Investigating the Complex Biology of Drug Resistance for Clinical Impact

April 02, 2024

Information

Yale Cancer Center Grand Rounds | March 22, 2024

Presented by: Dr. Katerina Politi

ID11532

To CiteDCA Citation Guide

00:00Good morning, everybody.
00:01Thank you for being here.
00:03Welcome to Grand Rounds.
00:06This is the this Grand Rounds is
00:09in a special location, obviously,
00:11because we are linked today to the
00:13first of what we hope will be a
00:16really successful series of annual
00:18translational science retreats
00:20meant to highlight the amazing
00:23resources that are present at Yale
00:26Cancer Centre for people who do
00:29translational science and also to
00:33highlight some of the amazing stories
00:35that that have come out of this work.
00:37And so no one better to to be our
00:41inaugural speaker than Doctor Katie Politi.
00:44Katie studied biology at the University of
00:47Pavia in Italy and then moved to New York,
00:50obtaining her PhD in genetics
00:53at Columbia University.
00:54She then joined Harold Varmus's
00:56lab at Memorial Sloan Kettering
00:58and began her life's work on the
01:01molecular basis of lung cancer.
01:04She continues this work at Yale,
01:05now as a professor in the Departments
01:07of Pathology and Internal Medicine
01:09in the section of Medical Oncology.
01:12Her laboratory is focused on studying
01:14the biology of lung cancer and
01:16uncovering mechanisms of resistance to
01:18targeted therapies and immunotherapies
01:20in in this disease.
01:22She's also got a keen knowledge of
01:26essentially every mutation that's
01:27ever been described in lung cancer.
01:29And I know that doctors often call
01:31her up and say what drug should I use.
01:34She Co leads the cancer signaling
01:36networks research program.
01:38She's the scientific director of
01:40the Center for Thoracic Cancers,
01:42Co Director of the Yale Sport in
01:44Lung Cancer and recently elected
01:46to the ACR Board of Directors.
01:48So we're really appreciative that
01:50you're going to kick us off today
01:53the the ID number there is to record
01:57your attendance and then we'll
01:59have questions both in the room
02:02and online when when we're done.
02:05Thank you.
02:10Thank you very much, Barbara,
02:12for that wonderful introduction
02:15and thank you very much for
02:17having me as a speaker today.
02:18It really always is, I think,
02:21very special to speak at one's own
02:24institution and then especially
02:26also associated with this first
02:28translational science retreat.
02:30So I'm really excited about this.
02:32And today what I'm going to do is
02:34I'm going to tell you about some of
02:37the work that we've been doing over
02:39the past few years in the laboratory.
02:46These are my disclosures.
02:50So we have a long standing interest
02:52in the lab on studying lung cancer.
02:55And as all of you know,
02:57there are several histological
02:58subtypes of lung cancer.
03:00But one of the things that we've learned
03:03over the past 20 or so years is that
03:05lung cancer is not one entity and that
03:08there are in addition to different
03:11histological subsets of the disease,
03:13there are also are a variety of laser
03:18pointer of molecular subsets and in
03:21particular in lung adenocarcinoma.
03:24Through various sequencing efforts,
03:26different mutations in genes that
03:30encode either receptor tyrosine
03:32kinases or downstream signaling
03:35components of receptor tyrosine
03:37kinase signaling pathways that
03:39regulate cell proliferation and cell
03:42survival have been identified as
03:44you can see here in this pie chart.
03:45And I think one of the things to
03:48really highlight is what we've
03:50learned over the years is that
03:52these mutations are in addition to
03:55being molecular to establishing
03:57molecular subsets of the disease.
03:59They really also are clinically
04:02relevant because different targeted
04:04agents have been developed that can
04:06you be used to block the activity
04:09of these mutated oncogenic drivers.
04:11And in particular and in the work
04:13that I'll tell you about today,
04:15for example,
04:15mutations were found 20 years ago
04:18now in Exxon's encoding the kinase
04:21domain of the epidermal growth factor
04:23receptor after in about 15 to 4050%
04:28of lung and nocarcinomas depending
04:31on which population you look at.
04:35And these are mutations that
04:39confer sensitivity to EGFR tyrosine
04:41kinase inhibitors.
04:42But there are many other
04:44targeted therapies as well.
04:45So you can have rearrangements in
04:49the anaplastic lymphoma kinase and
04:52targeted therapies that are effective
04:54in that and so on for a number of
04:57different oncogenic drivers and lung cancer.
05:00And so this has really transformed the field.
05:02And so if we look at this diagram here of
05:07approved FDA approvals for lung cancer in,
05:10in recent years,
05:11what you'll see is it really has
05:14been an explosion in FDA approvals,
05:16especially from the early 2000s in the
05:192000 and 10s and approvals now also
05:22in the first part of the twenty 20s.
05:25Most of these agents that were
05:27approved recently have been targeted
05:29agents and that really is linked to
05:31the discoveries of these molecular
05:33subsets of the disease.
05:34But also do I think one of the things
05:37that has been emerging also in the
05:39past 10 to 15 years really are the
05:42approvals of immunotherapies that
05:44we hear a lot about agents that
05:47are targeting immune checkpoints
05:48like the anti PD1,
05:50anti PDL ONE Access and CTLA 4.
05:53And so this has really been
05:55transformative in a lung cancer.
05:57And I'd like just like to point out
06:00how in recent analysis what we're
06:03seeing is that there's actually
06:05a decrease in mortality from lung
06:08cancer in recent years.
06:10And in the study published in the New
06:11England Journal of Medicine a few years ago,
06:13it was really shown that the
06:15decrease in mortality from lung
06:17cancer can't be accounted
06:19for just because of a decrease
06:21in incidence of the disease.
06:23But is likely reflects actually
06:26advances in the care and in the new
06:30therapeutics that have emerged,
06:32including in particular in the
06:33years that were studied in
06:35this paper for targeted agents.
06:37And so I think this is a really nice
06:41example of how what we've learned over
06:44the years from from the biology and
06:47from the genetic studies of tumors
06:50really is having a profound impact
06:52for patients with this disease.
06:55And of course I would be remiss if I
06:57didn't point out how immunotherapies
07:00have also been transformative.
07:02And I think the continued decrease
07:04in mortality that we are continuing
07:05to see is actually going to show how
07:08it isn't only the targeted therapies
07:10but also the immunotherapies that are
07:12really contributing to this decrease
07:15in and mortality from lung cancer.
