"Synthetic Lethal Therapy of TP53-mutated Head and Neck Cancer" and "Environmental and Sex- Specific Molecular Signatures of Glioma Causation"

December 02, 2020

Information

Yale Cancer Center Grand Rounds | December 1, 2020

Barbara Burtness, MD and Elizabeth Claus, MD, PhD

ID5950

To CiteDCA Citation Guide

00:00What you know it's 1202 or
00:02why don't we get started?
00:05And I know there are folks still logging on,
00:08so for those of us who are here,
00:12thank you for joining cancer grand rounds.
00:14I hope all of you had a restful
00:18and enjoyable Thanksgiving and.
00:20Obviously, I know we're all
00:22looking forward to year's end and
00:25hopefully celebrating a better 2021,
00:27but we're really very fortunate to have
00:31two exceptional speakers today and I'll
00:34start by introducing our first speaker,
00:36who frankly needs no introduction.
00:39Doctor Barbara Burtness is
00:41professor of Medicine Co,
00:43leader of the Developmental
00:45Therapeutics research program.
00:46And leader of the head neck cancer program
00:49and Yell Barbara is internationally
00:51known for her leadership in clinical
00:54development and research and
00:56understanding the biology of heading
00:58that canceran among her many accolades.
01:01We can add now only in the past
01:04month is the principle investigator
01:06of the head and Explorer of which we
01:10are just so proud of both Barbara
01:12and the entire team being awarded
01:15this really coveted an elite grant.
01:17Of which I think will be from
01:20not correct if I'm not mistaken,
01:22there only two head neck spores
01:24now in the United States are the
01:26leader of one of them,
01:28which is an extraordinary distinction for
01:30the people who work in this space at Yale.
01:33So Barbara was kind enough to
01:34share with us the work she's doing
01:37on head neck cancer and forever.
01:39Thank you for joining
01:40us today. Well,
01:41thank you for the invitation and for.
01:44All the support that's gotten us this far.
01:47So what I wanted to do was talk
01:49about P53 mutated head neck cancer,
01:52which is something I have a
01:54longstanding interest in.
01:55And. Obviously, P 53 is a very
02:00critical tumor suppressor gene.
02:03It's meant to be the cells way of
02:05reacting to cellular stress signals,
02:08and among these are many that we know
02:11are important in head neck cancer.
02:14So hypoxemia, DNA damage,
02:16replicative stress,
02:17and ideally in response to these P.
02:2053 is activated and promotes the
02:22transcription of target genes and
02:25domains of cell cycle arrest DNA
02:27repair a pop ptosis and others.
02:29However,
02:30in head neck cancer were aware that
02:33P53 functionally disrupted in the
02:36majority in HPV associated head neck cancer.
02:40P53 is wild type, but its degradation
02:43is fostered by viral proteins,
02:46an in HPV, negative head neck cancer.
02:50Over 85% have genomic disruption of P53,
02:53including in frame mutations,
02:55truncating mutations and missense mutations,
02:57and you can see here that many of
03:00these are clustered in the DNA binding
03:03domain and we know that this type
03:07of mutation is Villa terius for the
03:10Natural History of head neck cancer.
03:13So this figure comes from a large trial
03:16that the legacy Kog Cooperative Group
03:18ran over 500 respected head neck cancers.
03:22All respected to margin,
03:23negativity,
03:24and all offered appropriate risk
03:26based animal therapy is with
03:28standard at the time and then P 50
03:31three was sequenced and you can
03:33see here that long term outcome.
03:36Was worse for those patients
03:38who had P53 mutation,
03:40and if you classified the mutations
03:42as disruptive or nondisruptive,
03:44it was worse for those with
03:46disruptive mutation and the definition
03:48that was used in this paper.
03:51That was for disruptive was a
03:54mutation that was either truncating
03:56or in the DNA binding domain.
03:59So on the basis of these outcome data,
04:02we were interested in the cognitive
04:05Akron Head Neck Committee,
04:06which I chair in studying intensification
04:09of therapy for these poor prognosis
04:11patients with disruptive P53 mutation.
04:14But the first thing we wanted to
04:17do was examined how we really
04:19should be calling the P53 mutation.
04:22So we started with what we called
04:24the poeta rule,
04:26so those were the rules from the
04:28paper I just showed you and we
04:30compared them to 14 other cloud 13
04:33other classifiers that are out there,
04:35many of which are based on in
04:37silico predictions of disruption,
04:39some of which are based on experimental
04:42evidence actually of the decrease in
04:44OIF 1 activation for every specific
04:46mutation and then we also examine Dar
04:49poeta rules augmented with information
04:50about the splice site mutations.
04:52And you can see that this very simple
04:55definition of truncating or DNA
04:57binding domain actually outperformed
04:59in terms of clinical prognosis.
05:01All of the other indicators,
05:03and so in our clinical trial.
05:06We moved forward with this poeta
05:09rules plus splice site mutations
05:11and the trial that we're now about
05:14halfway through
05:15is a randomized phase. Two trial of.
05:18Post operative therapy for patients
05:20who meet the criteria for radiation
05:23but have negative margins,
05:25don't meet the criteria for chemotherapy,
05:27and then we want to ask in those
05:30patients with disruptive mutation,
05:32do we see an advantage for the
05:35addition of platinum that we
05:37don't see in other patients?
05:39This is supported by.
05:41Is Bisquick grant that takes care of
05:44all of the sequencing and we have
05:47two investigators who are doing
05:50the mutation calling in real time.
05:53So continue to support this trial
05:55and see this is kind of an important
05:58resource in terms of all the sequencing
06:01information that we're going to
06:03have on top of the clinical outcome.
06:07We also have support for a clinical
06:10trials planning meeting at the NCI
06:12which is going to happen in January.
06:14The goal of this is to write trials
06:16both for locally advanced and
06:19recurrent metastatic disease,
06:20identifying promising therapies
06:21for P53 mutated cancer.
06:23We also want to develop a
06:26national infrastructure.
06:27For the sequencing and mutation,
06:29calling with the consensus approach
06:30that all of the groups within the
06:33NCT and will accept the breakout
06:35groups for this have been meeting
06:38for about five months now.
06:39I can tell you that the focus is
06:42very strong and immunotherapy and
06:44synthetic lethal strategies and I'll.
