself was a win, to work with clinicians who

understand the impact of the disease in all people and try to provide access to these cutting edge technologies regardless of race or socioeconomic status. >> so there's a reception in the new faes space up the hall here, which you're encouraged to come and continue the conversation

with john carpten over coffee and cookies, but i think right now we should thank our speak he again for a very stimulating