07:17So if you know you look at this,
07:20there's really these advances
07:21have been tremendous.
07:22But what we do know is that both
07:25primary and acquired resistance
07:27to targeted therapies and to
07:30immunotherapies are common.
07:32And here you can see an example of
07:35scans from a patient with a tumors
07:38with AK Ras G12C mutation treated
07:41with AK Ras G12C inhibitor and
07:44you can see the tumor regresses
07:47but then comes back and you have
07:49this is acquired resistance.
07:51And here if we look at this plot
07:54taken from a review looking
07:57at studies of immunotherapies,
07:59you can see that across various
08:01different indications but including
08:03in lung cancer here that in clinical
08:06studies of immunotherapies,
08:07the response rates or to immune
08:09checkpoint inhibitors are not super high.
08:12We're not talking 7080% the way we're
08:14talking with some targeted therapies.
08:16Not only that,
08:17but also we see acquired resistance
08:19commonly emerging.
08:20So there's a lot of work that needs
08:22to be done to really understand and
08:25optimize treatments for both targeted
08:27agents and immunotherapies and to
08:30understand mechanisms of sensitivity
08:31and resistance to these agents.
08:33And So what do we do in my lab?
08:37And as part of the research program,
08:41we are really interested in understanding
08:45mechanistically biological processes
08:47that are involved in cancer.
08:49We like to integrate these with
08:53studying and addressing clinical
08:55challenges and investigating specimens
08:57and data from patients with cancer.
09:00And really the hope is that the work
09:02that we do collectively as a group,
09:04this is work that we do with many
09:07different people is to discover
09:09things that will discover findings
09:11that will lead to clinical trials and
09:14new therapeutic approaches to patients.
09:17Central to our research program is
09:20the use of biological specimens from
09:23patients and analysis of these specimens.
09:27And I think this slide is also going
09:29to be showed later in the day as an
09:31example of one of the resources that
09:33we have as part of the lung cancer
09:35program to really be able to collect
09:40and use specimens from patients.
09:42And this is just one of the examples
09:44of one of the resources I think
09:46you'll hear about a couple
09:48of others later on as well.
09:49But really an effort that started many,
09:51many years ago working initially
09:55with Scott Genger and Anna
09:58Wertz and Roy Herbst and many,
10:00many people in this room now with
10:03Sarah and many of all of the thoracic
10:06oncologists on the team and pathologists.
10:09Kurt for example,
10:11really working on collecting specimens
10:13from patients who have advanced
10:16lung cancer through treatment,
10:17especially at the time of resistance.
10:19So that then we can take these
10:21specimens and analyze them,
10:22generate patient derived models.
10:24And really these have contributed extensively
10:27to the work that I will tell you about today.
10:30And so I put a little cryovile here.
10:34And So what I'm going to do through the talk
10:37is when you see a cryovial on the slide,
10:40it actually is an example of data
10:44that we've been able to analyse and
10:46use because of the specimens that
10:48were collected through this approach.
10:50So you'll see that throughout the talk.
10:53So what what am I going to tell
10:56you about today.
10:56So I think as most of you know
11:00we have a long standing interest
11:02in studying the biology of EGF
11:04receptor driven lung cancer.
11:06And so when patients and really the
11:09focus that we've had at least in
11:13the in the past or until recently
11:14has really been and because of the
11:16sort of the clinical landscape has
11:18really been on advanced metastatic
11:20EGF receptor driven lung cancer.
11:23And so when patients are diagnosed
11:26with EGF receptor driven lung cancer,
11:29now they're mostly treated with tyrosine
11:33kinase inhibitors most recently and
11:35in the United States especially the
11:37tyrosine kinase inhibitor awesome.
11:39Merton if this is one of the newer
11:42generation of agents that has more
11:44activity on mutant EGFR compared
11:46to wild type.
11:48So hopefully decreasing its toxicity
11:50and has been shown to have superior
11:53progression free survival and overall
11:55survival compared to standard of
11:58care earlier generation tyrosine
12:00kinase inhibitors in this disease.
12:02And so this was an A really
12:05important advance in the field.
12:06However,
12:07what we do know is that still
12:09resistance or acquired resistance two
12:12asamertinib occurs almost inevitably
12:16and it actually isn't very commonly
12:20associated with on target EGFR mutations.
12:24And this is different from some of the
12:27earlier generations of tyrosine kinase
12:29inhibitors that instead where we saw
12:32commonly one most frequently observed
12:34on target EGF receptor mutation,
12:36the T79 TM mutation.
12:38But you see additional mechanisms of
12:41resistance met amplification for example,
12:44so a bypass signaling pathway
12:46being one of the more common.
12:48Then we see a histologic changes in
12:51the tumors that occur quite frequently,
12:53but then most of the mechanisms
12:55of resistance are really not known
12:57and poorly understood.
12:58And so one of the things that we've
13:01been interested from when as we
13:04think about these problems is really,
13:08really understanding these tough
13:10challenges like really understanding
13:12this part of the pie chart, right.
13:15What are these mechanisms of resistance,
13:17What is happening in these tumors
13:20where we don't really
13:22have a key genetic alteration that
13:24has changed that or clear process
13:27that is happening that we can target.
13:30And so just a couple of thoughts
13:33that sort of guide our thinking.
13:35Targeted agents are probably not sufficient.
13:38We need to discover new untapped
13:41vulnerabilities of oncogene driven lung
13:44cancers and then the tackling resistance
13:47requires new knowledge of the links between
13:50cancer cell plasticity and the tumor
13:53microenvironment and tumor heterogeneity.
13:55And so these are some of the and so I
13:57think of these that like the the not the
13:59low hanging fruit but the fruit really
14:00at the top of the tree that we're trying
14:04to really grasp and understand when we.
14:07And and really if we look at EGF receptor
14:10driven lung cancer and we think about it,
14:13one of the things that we know is
14:15that with with the targeted agents
14:17that I've told you about today is
14:20we do see this acquired resistance.
14:22But not only that.
14:23We also know that when we use the
14:25targeted agents they don't completely
14:27eradicate all the tumor cells and
14:30there's variability in the depth and
14:32duration of responses in patients.
14:34And you can see this really in this
14:36waterfall plot where there's some
14:38tumors that shrink dramatically
14:39and others that shrink less.
14:41And so we've been interested in the
14:44question of what accounts for this
14:47heterogeneity and disease progression and
14:49sensitivity to tyrosine kinase inhibitors.