06:46I'll mention both of those in in
06:49the remaining minutes of this talk.
06:51So head neck cancer is one of the
06:54cancers where it appears that increase
06:57tumor mutation burden is predictive
06:59of response to immunotherapy.
07:01And we know that this is a cancer with
07:04a higher number of nonsynonymous mutations,
07:07particularly in the HPV negative cancers.
07:10And in the platinum refractory setting,
07:12both for Pember Lizum app in this early
07:15single ARM trial and for development
07:18in a randomized phase three trial.
07:21In the also in the platinum
07:24refractory setting,
07:25in both cases we see that as
07:28tumor mutation burden rises,
07:30the likelihood of benefit
07:32from immunotherapy increases.
07:34So working with my long term
07:36collaborator at Fox Chase Circle,
07:39Columbus,
07:39we wanted to examine whether or
07:42not mutations in not only P.
07:4453,
07:44which is the most common most
07:47commonly mutated tumor suppressor
07:48and head neck cancer,
07:50but also CDK into a which is
07:53mutated in slightly over half
07:55of HPV negative cancers as well.
07:57See how these related to DNA damage
08:00as reflected in tumor mutation
08:02burden with the idea of establishing
08:05whether or not P53 mutated cancers.
08:07Would be particularly susceptible or
08:10appropriate for study with immunotherapy.
08:13We had access to a data set of 1010
08:16HPV negative cancers that have been
08:19profiled at Caris Life Sciences.
08:21There gene panel is about a 600 gene panel.
08:25They exclude HPV associated cancers
08:27with standard methods and then the
08:29CDK into a mutations that we saw were
08:32almost invariably truncations or deletions.
08:35So we included all of those,
08:37but for P53 we were interested once
08:40again in what's the best way of calling?
08:44Meaningful mutations,
08:44so we started with the American College of
08:48Medical Genetics variants calling
08:50this included essentially all the
08:52P53 mutations that that we SPA.
08:55We then looked for consensus between
08:57the ACM G and two other variant calling
09:01algorithms, interference linver.
09:02We use the International Agency for Research
09:05on Cancer guidelines for what was dominant,
09:08negative or loss of function.
09:11We then looked at the variance is defined
09:14by the poeta rules that I just alluded to,
09:18and then we called out those patients who
09:21seem to have gain of function mutations,
09:24most of which are defined experimentally
09:26across a range of publications,
09:28and TMB was.
09:29Measured just by counting all nonsynonymous
09:32missense mutations across the about 1.4
09:35mega bases that are included in this panel.
09:38This shows you the the patient
09:41characteristics so predominantly
09:42oral cavity in order.
09:44Pharynx cancers as we see in the clinic.
09:47Males outnumbering females,
09:48and as you see at the bottom,
09:52the number of patients who had P53
09:55mutation by the karris was higher
09:57than if we looked at the consensus
10:00calls or the disruptive call gain of
10:04function was less than 10% of all of
10:07the mutations that we saw in be 53,
10:11and indeed it turned out that either P.
10:1553 or CDK into a mutation.
10:17Was associated with an increase in TMB.
10:20Here we looked for threshold of 15
10:22per per per megabases as being likely
10:25predictive of response to immunotherapy.
10:28And you can see that across the board
10:31having both genes mutated was associated
10:33with higher TMB than having one or the other,
10:37and The only exception here was that
10:40those patients with gain of function
10:42mutations in P53 did not have an
10:45increase in tumor mutation burden.
10:47So you know,
10:48we concluded that mutation of P53
10:50or CDK in two ways associated with
10:53increased tumor mutation burden.
10:55This is highest when they're damaging
10:58mutations in both jeans and so just to
11:01kind of segue to the next part of the talk,
11:04where I'm going to talk a little bit
11:06more about synthetic lethal strategies.
11:09P53 mutated head neck cancer, I think,
11:12remains a really important subject for study,
11:14because it's common.
11:16It has a poor prognosis.
11:18We still don't,
11:19after many decades of people examining this,
11:22have agents which directly target mutated P.
11:2553.
11:25And so the increasing evidence that
11:29synthetic lethal strategies might have
11:31promise in these patients is has kind
11:34of attracted our attention in in the lab.
11:37And so one of the things that we
11:41know about disruptive P53 mutation
11:43is that you lose the cell,
11:46loses the ability to perform cell
11:49cycle arrest at the G1 S transition,
11:52and as a result it becomes much more
11:55dependent on transition at G2 M and so
11:58we I mean obviously many people have
12:01been interested in this across many cancers,
12:05but we were interested in
12:07examining some of the.
12:09Potential targets that regulate G2
12:11M We know that auroras increasing.
12:13I'll show you a little bit about this.
12:16We know that Aurora expression
12:18is increased in head neck
12:20cancer and. Aurora content
12:22will go up at the end of G2.
12:28Its activity is required to localize
12:32CDK one to the to the centromere to to.
12:39Foster mitotic entry Aurora also,
12:42in addition to its roles
12:45in centrosome maturation.
12:46It also has the property of.
12:51Activating the city.
12:52See 25 phosphatase,
12:54which removes an inhibitory phosphorylation
12:56from CDK one and on the other hand,
13:00it's important to know that that
13:04inhibitory phosphorylation is
13:06placed by the mitotic checkpoint
13:08kinase we want and so both we
13:11won an Auror recognized is up
13:14regulated and head neck cancer.
13:16Both of them are potential.
13:19Points of synthetic lethality
13:21in P53 mutated cancers,
13:23but they appear to have kind of
13:26contradictory or opposing roles,
13:28and so that the data that I'm going
13:32to show you now will try to make
13:35the case that by Co treating these
13:38cancers with an Aurora inhibitor,
13:41which will lead to.
13:43Abnormal spindle formation.
13:45Defective cytokinesis,
13:46but by inhibiting the rural will
13:49lose the ability to remove the
13:53inhibitory phosphorylation from
13:54CDK one and that will result in
13:58cell cycle arrest that we can
14:00counter that by inhibition of
14:03we won so that phosphorylation
14:05isn't placed and accelerate these
14:08cells into mitosis where given
14:11the spindle disruption that's been
14:13caused by the Aurora inhibition.
14:16They will be unable to complete a normal
14:19mitosis and instead will apoptose
14:22are undergo mitotic catastrophe.
