Identifying and Targeting DNA Repair Defects in Cancer

October 02, 2024

Yale Cancer Center Grand Rounds | October 1, 2024

Presented by: Dr. Shridar Ganesan

Information

ID: 12158
To Cite: DCA Citation Guide

Download Transcript

00:00A real pleasure to,
00:02have our speaker, but be
00:04we're here, and I will
00:06introduce him in just a
00:07second. I just wanna remind
00:08everyone,
00:10to fill out your
00:12DEI climate survey.
00:15They would very much appreciate
00:17it as would I,
00:19and so,
00:20please do it.
00:23So,
00:24I stood up and I
00:25said, you, I do without
00:27notes because I've known Sridhar
00:30Ganesan
00:31for a long time.
00:33And Sridhar has,
00:36over the course of his
00:37career, taken a trip up
00:39and down the East Coast.
00:41He He was an undergraduate
00:42at Princeton, a Yale, MD
00:44PhD student,
00:46and we actually overlapped here
00:48a little bit way back
00:49in some time in the
00:51past.
00:52And then,
00:54was a resident at the
00:55Brigham and a fellow at
00:57Dana Farber and stayed on
00:58there
00:59and worked for quite a
01:01number of years in the
01:02lab of the late David
01:04Livingstone,
01:05late and great David Livingstone.
01:08And Sridhar is now professor
01:10of medicine at Rutgers,
01:13has assorted titles which are
01:15up on the screen,
01:17and has focused his career
01:20on
01:22DNA damage
01:23repair
01:25and related topics.
01:28Back
01:29now probably
01:31close to twenty years ago
01:33when we were both at
01:35Dana Farber,
01:36I pulled Schreuder out of
01:38the lab.
01:39He was totally focused in
01:41the lab at the time,
01:43and he and two other
01:44people came to the clinic
01:46half a day a week.
01:48And I said, this is
01:49the way I'm mixing up
01:50the clinical people and the
01:52lab people, and I think
01:53it was a pretty successful
01:54experiment. I said, I don't
01:56care how many patients you
01:57see. Just show up.
02:00And to this day,
02:02Shreedhar still sees patients with
02:04breast cancer,
02:05and is interested in both
02:06breast cancer and
02:08rare tumors as well,
02:11which he may or may
02:12not comment about. And,
02:13today,
02:14his topic is identifying and
02:16targeting DNA repair mutations in
02:18cancer.
02:19Shreedhar, it is a pleasure
02:21to have you here. We
02:22actually have
02:24a plaque for you,
02:27which,
02:28you can,
02:30oh, and Roy's gonna take
02:31a picture of of the
02:32two of us with the
02:33plaque.
02:36Gotta have a butter here.
02:40Expect to see it hanging
02:41in your office next time
02:42we visit. Awesome.
02:44There you go. And I'll
02:45keep this for you, Ricardo.
02:47Thanks for being here. Alright.
02:49Thanks.
02:50That's a very kind introduction.
02:51Thanks so much. It's such
02:51a pleasure to be back
02:52here at Yale. I did
02:53spend eight years in lovely
02:54New Haven, even though that
02:55time I had long hair
02:56in a ponytail.
02:58Actually, that stayed actually doing
02:59fellowship and when I first
03:01met, Eric as well. And
03:02I am incredibly indebted to,
03:04both Eric and, Judy Garber
03:06for really bringing me into
03:07the clinic and, and and,
03:12have started my career in
03:13breast taking care of patients
03:14with breast cancer, which has
03:14been, you know, incredibly rewarding.
03:17So today I'm gonna talk
03:18to you it's a little
03:19bit of history of kind
03:20of DNA repair defects in
03:21cancer, some stuff that I
03:22did originally in in in
03:23David's lab and some more
03:25work right now that's we
03:26hope is a little more,
03:27clinically acable as well as
03:29some of the underlying basic
03:30science. And the idea is,
03:31you know, I have to
03:32tell this,
03:34audience
03:36alright. Ah, so these are
03:37my disclosures. So I do
03:39a lot of research. My
03:39wife is an employee of
03:41Merck.
03:42And, you know, I just
03:42wanna stop by. Yes. A
03:43lot of my work is
03:44done, with David Livingstone. This
03:46is David Livingstone in his
03:47natural habitat, which is surrounded
03:49by his former trainees.
03:50So there's,
03:51you know,
03:52Sharon Kanter who cloned,
03:54BRIP one or Fang j,
03:56Bing Shah who, cloned and
03:58characterized PALB two in the
03:59lab, and Lonnie Drapkin who,
04:01you know,
04:02contributed to identification of the
04:04cell of origin of high
04:05grade serous ovarian cancer,
04:07in his lab. And, David