14:52And so the first thing that I'm
14:54going to go through is some of the
14:56work that we've done to study how
14:58different EGF receptor mutations can
15:00actually have distinct properties.
15:02And so first of all,
15:06I've sort of told you about EGF
15:08receptor mutations and one could think,
15:09oh, we can lump them all together.
15:11But in reality,
15:13what we do know and what is becoming I
15:16think increasingly clear in recent years
15:19is that you have their different EGF
15:22receptor mutations and not only that,
15:24the different EGF receptor mutations have
15:28different properties both biological,
15:30biochemical and also in terms
15:32of TKI sensitivity.
15:34And so when we look at
15:36EGF receptor mutations,
15:37there are two major categories of mutations.
15:40There's the L858R point mutation and then
15:44there's a set of small in frame deletion,
15:46some of them more complex and Exxon 19.
15:50The most common of these is
15:52this E 746 to a 750 mutation.
15:54But then there are these other in
15:57Dells that are found at, you know,
15:59variable frequencies in these tumors,
16:00but they exist.
16:02And So what does it mean?
16:04Are all these mutations alike?
16:05Well,
16:06one of the things that we know is that
16:08even if you just broadly categorize
16:11the L858R mutations and the e.g FRXN 19
16:14deletion mutations and you look at the
16:17survival curves on ossumertinib from
16:19the trial of frontline osumertinib,
16:22you see that even just the
16:25Exxon 19 deletion mutations,
16:27the overall survival is about
16:2840 months in that study.
16:30But for the L858 Rs,
16:32it's about 33 months.
16:33And this is consistent over across
16:36different tyrosine kinase
16:37inhibitors that are used.
16:39So the L858R subset does worse with
16:43TKIS compared to the Exxon 19 subset.
16:46We also found several years ago in
16:50work that we did together with Sarah
16:53Goldberg and Mark Lemon is that that
16:56there's a small in frame deletion
16:59in a Proline insertion mutation and
17:02one of the Exxon 19 deletions that
17:05actually if you look at that mutation
17:07and you look in upon treatment with
17:09irlatinib this was a few years ago.
17:11So one of the early generation
17:14tyrosine kinase inhibitors that the
17:16progression free survival duration
17:17of a treatment overall survival were
17:20all worse for the for erlontinib in
17:22that subset compared to the more
17:25common Exxon 19 deletion mutation.
17:27And this along with some laboratory
17:30studies really piqued our interest in
17:32studying these differences a little bit more.
17:35And here you see the cryovile appear.
17:38This is also work that was Zenta Walther
17:42was really central to helping us
17:45identify these patients for this study.
17:48And so working with lots of different
17:52groups here we were able to show that
17:54this proline insertion for example what
17:57you see in Western blots is when you
18:00treat with tyrosine kinase inhibitors,
18:02it's less sensitive to various
18:05tyrosine kinase inhibitors compared
18:07to the canonical e.g.
18:09FRXN 19 deletion mutation.
18:11Not only that,
18:12when you actually go and look biochemically,
18:14and this is work that was spearheaded by a
18:17former student that Mark Lemon and I shared.
18:20Eris von Alderweil,
18:22von Rosenberg showing that this
18:24proline insertion mutation has AKM for
18:28ATP that is more more closer to the
18:31wild type in contrast to some of the
18:34other variants that instead are more
18:36sensitive to tyrosine kinase inhibitors.
18:38So really is that affinity of the
18:41kinase for ATP that is probably
18:43rendering it more resistant to
18:45these tyrosine kinase inhibitors.
18:46So really from the clinical observations,
18:49from some of the laboratory
18:51studies going to the biochemistry,
18:52we're really able to figure out what
18:55was happening with this variant.
18:57And this led to work that we did
19:00together with Mike Grant and Sarah
19:03Goldberg really putting together a multi
19:06institutional cohort of patients with e.g.
19:09Fr XL19 deletion mutations treated
19:11with asumertinib because we wanted to
19:13look at the tyrosine kinase inhibitor
19:15that was really clinically relevant
19:17for patients right now and that was
19:19being used to see what outcomes
19:21were for patients with this Proline
19:24insertion mutation with asumertinib.
19:26It's pretty rare.
19:27So you have to really work together and put
19:30together a cohort from various institutions.
19:33And so Mike and Sarah assembled
19:38this cohort including data from
19:40our Yale cohort and actually showed
19:43that in patients whose tumors have
19:46this proline insertion mutation
19:48treated with ossomatinib,
19:49you have worse progression free survival.
19:53Then if you look at the common e.g.
19:55Fr XM19 deletion mutation,
19:57the overall survival isn't quite
19:59statistically significant,
20:01but you can see that there is a trend
20:04in in in in worse outcomes there as well.
20:07And So what does this mean?
20:09What does this make us think?
20:11I think the message here is that
20:15not all mutations are the same.
20:17And now we have the tools and the drugs
20:20to better match mutations with therapies.
20:22We aren't the only ones who
20:24are thinking about this.
20:25There's some other work from
20:28Jacqueline Robichaud and John
20:30Haymack's group at MD Anderson,
20:32work from Christine Lovely at Vanderbilt,
20:35all really pointing in this direction.
20:37We need to know about the biology,
20:39the biochemistry of the mutations,
20:41and that can help us think about
20:44perhaps how to better optimize these
20:46therapies now that we have them.
20:48Another point, yeah,
20:50the structural and biochemical
20:51understanding of the effects of
20:53the mutation can guide predictions
20:55for TKI sensitivity and resistance.
20:57And of course,
20:58the other question that comes along
20:59is how do we translate to the
21:01clinic this to the clinic now what?
21:03What are the next steps that we can take?
21:05So we can test trials of like optimal TKI.
21:10So now we have all these reagents,
21:12we can test other agents and other
21:14drugs on these different variants
21:15and see if there's some that are more
21:18effective for specific mutational subsets.
21:20But then the other question is,
21:22are there other agents that we
21:24should be thinking about for certain
21:27subsets of the disease in combination
21:29with also Mertinib?
21:30And I think this will be a
21:32recurring theme throughout the talk.
21:33So for example, you know,
21:35should we be thinking about specific
21:37antibody drug conjugates or other
21:39approaches to target tumors with that
21:42don't do as well with monotherapy?
21:44Awesome.
21:45Or so after you know thinking
21:48about the different.