14:24So,
14:24um,
14:25it's been recognized that Aurora content
14:27is increased in the face of loss of
14:30P53 and their host of publications,
14:33which demonstrate that increased Aurora
14:35levels are correlated with poor prognosis.
14:38I'll show you some of our work.
14:41This is a panel of cell lines
14:43that that we use in the lab,
14:46all of which have either
14:47mutated or P53 null status,
14:49and you can see that all of them
14:51increase the expression of arorae
14:53relative to either fibroblasts
14:55or normal epithelial tissue.
14:57And when I was at Fox Chase,
15:00we worked on a Aqua essay and insight
15:03to fluorescence assay for Aurora that
15:06could be applied to tissue microarrays.
15:09And so you see here that green is
15:11for carrot and defines where these
15:14head neck cancer nests are within
15:17the tissue core Blues for dampit.
15:20So that will be your nucleus and red is
15:23for Aurora and in this Aurora high cancer.
15:27What you can particularly appreciate
15:29is the high level of expression
15:31of Aurora within the nucleus.
15:33When we looked at nuclear Aurora
15:35in the tissue microarray first
15:37for all cases we saw that high
15:40Aurora expression was associated
15:41with worse survival.
15:43This is also true,
15:44just is a reflection of Natural
15:46History in those patients who had
15:48had no post operative treatments
15:50and never been exposed.
15:52Any DNA damaging agents and we were
15:54able to show that this was entirely
15:57driven by the HPV negative cancers.
15:59So on the basis of this these
16:02data we went to Millennium.
16:04And argued that.
16:05Aurora could potentially be a good
16:08target in head and neck cancer.
16:11They were doing a trial of Al assertive,
16:14which is in Aurora a inhibitor.
16:17Val asserted monotherapy across
16:19a bunch of solid tumors,
16:21and we were able to convince
16:23them to add a head neck cohort,
16:26but this was crushingly disappointing
16:28because the response rate for Aurora
16:31monotherapy in in head and neck
16:33cancer turned out to be about 9% and
16:36given the increasing experimental
16:37evidence that are or inhibition may
16:40always be intrinsically limited.
16:42Limited by this kind of compens
16:44atory cell cycle arrest.
16:46We were then interested in what would be.
16:49The rational combination with
16:52Arorae inhibition that could.
16:55Optimize the targeting of what we
16:57continued to think was likely to be
17:00an important target in this disease,
17:03and so any Mendez and colleagues at the
17:06University of Washington together with
17:08Dell Yarbrough are former colleague
17:10here had undertaken a functional kind,
17:13ohmic screen.
17:14In P53,
17:14mutated head and neck cancer and
17:17actually Aurora came out of that screen.
17:20But another thing that came out
17:22was this mitotic checkpoint kinase
17:24that I just alluded to.
17:26We want and.
17:27People have been interested in
17:29the idea that inhibitors of G1
17:31will abrogate the G2 checkpoint.
17:34You have.
17:34The G1 checkpoint is already
17:36advocated by P53 mutation and that
17:39this might accelerate cell death,
17:41particularly in the presence of DNA
17:43damage such as you might generate with
17:46cisplatin and they showed in animal
17:48models that we one inhibitor MK 1775,
17:51which is now known as the data sorted,
17:54was synergistic with platinum in
17:56P53 mutated head neck cancer models.
17:59Eddie Mendez then took this forward as
18:01a window trial and head neck cancer,
18:04so a small number of patients treated
18:06with a DAB assertive together with
18:08low dose weekly chemotherapy.
18:10And you can see here that the majority
18:13of patients had some diminution in
18:15tumor size and a number of them had
18:19rather major pathologic responses,
18:21and most intriguingly,
18:22you can see that there was evidence
18:24of target engagement,
18:26and so among those patients who
18:28had responses,
18:29there was a decrease in
18:31phosphorylation of CDK.
18:33There was a decrease in
18:36fast focus Stone Age 3.
18:39And potentially you could see
18:42some increase in gamma, H2, ax.
18:45They also were able to correlate
18:48both pathologic and clinical
18:51response with the presence of P53
18:54mutation in the HPV negative cancers
18:58and across the board.
19:00These P53 mutations are
19:03disruptive or deletions.
19:05So we've been exploring whether or
19:07not you can combine Aurora A and we
19:11want inhibition and observe synergy
19:13in P53 mutated head neck cancer,
19:15and you see here a picture of John Wooley,
19:19my colleague in the lamp,
19:21who has done the majority of these
19:24experiments and so you'll see MLN,
19:27that's the Aurora inhibitor,
19:29Azd 1775 that's the wee one
19:31inhibitor and we see synergy.
19:34In terms of cell viability,
19:36soft auger oncosphere formation and
19:39this was present in two separate
19:42HPV negative head neck cancer cell
19:44lines that bear P53 mutations.
19:47Trying to figure out whether or
19:50not our our guests
19:52about the mechanism was correct.
19:55You can see here that when you give the.
19:59Aurora inhibitor there's a dramatic
20:02increase in phosphorylation of CDK one.
20:04This happens in a slightly different
20:07timeline in the two South and
20:09the two different cell lines,
20:11but seems to be a reproducible phenomenon,
20:14and that's abrogated by the addition
20:17of the wee one inhibitor and completely
20:20abolished when you gives it to together.
20:23This results in an increase in the number
20:26of mitotic figures that's abnormal to the.
20:30The. Presence of of really only
20:33single digit normal mitotic figures
20:35in the presence of the combination.
20:38So if you just walk through here,
20:40these are the normal mitotic figures.
20:42When you give the wee one inhibitor,
20:45you get some dis aggregation of
20:47chromatin reflected here in the
20:49fast food Stone Age 3 stain.
20:51When you give the Aurora inhibitor,
20:53you get the formation of these multipolar
20:55spindles three to four spindles,
20:57Purcell, and when you give the
21:00two together you get a, uh.
21:02Abnormal catastrophic mitotic figure.
21:06We also showed in using Annexin 5
21:09flow and looking for cleaved PARP that
21:13there's an increase in a pop ptosis.
21:17And we wanted to compare this
21:20to Aurora B inhibition,
21:21which completely cuts off mitotic entry
21:25by aggregating the phosphorylation
21:27of histone H3 and there was no
21:30synergy between these two agents
21:32and and you can see there.