04:09passed away unexpectedly a couple
04:10years ago, and the Farber
04:12is a,
04:13much quieter and less interesting
04:14place, I think, with his,
04:17with his passing. But I'm
04:18greatly, you know, a lot
04:19of what I work was
04:20really, he grabbed me when
04:21I was a senior resident.
04:23And,
04:24I had no idea who
04:25he was. And within five
04:26minutes of meeting him, somehow
04:27I'd agreed to work in
04:28his lap.
04:30Alright.
04:31And so, you know, obviously,
04:33you know, talk a little
04:33bit of background of BRCA
04:35and BRCA two. If you
04:35kind of look at families
04:37that have,
04:38high I don't know if
04:38there's a point here, who
04:39have, high risk disease.
04:41If you look at families
04:42in which there's multiple,
04:45members that have a breast
04:46cancer, a small proportion of
04:48them are due to Mendelian,
04:49you know, high penetrance,
04:51Mendelian genetics. So in those
04:53families that have high penetrance,
04:55about fifteen percent are associated
04:56with germline mutations in BRCA
04:58and BRCA two. There's a
04:59small percentage of others.
05:01PALB two is probably the
05:02most predominant one. We're talking
05:03about it's probably the closest
05:04thing we have to BRCA
05:05three in terms of being
05:05a high penetrance,
05:07gene. And then there are
05:08SNPs and other protected SNPs
05:10perhaps
05:11contributing from many of them
05:12and then unexplained or you
05:14know, breast cancer is common,
05:15so there can be clustering
05:16by chance in a common
05:17disease, that also can be,
05:19seen in a bunch of
05:20these. And for a for
05:22a while, you know, the
05:23I'm gonna talk a little
05:24bit about the the the
05:25BRCA one, cancer syndrome.
05:28Okay. You know, so women
05:30carrying a,
05:32mutant allele BRCA one have
05:33a greatly increased lifetime risk
05:34of breast and ovarian cancer.
05:36Actually, the relative risk is
05:37much greater for what we
05:38used to call ovarian cancer
05:39than breast cancer. Alright?
05:42Because breast cancer because ovarian
05:43cancer or a fallopian tube
05:44cancer is rare. The tumors
05:46arises when we all have
05:47undergone
05:48lost heterozygosity. We've lost the
05:50wild type,
05:51allele, implying the BRCA one,
05:52and we'll talk about this,
05:53really functions as a classical
05:54tumor suppressor. And men with
05:56a interestingly, with the BRCA
05:57one mutation have little or
05:58no phenotype.
06:00That's discernible.
06:01Right?
06:02And the cancers that arise,
06:04you know,
06:05tend to have early onset.
06:06So this is kind of
06:07the distribution of age of
06:08onset. The bottom line here
06:09is kind of the incidence
06:10in the general population over
06:12time of breast cancer. And
06:13up here is BRCA one.
06:14You see the instance starts
06:15to rise in the thirties
06:17and really peaks in the
06:18premotor puzzle time. And then
06:19after about age fifty, you
06:21can see these these curves
06:22are actually parallel. So then,
06:23you know, the bulk of
06:24the risk is really here.
06:26Alright? This is BRCA two
06:27and PALB two. So BRCA
06:29one BRCA two and PALB
06:30two have a slightly different
06:31kind of age of onset.
06:32It it and the risk,
06:34occurs a little later and
06:35continues to increase, and PALB
06:37two is very similar to
06:38BRCA two. Alright?
06:40And these are high risk.
06:41These are much less penetrant,
06:42and we won't talk about
06:43them all, not that much
06:44today.
06:45Alright? And so when BRCA
06:47and BRCA two were first
06:48cloned, everyone assumed
06:50that because it was associated
06:51with breast and ovarian cancer,
06:52they would have some sort
06:53of lineage specific role. That
06:55it was somewhat to be
06:56something with the biology of
06:58the breast or the ovary.
07:00And, when I was a
07:01senior resident, we made the
07:02first antibodies against BRCA one
07:04that actually were worked and
07:05were validated. And we found
07:06to our surprise back then
07:08in David's lab that BRCA
07:10one instead was expressed was
07:11not lineage restricted, but cell
07:13cycle restricted. Any cell that
07:15passed the g one transition
07:17expressed BRCA one in these
07:18beautiful little nuclear foci.
07:20Alright? So it was cell
07:22cycle restricted, not lineage restricted.
07:24And for tumors to purposes,
07:26it was kind of unusual