21:49So we talked about how different EGF
21:52receptor mutations themselves can
21:54have an impact and have distinct properties,
21:56but what about Co mutations?
21:57How can Co mutations influence tumor
22:01progression but also TKI sensitivity.
22:05And so many years ago now,
22:07I probably started working on this
22:09actually almost exactly 20 years ago
22:12when EGF receptor mutations were discovered.
22:14I think it was May 2004 that I started
22:18generating these mouse models.
22:20We generated genetically engineered
22:24mouse models of EGF receptor driven
22:27lung cancer in which we could express
22:30the EGF receptor mutants inducibly
22:32in the lung epithelium.
22:33And this was really these were really
22:35to be able to study the biology
22:37of the disease.
22:38And we've used these mice extensively
22:41over the years to study signaling by
22:44mutant EGF receptor discover resistance
22:46mutations to tarsine kinase inhibitors,
22:49identify therapeutic strategies to
22:51overcome or prevent and or prevent
22:54drug resistance and study the
22:56effects of targeted therapies on
22:58the immune microenvironment.
22:58And here you can see MRI images.
23:00We use MRI imaging for our mice to
23:04look at the lungs and see or you can
23:05see lungs full of tumors you treat
23:07them with a tyrosine kinase inhibitors,
23:09the tumors shrink and go away.
23:12Over time the tumors come back and
23:14we can study those resistant tumors.
23:16So a few years ago we decided to
23:20upgrade our our mouse model and
23:25use a slightly different system
23:27that would allow us then also to
23:29be able to modulate other genes.
23:31Because we know that EGF receptor mutations
23:33and lung cancer don't occur in a vacuum.
23:35There are other mutations in the tumors there
23:38and we wanted to be able to model that.
23:40So we decided to take this still
23:44this tetracycline inducible EGFR
23:46allele across it to another mouse.
23:50That in which using Cree recombinase
23:54you can then turn on expression of the
23:56reverse tetracycline transactivator
23:58which can bind the tetromotor in
24:00the presence of doxycycline and
24:02induce expression of EGF receptor.
24:04And we also crossed it to AP
24:0653 phloxed allele.
24:07But using this mouse what happens
24:10is we can deliver Cree recombinase,
24:12we deliver it with a Lantivirus
24:16into the lungs of mice,
24:18turn on mutated EGF receptor.
24:20Simultaneously we can delete P53.
24:23And here's some images,
24:24these are the lungs of mice.
24:26You can see the by MRI,
24:29you can see here by Histology and a a
24:32bigger magnification of the Histology.
24:35So we said OK,
24:37so we have this mouse model with now
24:40EGFR and mutants and P53 deficient tumors.
24:42The P53 deficient tumors are higher grade,
24:45they're nastier.
24:46I see Rob Homer here.
24:47He has helped us extensively over the
24:49years characterize and study these tumors.
24:52And so one of the questions that
24:54we had is well in addition to P53,
24:56what role do other mutations in
24:59EGF receptor play in EGF receptor
25:01driven lung cancer?
25:02How do they affect tumor progression?
25:04How do they affect TKI resistance
25:06and how do they affect the molecular
25:08properties and phenotypes of the tumors?
25:11And So what we did is we worked with
25:13a colleague at Stanford University,
25:15Monty Winslow,
25:15who had developed an approach in and
25:19used it in K Ras driven tumors to
25:21really be able to inactivate using CRISPR,
25:25CAS 9 technology,
25:26different tumor suppressor genes
25:29simultaneously in the lungs of mice.
25:32So not all of them in the same cell,
25:34but you can deliver this kind of
25:37pool of lentiviruses and in different
25:39cells you can then inactivate
25:41different tumor suppressor genes.
25:42And then you can use a computational
25:45approach that he developed called
25:47tumor barcode sequencing which
25:49based on various controls that are
25:51spiked in and based on barcode IDs.
25:54You can actually look and quantify
25:57the effect of inactivating that tumor
26:00suppressor gene on the number and
26:02size of tumors in in, in a screen.
26:05It's essentially a way of doing
26:06an in vivo screen.
26:07And so we applied,
26:09we took this pool of lentiviruses
26:13targeting different tumor suppressor genes
26:15that were frequently altered in lung cancer,
26:19not necessarily in EGF receptor driven
26:21lung cancer but in lung cancer and
26:22he had used it in the K Ras model
26:25previously and so we applied it to our e.g.
26:27FRL 850 at RP53 model and in particular
26:31we had also crossed the model that
26:33I just told you about with one
26:34that has an inducible CAS 9 Ileo.
26:36So these are experimental animals here.
26:39These are controls because
26:40they don't have CAS nine.
26:41You can't do CRISPR CAS 9 mediated genome
26:44editing when you don't have CAS 9:00.
26:46So we transduced the lungs of the mice,
26:50waited 11 weeks and then took the lungs
26:53of the mice and did tumor barcode
26:57sequencing in our control animals.
26:58When you look at the relative
27:00tumor size compared to controls,
27:01you don't really see any.
27:03The tumor suppressor gene
27:04inactivation doesn't have any effect,
27:06but that's because you don't have CAS 9,
27:08so you shouldn't see anything.
27:10So that was reassuring.
27:11What do we see in the mice with CAS 9?
27:13So one of the things that we saw is
27:15that when you inactivate APC from the
27:19wind signaling pathway RBM 10 and RB1,
27:23these three tumor suppressor
27:25genes when inactivated had the
27:28biggest effect on tumor growth.
27:30So the tumors grew faster when
27:32you were inactivating these tumor
27:34suppressor genes compared to controls.
27:37We also noticed interestingly
27:40that SET D2 and LKB 1,
27:43both of these putative tumor
27:45suppressor genes I'd say actually had
27:47a negative effect on tumor growth,
27:48which was quite interesting
27:50and is and I'll go,
27:51I'll tell you a little bit
27:52more about that in a minute,
27:53but it's a topic of interest,
27:55interesting work that we're doing.
27:56And then there were a number of
27:58tumor suppressor genes that really
28:00had no effect on tumor growth.
28:02We went ahead and we validated
28:04these using single SGRNAS.
28:06This is towards APC and this is
28:09to RBM 10 which is an RNA binding
28:12protein and a splicing factor.
28:14And you can see that when you
28:16inactivate them you see these bigger
28:19tumors and tumors progress faster
28:21than in the EGF receptor P53 model.
28:24So what does this mean though
28:26in the context of human cancer?