21:34The lack of fastball, histone H3,
21:38an increase in DNA damage.
21:42Taking this into xenograft models here at
21:45either of two doses of the wee one inhibitor,
21:49the standard dose of the Aurora
21:51inhibitor tumors continued to grow
21:53not too differently from vehicle,
21:56but when we gave the two together,
21:59there was control of tumor growth and
22:02actually statistically significant
22:04improvement in survival for the animals.
22:06Looking at the tumors under the microscope
22:09when we gave the two agents together,
22:12there was a decrease in proliferation
22:14reflected in decreased Ki 67,
22:16there was increased cleaved caspase and
22:19there was a decrease in fast for CDK,
22:22one within tissue and if we did
22:24Aquaphor phospho CDK one and
22:26counted the amount of phosphorus,
22:29IDK one signal in the tumor leading edge.
22:32You can see this was dramatically decreased.
22:35Ellisor to has been a difficult
22:38drug to work within the clinic.
22:40It's associated with Mila suppression,
22:42and there's been a negative
22:44phase three monotherapy trial,
22:45and lymphoma,
22:46and so we were concerned that the development
22:49of that agent might not go forward.
22:52However,
22:52there's been a number of 2nd generation
22:55or or inhibitors that have come
22:57forward and we've had access to a
23:00compound from Taiho called task 119
23:02recently been acquired by Bit Track.
23:05And it's gonna be called Vic 1911
23:07moving forward and once again across a
23:10range of P53 mutated cell lines we see
23:14dramatic synergy for the two agents.
23:16Once again,
23:17we see synergy in xenograft models.
23:20This is confocal microscopy that
23:23again shows you the multipolar
23:27spindle formation with the use of
23:30task 119 is the
23:33Aurora inhibitor.
23:34But with the cells really arresting in
23:37that or becoming quiet sent in that
23:40multipolar spindle state an as they
23:42then attempt to enter mitosis in the
23:45in the presence of both the wee one
23:48inhibitor and the Aurora inhibitor,
23:50developing these very catastrophic
23:51mitotic phenotypes, an notice that
23:53I'm sort of running out of time here,
23:56so I won't March you through this,
23:59but the mechanism looks to be identical here,
24:02as what we saw with assertive.
24:05I'm working with our Columbus
24:07is Lambert at Fox Chase.
24:08We undertook a high throughput
24:10screen to see if we could find
24:13additional partners that would be.
24:15Both hindering and an fostering mitotic
24:18entry again with the attempt to exploit
24:21these multiple regulators of G2,
24:23M and another hit that appeared
24:26very strong was the check one
24:29inhibitor Prexasertib agent.
24:30It's not really moving forward in
24:33the clinic because of its toxicity,
24:36but I wanted to show this just
24:39because with very low dose Ng and
24:42a single dose we saw a profound.
24:45Energetic survival effects that make us
24:47hopeful that with a number of these pairs,
24:49we might be able to go to very
24:52low doses in the clinic.
24:54So test 119 has completed
24:56two clinical trials.
24:57There's a recommended phase two dose.
25:00The toxicity seems to be very manageable
25:03with diarrhea and eye disorders.
25:06Probably the prominent most
25:08prominent side effects,
25:09and so we are moving forward
25:12with a window trial in HPV,
25:15negative head neck cancer that will
25:17have both an initial dose escalation.
25:20Looking at the combination
25:22of Vic and DAB assertive.
25:25And followed by a dose expansion and
25:27that will be part of Project two of
25:29our head next 4 so I wanted to leave
25:32a couple of minutes for questions,
25:35but I didn't want to end without
25:37first of all calling out all of
25:39the fabulous colleagues who were
25:41part of the team that they put
25:44the head and explore in and then
25:46acknowledging all the people whose
25:47work I've just talked about,
25:49particularly John Wooley Jannike Parameshwar
25:51on in Teresa Sandoval Schaefer in the lamp.
25:54So thank you very much.
25:55Barbara,
25:56thank you.
25:56That's fabulous work.
25:58Congratulations on all of it and
26:00and folks can submit questions on
26:03the on the chat box of of the zoom,
26:06but I wanted to ask, you know,
26:09with regard as you look through
26:11the combination of an Aurora
26:13kinase and we want inhibitors,
26:15do you have a sense of 1st what might emerge?
26:19I'm even when you're getting response
26:21because of the complexity of those pathways.
26:24What might emerges mechanism?
26:25Resistance that will occur when
26:28you have dual inhibition of any and
26:30then the second question I have is
26:33what do you anticipate will be the
26:35toxicity profile or the therapeutic
26:36therapeutic window for the combination
26:38clinically. So the we want inhibitors
26:40been quite tolerable 'cause I'm
26:43going to take the second question
26:45first 'cause I've already wrestled
26:46with X and a lot that we want.
26:49Inhibitors been quite tolerable
26:50in the clinic but when it was
26:52combined with PARP inhibition,
26:54diarrhea really became the dose limiting.
26:56Side effect, and so this second
26:59generation Aurora inhibitor did have
27:01about a 25% rate of high grade diarrhea
27:04at the recommended phase two dose.
27:07So the two things that were
27:10sort of hoping is 1.
27:12That will get away with lower doses as we
27:15have in the animal models and second of all,
27:19the diarrhea as it dose limiting
27:21toxicity is one of the easier ones to
27:24manage and so that if we're on top of
27:27this with an Imodium regimen early on,
27:29hopefully that will be helpful in terms
27:32of resistance mechanisms with this
27:34is not something that we've really
27:35gone into with the combination yet,
27:38but is well studied for both of the
27:41agents independently and one of the.
27:43Resistance mechanisms to the Aurora
27:45agents has been a kind of conformational
27:48dependence on the inhibitor,
27:50so inhibitor binds to the activated
27:53form of Aurora A and if you get a
27:56an adaptive process where the cell
27:59just generates more inactive Aurora,
28:02the current generation of inhibitors may
28:05not work as well and there is a group.
28:09Kevan Shokat lab has been developing
28:11novel Aurora inhibitors that maybe.
28:14More able to bind the inactive
28:17confirmation as well. And in terms
28:21of we want inhibitors there there is.
28:28Suggestions that? The the.
28:36Do you need damage effects of?
28:41That we want inhibitors may have an S phase.
28:45Could actually upregulate some checkpoints
28:49that are earlier in the cell cycle.