28:28And so if we,
28:31what we did at that time is we
28:34actually interrogated the ACR
28:35Project Genie database,
28:37which is a large data set that has a
28:39lot of mutational information that
28:41has been contributed to this data
28:44set from various institutions that
28:46are from their tumor sequencing
28:49efforts at their institutions.
28:52And when we look in this data set at e.g.
28:55F RP53 driven tumors and we look at
28:58the frequency with which there are
29:00alterations in this Co occurring
29:02tumor suppressor genes,
29:03you actually see that the top hits
29:06RBM 10 RB one and APC are where the
29:09top hits in our functional screen in mice.
29:13So we think that our screen in mice
29:15is actually telling us something
29:17about the functional relevance of
29:20these alterations in the human
29:21tumors and arid 1A didn't come out
29:23in our screen at 11 weeks,
29:26but we actually did another time
29:27point at 19 weeks and it popped up.
29:29So perhaps it's more important later
29:32in tumorigenesis And interestingly
29:34Genes SDK 11 is LKB one,
29:36it's really not frequently altered
29:38and that was the one that I showed
29:41you seemed to have a negative effect
29:43in our in vivo screen.
29:45So we've actually,
29:46this has been a really powerful
29:49system and we've actually been able
29:52to do broader screens with more
29:54genes and try to learn a little bit
29:57more about what genes are important
30:00for the progression of these tumors.
30:02And I'd just like to highlight
30:04an example of work that we
30:08did continuing this continuing
30:11this effort with D2G Oncology,
30:14a company that was founded Co
30:16founded by our collaborators
30:18Monty Winslow and Dmitry Petrov.
30:20And we work together on doing
30:23this screen of additional tumor
30:25suppressor genes in the context of
30:28EGFR tumors but also in the context
30:30of K Ras driven tumors for example.
30:32And you know I just like to go back to
30:36LKB one for example showing how this
30:38has a negative effect on EGFR driven tumors.
30:41It's not really a contributing,
30:45it doesn't really Co occur
30:47mutationally with EGFR driven tumors.
30:50So it seems to be like a synthetic
30:52lethality with these tumors.
30:54But it's an amazing contrast with what
30:56we see in Keras driven tumors where it
30:58is one of the major drivers of tumor growth.
31:01And so this is I think telling
31:03us and it's frequently mutated
31:04with Keras in human tumors.
31:06So we're really,
31:09we're really think that this is a
31:11cool system to be able to understand
31:14how Co occurring alterations
31:16impact the fitness of tumors.
31:18And Fran Exposito in the lab is
31:21really working a lot to understand
31:23this synthetic lethality and is
31:25doing experiments to knock it LKB
31:29one out and established EGF receptor
31:31tumors and see what happens and
31:33also to understand mechanistically
31:35what is happening in these tumors.
31:37So stay tuned for for data on
31:40these studies that I think will
31:42be really fascinating.
31:43And there are some other targets that
31:45we're studying along these lines as well.
31:48So I think a very powerful system.
31:50We've also used this approach not just
31:54to study mechanisms of tumor progression,
31:57but also use this type of approach
31:59to really understand what genes
32:01can modulate the sensitivity to
32:03tyrosine kinase inhibitors.
32:05So we did the same experiment and instead
32:09of just waiting and collecting the tumors,
32:12what we did is we also had an arm
32:14where we treated for two weeks with a
32:16tyrosine kinase inhibitor osumertinib.
32:18You see here the tumors go away
32:20or they're shrinking mostly.
32:21They're not completely going away at
32:23two weeks, but you do see a response.
32:25And so we did the same tumor bar
32:27code sequencing and what we found
32:29here is so this is the,
32:30this is the plot that I showed you earlier
32:33looking at what is affecting tumor growth.
32:36Well,
32:36when we add Asamertinib,
32:38one of the things that we saw is that
32:41keep 1 the tumor suppressor gene,
32:43keep one that really didn't have much
32:45of an effect on the growth of the
32:48tumors in the absence of drug now
32:50limits the sensitivity to Asamertinib.
32:52In other words,
32:54the tumors aren't shrinking as
32:56much as wild wild type
32:58or control tumors do
33:00when keep one is present.
33:02What do we think is happening here?
33:03Well, we know that keep one is important
33:07to sequester NRF 2 in the cytoplasm.
33:10When you knock out KEEP 1,
33:12NRF 2 can then go into the nucleus and
33:16activate antioxidant response elements and
33:18those gene expression programs that allow
33:22cells to really withstand oxidative stress.
33:24And when we take our mice and we just use
33:28an individual SGR and a targeting keep one,
33:31these are the control mice
33:32that don't have CAS nine,
33:34you use Asamertinib, the tumors go away,
33:37you don't really see anything
33:38left in the lungs.
33:39But if you have the experimental mice
33:41that have CAS 9 and you use the SGR and a
33:44targeting keep one treat with Asamertinib,
33:46you see tumors are still left over.
33:49And so again,
33:50what does that mean for patients?
33:52So at the time what we did is we
33:55worked with Jessica Hellier and Heather
33:58Wakeley at Stanford University who had a
34:01collection of data from patients with e.g.
34:04F RP53 driven lung cancer and looked at
34:07whether there were mutations in genes
34:09in the keep one access in these tumors.
34:12And you can see here in this blue line,
34:14the patients who had mutations in the
34:17keep One access in their tumors had
34:19a shorter time to treatment failure
34:22compared to controls suggesting that if
34:26you have alterations in this this program,
34:30this antioxidant response response program,
34:34you're going to have limited sensitivity
34:37to tyrosine kinase inhibitors.
34:40And so I think one of the things that
34:43we're really seeing emerging from this
34:46work looking at the tumor suppressor
34:49genes is that when you do have mutations
34:52or you have alterations that Co occur
34:55with EGF receptor and with EGF receptor
34:59P53 these can modulate both the growth
35:01and sensitivity to these agents.
35:04We we were interested in looking
35:06further and in work that Paul
35:09Stockhammer who was a resident is
35:12now a hospitalist here and is an
35:15incoming he monk fellow did recently.
35:19He looked at both our Yale internal data
35:24from our tissue collection program.
35:26You see the cryovial here,
35:29but also at the ACR project gene data set
35:32and looked at outcomes for patients on
35:38tyrosine kinase inhibitors whose tumors
35:41had different combinations of mutations.
35:44And I think the take away here is he
35:47was able to look at tumors that had
35:49mutations in a subset of tumor suppressor
35:52genes because tumors had been analyzed
35:54across a wide variety of different platforms.
35:57So we had to sort of focus in on the the,
36:01the common subset of tumor suppressor
36:03genes that were looked at across platforms.
36:05But essentially if tumors had both
36:10P53 mutations and a mutation,
36:12at least one of these tumor
36:13suppressor genes that he looked at,
36:15they had worse outcomes.
36:17These are EGFR mutant tumors even
36:19compared to mutations that just had
36:22TPF 3 mutations and were wild type for
36:25those different tumor suppressor genes.
36:28And So what does that mean?
36:29Again, I think we're identifying a subset
36:32of tumors where there may be a benefit
36:35from adding a different therapy or it
36:38should be at least be investigated from
36:40the get go because they are likely to
36:44have worse outcomes with monotherapy
36:46tyrosine kinase inhibitor treatment.
36:48And this is very relevant right now
36:50at least in the field of EGF receptor
36:52driven lung cancer because there are
36:54studies of chemotherapy plus asamartinib
36:56in the first line that are positive.
36:59But people are very reluctant to
37:01give that combination to everybody.
37:03If we can identify people who might
37:05benefit more or might need it more than
37:08that could be really helpful for deploying
37:10these different strategies in the clinic.
37:13And then I think another point is that
37:16we're really learning the Co mutations
37:19can affect therapeutic sensitivity
37:20and it isn't only in the context
37:23of EGFR tyrosine kinase inhibitors.
37:25This is happening in multiple contexts
37:28and with with multiple agents.
37:30So here an example,
37:31I'm just just giving you a few examples.
37:33There are many more in the literature.
37:35But if we look at keep one,
37:37going back to keep one, keep one,
37:39alterations seem to have been negative
37:43for response rates to Sotirasip
37:46in K Rash G12C driven lung cancer.
37:50Worse,
37:51you know higher local recurrence
37:54with chemo radiation in the context
37:58of immunotherapy LKB 1 mutations
38:01actually seem to be worse confer,
38:04you know be worse for or describe,
38:06define a word a subset that
38:09does worse with immunotherapy.
38:11And so in conclusion for this
38:14part of the talk,
38:16the nature of the oncogenic mutation and
38:18Co occurring mutations effects sensitivity
38:20to Tkis and mechanisms of resistance.
38:23We've developed a new generation of
38:26genetically engineered mouse models that
38:28can be used to study these complex genotypes.
38:31And I'd like to point out that
38:33really we have a lot of work that
38:35is happening now studying these
38:37individual different components.
38:39Mariana Do Carmos,
38:40an MD,
38:41PhD student in the lab.
38:42She's studying the role of RBM 10
38:46in EGF receptor driven lung cancer
38:49working with Luisa escobarahoyos lab.
38:51Because we really can
38:55join forces and Luisa is an
38:57expert in splicing and this is
38:59really important gene protein
39:01that is involved in in splicing.
39:03So we're doing that.
39:04I told you about Fran's work.
39:06We have Kita who's working on KMT 2D,
39:10which I didn't tell you about
39:12another potential target
39:13that came out of this screen.
39:14So really we can really study
39:16these different genotypes and
39:18understand the biology of these
39:20different complex genotypes,
39:21which is really exciting.
39:24We have found out that an activation of
39:27these different tumor suppressor genes
39:29can have different effects on both
39:31tumor growth including positive and
39:32negative effects and TKI sensitivity
39:34depending on the oncogenic context.
39:37We showed that keep one loss limits
39:40sensitivity to osmertinib in mice
39:42and in patients and think that
39:45this is really potentially a bad
39:47actor if there's Q1 alterations
39:50either at the genetic level or
39:52also alterations in the pathway.
39:54The pathway can be modulated
39:56in many different ways,
39:57and tumor suppressant gene mutations
39:59can be used to identify patients,
40:02subsets of patients who are likely
40:05to have worse outcomes and could
40:08be considered for additional
40:10therapeutic interventions.
40:11So in the last part of the talk,
40:15I'd like to tell you about some
40:18other work that we've been doing
40:21more recently to study non mutational
40:24mechanisms of resistance and I'd
40:26say also of persistence.
40:28So on tyrosine kinase inhibitors.
40:30And So what are some of the things
40:34that we're thinking about broadly
40:36in the lab when we think about this
40:38problem of this 50% of tumors that
40:40we don't what for which we don't
40:43know why a resistance emerges.
40:45So some of the things that we're
40:47really interested in in understanding
40:49and studying are how the tumor
40:52microenvironment effects resistance
40:53and persistence.
40:54And this is work that we're doing
40:57collaboratively,
40:57Jake Schillo in the lab doing
41:00collaboratively working with Don
41:02Nguyen's lab.
41:03We are studying lineage plasticity
41:06and tumor heterogeneity.
41:08And I'll tell you about an example
41:11of this that was just recently
41:13published this month and that comes
41:16out of work studying mechanisms
41:19of tumor persistence.
41:20And of course another area that
41:22we're really interested in is while
41:24we've we're talked a lot about genes
41:26and mutations and genetics here,
41:27but are there ways of reading out
41:30pathways and learning about how
41:32pathways are altered in tumours
41:34which might be an important way
41:37of understanding resistance
41:38and persistence as well.
41:40And so one of the non mutational
41:42mechanisms that we recently
41:44discovered and published on,
41:46I'm not going to tell you about that
41:48today because I don't really have
41:50time is that we identified a role
41:52for the ATP as of the SLY sniff
41:55complex in mediating resistance
41:57to tyrosine kinase inhibitors and
42:01SMARCA 4 is actually usually lost,
42:04you have loss of function mutations
42:06in tumors.
42:06One of the things that we found
42:08which was really interesting is that
42:11actually it seems to be important
42:13for the resistance phenotype because
42:15in resistant tumors it actually can
42:18promote accessibility of chromatin
42:20at both cell proliferation genes but
42:23also at genes it are NRF 2 low size
42:27so that allow activation of genes
42:30that are antioxidant genes with that.
42:32So it links to that keep one,
42:35keep one finding that we had in
42:37our tumor suppressor gene screen.
42:39So I'm not going to tell you about this,
42:41but I did want to highlight it
42:43as as one of the some of the work
42:46that we have done recently on non
42:48mutational mechanisms of resistance.