28:53But it's a good question.
28:55Probably something we should
28:56devote more effort to.
28:57Yeah, well, I'm sure it'll
28:59it'll definitely emerge.
29:01Emerge as you get samples from your trial.
29:04So it's really exciting.
29:06And congratulations so I know we're at 12:31.
29:09Wherever so why don't we
29:11will turn out to people?
29:13Can submit questions to Barbara Online,
29:15but will turn now to our second speaker
29:18and very fortunate to have another
29:21valued member of our faculty speaking.
29:23Doctor Elizabeth Klaus is a professor
29:26of Biostatistics and neurosurgery.
29:28Focused not only on brain tumors
29:30but also the Epidemiology,
29:32most notably the genetic
29:34Epidemiology of these malignancies.
29:35She received her MD and PhD from Yale
29:39and completed her surgery here in
29:41their surgery and through her work
29:44she really has been an international
29:47leader in in the investigation of
29:49the Epidemiology of CNS Malignancy's,
29:52most notably serving as the leader
29:54of the Meningioma consortium,
29:56that meningioma Genome Wide
29:58Association study.
29:59Also,
29:59a leader of the AL Acoustic neuroma study,
30:03and we again were so pleased to have
30:06talented people who bridge the gap
30:08of Epidemiology in biology of cancer
30:11and Elizabeth thank you so much for
30:14sharing your work with us today.
30:17Thanks very much.
30:18Can you see my slides?
30:20OK,
30:20yes great.
30:21So I'm going to
30:22talk a little bit about something
30:24we've been working on an I do want
30:26to note that this is work done in
30:29collaboration with Jeff Townsend's Group
30:30who I think you all know very well.
30:33And Vincent Kenna Tarot as
30:35well as Steven Gaffney.
30:36So despite all the things
30:37that we've attempted to do,
30:39we still don't know much about
30:41risk factors for glioma.
30:42And we wanted to take a look
30:44and see if they were different
30:46methods that we could use.
30:48To see if we could tease out
30:50both environmental and then also
30:52another hot topic is sex specific
30:54signatures of glioma causation.
30:56So you all know that gliomas are the most
31:00common type of malignant brain tumor,
31:02accounting for about 1/3 of all brain tumors,
31:05and the majority of malignant tumors.
31:07But they proved to be very
31:09heterogeneous and we have not done a
31:12great job identifying risk factors,
31:14be they genetic or environmental for glioma.
31:17And so we were interested in doing that,
31:19particularly in light of the poor outcomes
31:22that we see with this group of patients.
31:24So we do know that there are sex specific
31:28differences in glioma risk and outcome and
31:31the plots I have here are for all gliomas,
31:35an then glioblastoma or
31:36sort of an IDH positive.
31:38Excuse me,
31:39IDH negative tumor and then
31:42lower grade gliomas.
31:43The males being the blue,
31:45the females being the red.
31:47And it's interesting in that we
31:49see this sex specific difference
31:51across the entire age range,
31:53so it's a little bit different
31:55than we see with,
31:57for example,
31:57meningiomas where we see the
31:59women having greater risk,
32:01but the risk difference decreasing once
32:03women passed through the menopause.
32:05Whereas here we see the sex
32:07differences for glioma across the
32:09age spectrum and across all subtypes,
32:11and so that obviously suggests
32:13that other mechanisms in addition
32:15to a good sex hormones must be.
32:17Are behind the difference men are
32:19at greater risk of being diagnosed
32:21with the disease.
32:22And again,
32:23that's across pretty much all
32:25the subtypes and they also have
32:27lower survival in general then
32:29for females across all subtypes.
32:31So we've looked at this a little
32:33bit and I've been lucky enough
32:35to collaborate with a group of
32:38individuals called the glioma.
32:40International Case Controls Consortium,
32:42and that's led by Melissa Bondy,
32:44initially at MD Anderson, then it Baylor.
32:47Now she heads up the Epidemiology
32:49section at Stanford,
32:51but we were able to gather over
32:5310,000 cases and 10,000 controls,
32:55and so these are essentially looking
32:58at constitutional or germline
33:00risk alleles by sex.
33:01So if I can draw your attention
33:04to the table over here,
33:05these are the variants that we
33:07found to be significantly different
33:09males versus females.
33:10Males being the blue,
33:12an females being the red.
33:13So we did certainly find differences
33:15at the germline level,
33:17but we were also interested in
33:19looking at things at the tumor
33:21or their cinematic level.
33:220 Sex is a biologic variables I mentioned.
33:25This is a very hot topic.
33:27Now we obviously know there are biologic
33:30differences between males and females.
33:31There's also some thought as to
33:33whether there's variation in
33:35the prevalence of risk factors,
33:36and then also whether there's
33:38a difference in sort of a gene
33:40by environmental interaction.
33:41So, for example,
33:42and this has long been postulated,
33:44but it's really been pretty difficult
33:46to prove that males in particular are
33:49more likely to be exposed to work like
33:51toxins that might be associated with risk,
33:54and so that was one of the things
33:57we wanted to look at as well.
33:59And in part, why we divided our analysis.
34:02By sex.
34:03So there's two goals and what
34:05I'm going to talk about today.
34:08We wanted to look at the relative
34:10contribution and this is based on
34:12some of the work that I know you've
34:15already appreciated with Jeff Townsend,
34:18but we're applying it specifically
34:19now to glioma,
34:21but looking at the relative
34:22contribution of cancer cell lineages,
34:24proliferation and survival of
34:26single nucleotide mutations,
34:27and we divided our study subjects
34:29up by IDH mutation.
34:31And, as most of you know,
34:33IDH mutation is one of the
34:35key dividers into the.
34:37The higher in the lower grade tumors,
34:39and certainly a prognostic factor,
34:41as well as a factor in response to treatment.
34:44We also wanted to quantify,
34:45and this is something that is a
34:48little bit new to Epidemiology
34:49in terms of how we've tried to
34:52identify risk exposures.
34:53Typically we've done things like
34:55large case control studies where
34:57we look at large numbers of people
34:59that have the disease,
35:00compare them to large numbers
35:02of people without the disease,
35:04and look at things like questionnaire
35:06or work pic.
35:07Exposure and see if we can figure out
35:10differences between the cases and controls.