42:50What I really wanted to focus the last
42:53few minutes of the talk on is telling
42:55you about some work that we've been
42:57doing to study tolerance and persistence
43:00to tyrosine kinase inhibitors.
43:02And you saw this waterfall plot earlier.
43:06But one of the and one of the
43:08questions that that we've had and I
43:10think that is a prominent question
43:11in the field is why aren't all cells
43:14eradicated upon TKI treatment,
43:15right, Because if we could get rid
43:17of all of the cells from the get go,
43:19we wouldn't have the problem of acquired
43:21resistance. And here's some scans.
43:23You see the tumor and you see several
43:25months later the tumor is still there,
43:27there still is some residual tumor leftover.
43:30So what is the biology of residual disease?
43:33Well, we decided and this is work from
43:35a former graduate student in the lab,
43:38Boom Yao,
43:39who who is now in Arno Osher's lab
43:41as a post doc.
43:42And I think Boom Yao is here.
43:43I thought I saw him.
43:45And So what Bom Yao did is he took
43:48advantage again of our collection
43:50of specimens from patients.
43:52And he said, well,
43:53what happens if I implant these
43:56PDXS that we've generated,
43:58treat them with a tyrosine kinase inhibitor
44:00and then look at residual disease?
44:03We can harvest that.
44:04You know, we take it at a plateau, right?
44:07Once the tumors aren't shrinking anymore,
44:09that's what's left over.
44:10And can we we it's really hard to
44:12study residual disease in patients.
44:14We can't really easily do biopsies
44:16on treatment,
44:16but this is as a surrogate of that.
44:19And so here are some examples of
44:22the PDXS that Boom Yao studied.
44:25So he took these PDXS,
44:27treated them and then took what was
44:29leftover after four to six weeks
44:31of treatment when they plateaued.
44:33And you can see in all of the cases
44:36there was tumor leftover after treatment,
44:38varying amounts of tumor and in
44:40some very little,
44:41very small islands of tumor,
44:43but there was tumor leftover.
44:44And I'd like to highlight an example
44:46of one of the things that we found
44:48from one of these PDXS that we
44:50studied in a little more detail.
44:52We found that in one of them we
44:54saw up regulation of Ascl 1.
44:57ASCL one is a basic Helix loop
45:01Helix transcription factor.
45:02It has a role in neuronal differentiation
45:04and its expression actually identifies
45:06a subset of small cell lung cancer.
45:09So it was really up in the residual
45:12disease in this tumor and not only
45:14was it up at the transcriptional
45:17level and the signature was was
45:19enriched in the residual disease,
45:22but it's downstream targets rat BCL two
45:25and DLL three were also all turned on in
45:29the residual disease in in that tumor.
45:32Ossumertinib was working really well.
45:34You can see phospho EGFR is gone here.
45:37And so this was really interesting to
45:39us because we know that a subset of
45:42EGFR driven tumors when they're treated
45:44with osumertinib can actually undergo
45:48neuroendocrine differentiation and
45:49transformed to small cell lung cancer,
45:53a subset of which are ASCL 1 positive.
45:56And so this kind of piqued our interest.
45:59And so one of the first questions that
46:02we had was are these ASCL one cells
46:05present in the tumor pretreatment.
46:07And so when we looked and we did
46:09single cell RNA sequencing,
46:11we actually saw that the if you look at
46:15the pretreatment specimen here in blue,
46:17there is a subset of these cells that
46:20is present that is ASCL 1 positive.
46:22So we think that those cells
46:25were present beforehand.
46:26Whether other cells then turned it on,
46:29we can't really tell from the
46:30types of experiments that we did.
46:32But we do know that there was a
46:34population that was there pretreatment.
46:36And so our next question after that was
46:39well how is ASCL 1 conferring TKI tolerance,
46:42what is happening.
46:44And so we said OK,
46:46let's turn to our human EGF
46:48receptor driven cell lines and let's
46:50express ASCL one in these cells.
46:52And so one of the first things that we did,
46:54we expressed ASCL one in the cells and you
46:57can see here in this HCCA 27 cell line,
47:00we expressed it and we saw more colonies
47:02and you can see this quantified here
47:05after treatment with osmertinib
47:06compared to the empty vector control,
47:08we did this across in another cell line
47:11and we saw no effect of ASCL one expression.
47:14And so this was also interesting and we said,
47:17OK,
47:17so why does ASCL one having a
47:20phenotype has a phenotype in one
47:22cell line but not the other.
47:24We did gene expression profiling and what
47:26we saw is that in the permissive cells,
47:29these HCC 827 cells,
47:32you actually saw that ASCL one could
47:35lead to an EMT gene expression
47:38program was it had no effect at
47:41all in the PC-9 cell line.
47:43And we went on and we looked with ataxiq
47:47at chromatin accessibility at EMT genes
47:51and we see that upon ESAS CL1 expression,
47:54you do see changes in chromatin
47:57accessibility at both epithelial
47:59genes and mesenchymal genes when
48:02you put Ascl one into these HCC
48:06827 cells that are permissive,
48:08but you don't see any changes
48:09in the PC-9 cells.
48:11And So what do we think is happening?
48:14So we think that when you have,
48:18when you don't have ASCL 1,
48:20the TKI can work and you see death
48:22of the EGF receptor driven cells.
48:25If you have a permissive cellular
48:28context what happens is that
48:30you can have ASCL one can turn on
48:33or can lead to an EMT program and we
48:36know that that is associated with
48:38resistance to tyrosine kinase inhibitors.
48:41In a non permissive cellular
48:43context though that you don't have,
48:46you don't turn on this program so
48:48you don't have a difference in ASCL 1
48:51expressing versus non expressing cells.
48:53We also found that pre-existing
48:55cells with transcriptional features
48:57of drug tolerant cells are present
49:00in the untreated tumors.
49:01And I think one of the questions that
49:03we've we're really interested in is you
49:05know why are some cells permissive or not.
49:08I think this is sort of one of
49:10the major problems in cancer,
49:11one of the things that has been a
49:13mystery in cancer over all of the years.
49:14Why do you see certain phenotypes when
49:16you have certain settings and not others?
49:18And in the case of ASCL one,
49:20this is very reminiscent of
49:22reprogramming because it's known,
49:24for example,
49:25that you can put ASCL one into
49:29fibroblasts and reprogram them to neurons,
49:32but you put them when you put them
49:34in keratinocytes.
49:34You can't and this has been shown
49:36to be due to actually the chromatin
49:39landscape at Ascl,
49:40one target genes in the different cells.
49:42So could something like that be
49:44happening in the cancer cells as well?
49:46And one of the other questions of
49:48course that we have is since Ascl
49:50one is important for and neuronal
49:54differentiation,
49:55it's associated with neuroendocrine
49:57differentiation, Is it poising these cells?
49:59We didn't see any other, you know,
50:02neuroendocrine markers on,
50:03but is it poising the cells to
50:06undergo that type of change?
50:08And so,
50:10so some of the things that we're thinking
50:12about now and we have experiments ongoing,
50:15we have Mark Wiesehofer in the lab
50:17who's been thinking about this and
50:20working about on this in the context
50:22of both prostate cancer where very
50:25similar things happen and lung cancer.
50:27We're asking how does a chromatin
50:29state of a cancer cell affect
50:31responsiveness to therapy and plasticity.
50:34And so you can have these different cells,
50:35you add ASCL one and you can
50:37see different things happen in
50:38these different cells.
50:39And why is that happening?
50:41And is there something that we can
50:43learn from these cells that then
50:44we can apply to human tumors and
50:46could we use this information?
50:47I'm thinking far a little bit far ahead,
50:50but it's something that's in the back of the,
50:51my mind is can we predict how a tumor
50:54will evolve on treatment with this knowledge.
50:58So finally a couple of final thoughts.
51:02So what have I told you today,
51:04baseline mutations and Co mutations
51:06can affect disease progression,
51:08drug sensitivity and mechanisms
51:09of drug resistance and how can we
51:12incorporate this knowledge into
51:14clinical investigation and practice.
51:16This is something that we think about a lot.
51:19There's a vast heterogeneity and
51:21complexity of non mutational resistance
51:24and persistence mechanisms and
51:25we're working to identify them,
51:28establish when they are relevant
51:29for specific tumors and find
51:31vulnerabilities of these and be
51:33happy to talk more about these
51:35thoughts throughout the day.
51:36Today I there are a lot
51:39of people to acknowledge.
51:41Here are some pictures of lab
51:44members throughout the years.
51:46Here's a particularly fun one.
51:50This was a fundraising picture
51:52for a closer to free team that so
51:55I thought that was pretty cool.
51:57These are Halloween,
51:59one of our Halloween parties and
52:02other pictures from the we have the.
52:05All of the lab has contributed
52:06tremendously to all of these
52:08efforts over the years,
52:09and I'm so grateful to have
52:12the opportunity to work with
52:13so many talented people.
52:15There are lots of people to acknowledge
52:18who have contributed to this work
52:20in addition to members of the lab,
52:23so many collaborators outside of Yale,
52:27but in particular everybody here at Yale,
52:29which I, I, I really,
52:31I hope everybody is on this slide.
52:35It's one of the things that I was
52:37worried about but want to make
52:39sure that everybody is acknowledged
52:40here because of the tremendous
52:42contributions that makes it such
52:44an amazing place to work together.
52:47A couple of things that I'd like to say,
52:49we have a retreat too on thoracic cancers.
52:52On Monday, it's retreat season.
52:55It is at West Campus,
52:57so you're all invited to join us.
53:01We have a team that has been working.
53:04Sarah's in here, I think.
53:05Sarah Goldberg, Justin Blasberg.
53:07We have Glynis Arnold and Melody
53:10Noga MENA who's been working
53:12to organize this retreat.
53:14So we hope you can join us and then
53:18save the date for our annual lung
53:20cancer workshop on June 12th and 13th.
53:23It is also going to be at West
53:25Campus here and it's particularly
53:28special this year because we are
53:31going to be recognizing the 20th
53:33anniversary of the discovery of EGF
53:35receptor mutations and lung cancer,
53:37which has really transformed the field.
53:38It's near and dear front to my heart
53:41as you can imagine from the talk,
53:43but it's really going to be I think a
53:46spectacular event with lots of people
53:48coming from all over to mark this,
53:51this moment.
53:52And so we hope that you can
53:55participate in that too.
53:56Thank you very much and I'll
53:58be happy to take questions.
54:10Thank you so much, Katie.
54:11That was wonderful.
54:12Are there questions in the room?
54:16Maybe I'll start as a person who
54:18knows more about squamous cell
54:20cancers than adenocarcinomas.
54:22When you talk about P53 mutations,
54:25are they always the same
54:27in adenocarcinoma patients?
54:28And we spend a lot of time
54:30in the squamous world talking
54:31about disruptive mutations,
54:32gain of function mutations. Yeah,
54:36we have, I think there's a wide variety of
54:40P53 mutations that you see in lung cancer.
54:44So they're like different types and
54:47have you dissected out if they
54:49have different implications.
54:50We think the gain of function
54:52mutations don't lead to as much
54:53genomic instability for example. Yeah,
54:55those are things that we
54:57haven't studied that much.
54:58I think Paul had looked at the
55:01different mutations a little
55:02bit in terms of outcomes,
55:04Paul Stockhammer and I don't
55:05think he had found differences in
55:07terms of outcomes with Tkis with
55:09the different classes mutations.
55:12So is the polycommers suppressor
55:16name screen that your
55:19biggest hit at least in one
55:21of the assays was loss of RB,
55:24but it looks like in the in the cancers
55:27RB loss was relatively infrequent.
55:29Does it does that suggest or have
55:31you looked at whether there's other
55:33dysregulators of the RB pathway that
55:35are more common in lung cancer like
55:37the Cyclone CDK pathway and that's
55:39a potentially targetable approach?
55:42Yeah, that's a great question.
55:43So it's interesting because RB as you
55:47said RB one loss is one of the biggest
55:51drivers of tumor growth in our screen.
55:55It is also if you look at how frequently
55:59it Co occurs with EGFR and P53 mutations,
56:02it's one of the tumor suppressor genes
56:04that is most frequently Co altered.
56:06So none of them go really
56:08above the like 10% threshold.
56:12We do know, we haven't really looked at
56:15other ways in which the P50 in which the
56:17RB pathway could be altered in tumors.
56:19We haven't really looked at that.
56:21What we do know is that if
56:25you have tumors with e.g.
56:27F, RP53 and RB alterations,
56:31those are the ones that have the
56:33highest likelihood of undergoing
56:35that neuroendocrine differentiation.
56:37And so like 1/4 of those will undergo
56:39the neuroendocrine differentiation.
56:45Any other questions from.
56:47OK, Thank you again so very much. Thank you.