35:12So what we're doing now,
35:14and this is sort of an emerging
35:16field in cancer Epidemiology,
35:17is to look at the cosmic cancer
35:20mutational signatures in tumors and
35:22see if we can then backtrack match
35:24it to possible risk exposures,
35:26and one of the things we're hoping
35:28to do in the future is to go back to
35:31our cohorts and studies for which
35:33we collected good occupational data
35:35and see if we can match it up to.
35:39Mutational signatures So the methods
35:42I'll talk a little bit about.
35:44I am highlighting here.
35:46Jeff's paper that he had in J&CI
35:48two years ago,
35:49and I think you've seen some of
35:51these sorts of methods applied,
35:53in particular to actually head
35:55and neck cancer.
35:56So the Cancer Genome Atlas,
35:59and others, including the glioma
36:01Longitudinal Analysis Consortium,
36:03or Glass, which is led by roll.
36:06Their Hokage,
36:07Jackson Labs and which I'm also a member of.
36:11So these groups have identified
36:14the most common genetic changes in
36:17primary glioma tumors including TP 53,
36:20IDH,
36:20EGFR with the relative importance
36:22of these mutations and how they
36:25relate to tumorigenesis.
36:27Is not well known,
36:28so one of the things that we've
36:31been working on,
36:32and Jeff has been a leader in
36:35is defining this
36:36cancer affect size.
36:37So this metric of the relative
36:40overabundance of variance due to
36:42their contributions to survival,
36:43indivision versus what you're
36:45actually seeing in the tumor.
36:47So we're quantifying the cancer affect size.
36:50We're using single nucleotide mutations,
36:52and then we basically do a scaled
36:55selection coefficient for the for the
36:57different variants we look at it.
36:59By sex and by IDH subtype.
37:01And so we're trying to get a feel for
37:04whether this would help us explain
37:06any differences in the glioma,
37:08risk and outcome that we see by sex.
37:11And then we're going to move on to
37:14the cosmic mutations so I won't go
37:17into the gory statistical detail.
37:19This is drawn from Jeffs paper,
37:21but basically you're comparing
37:22expected to observed so expected
37:24number of synonymous mutations,
37:25and then we're looking at the
37:27rate at which the mutations.
37:29Actually occur.
37:30The data that we're using here,
37:33our whole exome sequencing data from a pretty
37:36good size data set in terms of glioma,
37:38so about 1100 and these are
37:40all adult patients.
37:41There's no pediatric patients
37:42in here and we drew it from the
37:45Cancer Genome Atlas study.
37:46And as I mentioned,
37:47I know some of you may be aware of
37:50what glasses, so it's an effort.
37:52As I mentioned led by role Verhaag,
37:54but which yell is also a member of
37:57looking at not only the initial tumors,
37:59but the humours overtime.
38:01So how do they change?
38:02In terms of their genetic makeup,
38:04when we do nothing to them when we
38:07do chemotherapy or we do radiation,
38:08or a combination of all the above and
38:11what changes do we see and what do
38:14we learn from that in terms of what
38:16we should or should not be doing?
38:18And then we also used a lot of data.
38:21All of this is readily available
38:23off the Internet,
38:24but we use tissue specific mutational
38:26covariance and this helped.
38:27Just figure out what sort of mutation
38:29rate calculations we should use.
38:31Gave us a little bit of information
38:33about replication timing.
38:34And some of the other datasets
38:36that are listed here.
38:37So here's some of the results.
38:39Just to take you through it a little bit.
38:42So I have a divided by tumor type
38:45and it's by seksan by mutation.
38:47So the wild type tumors who would
38:49be considered the higher grade are
38:52primarily the glioblastoma tumors are
38:54in the first 2 rows and the IDH mutant,
38:56which would more typically
38:58be the lower grade tumors.
39:00And then I have males versus females,
39:02males versus females,
39:03and then there's sort of a
39:05cancer affect size here.
39:07The blue is non coding region.
39:09And the red is coding so you
39:10can see the patterns are quite
39:12different for what might be called
39:15the low and the high grade,
39:17the IDH mutant tumors had few
39:19unique recurrent substitutions.
39:20All of them were in coding regions,
39:23whereas the wild type tumors,
39:24and obviously this is in part what
39:26makes them so hard to manage is
39:29they exhibited many substitutions,
39:30but they were primarily
39:32in non coding regions.
39:34So here's another picture.
39:35A little busy but divided once again,
39:38the IDH mutant or the lower grade
39:41tumors are presented first.
39:43The wild types are second,
39:45and there's female male, female, male,
39:47and So what we're looking at here is that.
39:51Items that top the list
39:53are the most important.
39:55The size of the circle that is attached
39:57to them measures the prevalence
39:59so there can be kind of this.
40:02Disconnect as to what is important
40:05and how frequently it occurs so
40:07we can see that in the low grades
40:10it's pretty much as expected.
40:12Previously reported mutations
40:13in IDH one and two TP 53.
40:15Some of the other classics were confirmed,
40:18but what's interesting is
40:20if we go here to the IDH.
40:22Wild type tumors the most important
40:25with respect to cancer affect
40:27gene is this low prevalence right?
40:29You can see that the circle
40:31that matches up to it is small,
40:34not large like we see for IDH.
40:37Is this B RAF V 600 E so we
40:40know that it's important.
40:42It turns out that it looks
40:44like it's the most important,
40:46but obviously it doesn't occur
40:48that frequently but interesting.
40:50What drives some of these gliomas here?
40:53The other thing we looked at is do
40:55males and females show the same pattern
40:57of what significantly overburdened,
40:59and there were a lot of similarities
41:02the way that we have this broken
41:04up here is each panel is a gene.
41:07The mutants come first in each
41:09panel and then within each panel
41:12we've got the females in the mails.
41:15So we did see some differences,
41:17although overall most the things the
41:19males and females showed were similar,
41:21but we did see differences in the P3K
41:24pathway, so an IDH mutant, tumors the PK.
41:27Three CA mutations were located in
41:30the helical domain for females,
41:32and the kinase domain for the males,
41:34and so that's appear.
41:36This panel here.
41:38OK,
41:38so it's the mutant and non mutant
41:40and then the variance of import
41:43also differed by sex for PK3R1
41:45and so that's interesting in part
41:48because we know that the way in
41:50which these areas are targeted by
41:52various chemotherapies does differ.
41:54We looked in the literature an we
41:57don't see too much reported honest.
41:59We did find a paper by Dan Cahill
42:02at all at mass general and although
42:05they didn't report it as such,
42:08they found something similar
42:09where the females tended to have.
42:11Variations in the he local domain and
42:15the males had them in the kinase domain,
42:19and so as I said,
42:21although both domains are
42:23involved with glioma Genesis,
42:25there is differential amounts of
42:28potentiated by these two regions.
42:30And obviously there's different
42:32sensitivity to various treatment
42:35types depending upon domain.
42:37So back to environmental exposure.
42:39We have searched and not just our
42:41group of many groups have searched
42:43long and hard for environmental and
42:46genetic risk factors for glioma.
42:48In terms of genetic risk factors,
42:50we have found small numbers of families
42:53with high risk but typically that does
42:55not relate to the general population
42:58and so no genetic risk factors
43:00really explain a large proportion of
43:02inherited risk and other than high dose
43:05radiation to which not many people.
43:07Thankfully are exposed.
43:08We really haven't found much of an
43:11Association between environmental
43:13risk factors in glioma risk.
43:15There has been reported a fairly
43:18consistent but low effect,
43:20an inverse Association with
43:22history of allergy.
43:23So the question comes,
43:25why haven't we found anything?
43:27Is it that there is no Association?
43:29Or is it basically statistical power
43:31that there's so few cases of glioma
43:34relative to other things we've looked at?
43:36For example,
43:37I started my work with breast
43:39cancer and even just using the
43:42state of Connecticut as a base,
43:44you would have enough cases
43:45for a large study for glioma.
43:48That is not true and also likely a
43:50lot of the exposures that we think
43:53are causing risk are themselves rare.
43:56So one of the things that people
43:58have been thinking about doing,
44:00is there another way to do this now?
44:03So now that we have these mutational
44:06signatures that the are listed in
44:09the Catalogue of Somatic mutations
44:11and cancer or cosmic can use that
44:13as a way to match up to exposure,
44:16particularly if you have previously
44:18obtained environmental or other
44:20exposure history in the patients.
44:21So we did that here with the
44:241100 cases that we mentioned,
44:26we group Jack Sonic.
44:27SNV and tried to match him
44:30up to what is in cosmic,
44:32and so you know that the cosmic
44:34catalog is rapidly changing.
44:36New things are always being added,
44:38but we looked at what existed at this point,
44:41and obviously they have previously found
44:43a match up over environmental exposure to
44:46signatures not only in head and neck cancer,
44:49but smoking and lung cancer,
44:50UV exposure,
44:51and so we looked at that for glioma.
44:55And so again,
44:56here's our slide here again,
44:58broken into IDH Mutant,
45:00which is the top row IDH,
45:03Wildtype bottom row and then females
45:06or first column mails or second
45:09column with each of these bar charts
45:12relates to is the proportion of our
45:15cases for whom the majority seem to be
45:18associated with a certain signature.
45:21And the overall news is a little
45:24bit depressing in the sense that
45:26the primary molecular signature
45:28identified was age related mutagenesis.
45:30Basically the older you get,
45:32the more at risk you are,
45:35but we did find one thing that
45:37was quite interesting,
45:39particularly in light of their such
45:41a positive risk factors identified
45:43for glioma and that was occupational
45:45exposure to something called Halo alkanes.
45:48Pretty much true across whether
45:50you are male or female.
45:52And whether you were IDH,
45:54mutant or not,
45:55we did find a little greater rate
45:58of the signature showing up in
46:00the males versus the females.
46:02But we certainly saw them in both
46:04and then we also saw which we
46:07haven't quite figured out how to
46:09explain yet. These UV light signatures an
46:12it's interesting because glioma has been
46:15associated in the number of instances
46:17with Melanoma and also with the B RAF.
46:20So we're trying to sort out whether that.
46:23Has anything to do with why we're
46:25just seeing some of those signatures?
46:27So hello, alkanes are basically used for
46:30many industrial and day-to-day purposes.
46:33Of interest there seen in refrigerants,
46:36fire extinguishers, flame retardants,
46:37and we thought this was very interesting
46:41because there's always sort of been
46:43this theory that in some of these.
46:46Occupations including for firemen
46:47and that has been reported that
46:50there is an increased risk of glioma,
46:52and so, whether or not this ties
46:55things together or not is unclear,
46:58so the signature was basically developed
47:00by looking at cholangio carcinoma in
47:02a group of workers that were exposed,
47:05known, exposed to hello Alkins in
47:07Japan and so essentially they had
47:10111 workers that were exposed.
47:1217 developed.
47:13Would you all know to be a pretty rare?
47:16Cancer,
47:16so it was quite unusual that this number
47:19of individuals was diagnosed with it.
47:22They all were working in printing
47:24companies and they all were known
47:26to have occupational exposure and so
47:28essentially what they did was they took
47:31the tumors from these individuals,
47:33looked at the molecular.
47:34Pattern and developed this signature.
47:37So that's essentially how that the
47:39signature was initially identified,
47:41and so that's what we're seeing.
47:44Basically in our data.
47:46So conclusions here that the majority
47:48of cancer causing mutations in these
47:51gliomas we're seeing primarily as
47:53a consequence of endogenous,
47:55rather than actual, exogenous exposures.
47:57We did think was interesting that
48:00different domains of jeans in the P3K
48:03pathway were different for males and females.
48:06For those of us that have searched
48:08long and hard for some of these
48:11risk factors for glioma,
48:13we are excited that at least potentially,
48:15there's a new means to try
48:17and identify even if rare,
48:19these environmental risk factors
48:21and it's sort of a whole new aspect
48:24of glioma that were looking at so
48:26some of our future directions.
48:28We're looking now to partner with
48:30colleagues who have worked with us
48:33both in the meningioma consortia man,
48:35the Glioma international Case
48:36Control Consortium.
48:37And we also in our international
48:39low grade glioma registry.
48:41Looking at cohorts in which we
48:43have a good occupational history.
48:46So the San Francisco Bay Area
48:48Adult Glioma study,
48:49which is led by Margaret Wrench
48:52and John Winky,
48:53they collected extremely detailed
48:55occupational history for their cohort,
48:57and they have all the tumors.
48:59So we're going to try and go back and
49:02Geno type those tumors and see if
49:05we can confirm these associations,
49:08which they found with firefighters.
49:10And glioma.
49:11And also they found it with painters as well.
49:14And so we are also collecting glioma
49:17patients with occupational histories
49:19and just sort of throwing it out to people.
49:22If you're aware of any firefighters
49:24or similar occupied individuals with
49:26glioma would love to try and get
49:29a cohort
49:29together. The other thing that was
49:32just sort of luck this past semester.
49:34So I teach over at the school,
49:37public health and everything has been remote.
49:40And so as I was meeting via zoom with
49:43one of my students for her final project,
49:46she revealed that she was actually
49:49the principle project director for
49:51the Firefighters Cancer Cohort study.
49:53So we're also hoping to parano.
49:55NIH is a big directive to try and
49:57look further at environmental
49:59exposures and cancer,
50:01so we're hoping that we can partner with
50:03some of these folks to look at individuals,
50:07either living or dead that
50:09may have undiagnosed.
50:10With glioma that we now have this exposure.
50:13So thank you all for your time.
50:16I wanted to also thank Jeff Townsend,
50:18Vinston, Canna Terra,
50:19who was a postdoc in Jeffs lab but now as an
50:23assistant professor of biology up the road,
50:26a little bit of Emmanuel
50:27College and I have to thank him.
50:30He made all the beautiful
50:32pictures an Steven Gaffney,
50:33who also works in Jeff Slab.
50:35Thank the various.
50:37Brain tumor associations,
50:38including the ABCA in the
50:40NBTS as well as Luglio Anna,
50:42a Dutch group called Stop Brain
50:44Tumor for their support and then
50:46also thank you for Doctor Rolled
50:49their hacking the Glass consortium
50:50who allowed us access to the data.
50:53So happy to take any questions.
50:57Elizabeth, thank you.
50:58That was a terrific summary of your
51:01work and obviously will open it
51:03up to questions on the chat line.
51:06But let me ask.
51:07I found it interesting the the
51:09observation I guess from Asia
51:11about the Association of Halo
51:13alkanes with cholangiocarcinoma.
51:15As you may know,
51:17there's there's a biologic
51:18difference between intrahepatic,
51:20an extra paddock cholangio,
51:22where extra panic actually gave
51:25IDH mutations but insured don't
51:27wear the cases that they found
51:30in Asia with a extra paddock.
51:32You
51:33know, I don't know the answer to that.
51:35I gave a similar talk at UCSF
51:37and they mentioned this as well,
51:39so we're trying to gain access
51:40to some of that information,
51:42but I don't know at present.
51:45And then. With regard to the finding of
51:48the potential differential in mutations
51:50within pick three CA by gender by sex,
51:53is there an understanding of
51:55why those two domains would be
51:56different between men and women?
51:58No, and you know,
51:59we started to look at that a
52:02little bit and we collaborate
52:03a bit with Dan Cahill.
52:05As I mentioned up at mass
52:07general so we don't know yet,
52:09but he's going to try to take a look
52:11into that he he presented the data
52:13but didn't note the differences,
52:15so he's going to try to take
52:17a look and see what that.
52:19Might entail.
52:21And then my last question,
52:23and this is gonna show my naivete
52:25and understanding brain tumors.
52:26But instead of the Natural
52:28History of the low grades.
52:30Is there an evolution of the
52:33semantic events such that they
52:36look more like high grades?
52:38So it depends.
52:39They generally remain quite different.
52:41The IDH mutation stays constant
52:43throughout and so that's sort of
52:45been one of the issues is what
52:47you show up to the party with
52:49tends to be what you stay within.
52:51That makes it a little bit
52:53different to manage them.
52:55We didn't found find in some of
52:57the glass consortium work that
52:58we've looked at that really things
53:00changed that much whether you
53:02gave them treatment or whether
53:04you didn't give them treatment.
53:05Is a little bit disheartening,
53:07but we're going to try to look a
53:10little bit further at that.
53:11Yeah, yeah, you know,
53:13judging by the way you describe
53:15for the presence of Halo alkanes,
53:17you could imagine they
53:19may be more ubiquitous.
53:21In our environment than we might
53:22otherwise appreciate given
53:23all the things they are in
53:25absolutely and it doesn't
53:26have to just relate to glioma.
53:28You know could relate to
53:29lots of different things so.
53:32Well, very interesting.
53:33You know we're just about out of time.
53:36An really appreciate.
53:37Oh actually, JoJo contest
53:39hasn't question, forgive me.
53:41So Joe's question is high dose
53:43radiation therapy delivered to
53:45pediatric patients can lead to glioma?
53:47Have you found evidence that medical
53:50imaging and radiation exposure in
53:52this setting is associated? So
53:54there's actually, and you probably
53:56even know of these two studies.
53:58There's a big cohort from Australia as
54:01well as a second cohort from England,
54:04and they did find that even exposure to
54:07head CT's at an early age in children
54:10was associated with the I mean.
54:12It's a very small increase in risk,
54:15but a definite increase in risk
54:17of both glioma and meningioma,
54:19and then anything we looked at
54:21we did find it was a fairly hotly
54:24contested topic we did find.
54:26And exposure to bite wings was associated
54:29with an increased risk of meningioma,
54:31but that's sort of exposure
54:33level in terms of dental X,
54:35Rays generally doesn't exist now.
54:37But yeah, in terms of head CT's,
54:39the two big cohorts from Australia,
54:41anyone to do suggest that,
54:43although you know,
54:44even though the risk is increased,
54:46the absolute numbers
54:47are small. And then Antonio
54:49Murray asks, is great talk.
54:51Have you looked at thyroid hormones,
54:53thyroid disease and differences
54:55between men and women?
54:57So we haven't but one thing that
55:00is very interesting, and it relates
55:02a little bit more to meningioma.
55:05Is a gene that we found,
55:07and this is a constitutional
55:10gene on chromosome 10.
55:11We've found to be associated with meningioma,
55:14breast, ovarian and also now
55:16thyroid tumors. Interest.
55:20Elizabeth, thank you.
55:21We are at the top of the hour.
55:24Appreciate both your talk and
55:25Barbara's really extending work.
55:26Thank you for sharing all of it with
55:29us and to everyone who joins us today.
55:31Thank you for taking the time
55:34to join grand rounds and we'll
55:36see you all again next week.
55:38Have a good day.