07:27because you took out BRCA
07:28one, the tumor didn't grow
07:29bet cells didn't grow better.
07:30They all crashed. And, partly
07:33PFT dependent arrest within a
07:34few cell cycles, we took
07:35out BRCA
07:36one. Alright? So it was
07:38cell cycle dependent, and without
07:40it, cells just crashed. Alright?
07:43At that time, BRCA two
07:44was also cloned and identified.
07:46And if you remember, BRCA
07:47and BRCA two were identified
07:48by classical genetics by Mary
07:50Claire King. Alright? And, when
07:53ultimately BRCA two is cloned,
07:55I always say BRCA one
07:56and BRCA two are, you
07:57know, eighty percent identical by
07:59name,
08:00but zero percent identical by,
08:02you know,
08:03sequence or anything else. They're
08:04completely different proteins.
08:06Alright? And with little homology
08:08to any other known proteins.
08:09And it's another large gene.
08:10And interestingly,
08:12despite the fact that they're
08:13completely dissimilar proteins,
08:15some of the characters of
08:16BRCA also, these are also
08:17not lineage restricted,
08:19but came up during s
08:20phase in beautiful nuclear foci
08:23that colocalized with BRCA one.
08:25This is work by,
08:28Daniel Silver and Junji Chen
08:30in David Slab.
08:32Alright? So, and people all
08:34talk about Bracker, you know,
08:35the BRCA1, but they're actually
08:36very complete different genes. They
08:37seem to they're just in
08:38the same pathway. It was
08:39not necessarily obvious that they
08:40should they would.
08:41Right?
08:43And even though a lot
08:43of work has gone into
08:44what the fundamental problem function
08:46in which BRCA one, works
08:48in, so there's lots of
08:49data on transcriptional regulation, genomic
08:51regulation,
08:52epigenetic regulation.
08:54The fundamental thing is that
08:55without BRCA one, if you
08:57have a cell that goes
08:58without BRCA one through a
08:59couple of s phases, the
09:00chromosomes fall apart.
09:02Alright? So these are mouse
09:03chromosomes in which we've taken
09:04out BRCA one, and this
09:06is after a few cell
09:07cycles. The
09:09normal chromosomes in mice should
09:10look like these acrocentric. Alright?
09:12But you take them out
09:12also and the chromosomes fall
09:14apart. You see quadradials,
09:16chromatin deletions.
09:17Alright?
09:18So BRCA one is
09:21incredibly important for normal s
09:22phase,
09:25maintenance. Alright?
09:28Other thing is that, you
09:29know, what we realized from
09:30this is that BRCA one
09:31is not as what I
09:32consider an accidental tumor suppressor.
09:35Alright? Its main function is
09:36an s phase to keep
09:37the chromosomes alive. A rare
09:39consequence of its dysfunction
09:41is cancer. If you look
09:42at women,
09:43with a BRCA one mutation,
09:44they have about a fifty
09:46to sixty percent lifetime risk.
09:47So imagine that in their
09:48in that,
09:49a breast cancer. Imagine that
09:51all trillion of her cells
09:52are heterozygous.
09:53There's only a fifty to
09:54sixty percent chance that one
09:55of those trillions of cells
09:56will acquire the changes necessary
09:58to get cancer. Right? So
10:00this is not p fifty
10:01three. You get multiple cancers
10:02early in life. This is
10:03not APC, where by the
10:04time you're twenty or thirty,
10:05your colon is filled with
10:06premillion polyps. This is not
10:08NF one.
10:09Alright? This is a
10:12per cell basis, very rare.
10:13Obviously, with a trillion cells,
10:14it's, it's inevitable.
10:16So always consider BRCA one
10:18a rare consequence of this
10:19dysfunction is cancer. Right? I
10:21tell my students actually, it's
10:22like, you know, having a
10:23shoelace. If you have your
10:24shoelace is untied, the chance
10:25of it catching in a
10:26bicycle gear is greatly increased.
10:29But the shoelace is not
10:30a bicycle chain, you know,
10:31catcher suppressor.
10:33Alright? It is a rare
10:34consequence of its dysfunction, you
10:36know, unlike many other classical
10:37tumors. That's the way I
10:38think about it.
10:41So, early on, some insight
10:43to that role, of BRCA1
10:46and DNA repair was actually
10:47found in it was I
10:48was lucky enough to work
10:48with Ralph Scully early on,
10:49again, as a resident and
10:50fellow. We found that, hey.
10:52One of the, that BRCA1
10:53is really involved in esophage,
10:56in part,
10:57doing a DNA repair and
10:58specific DNA double strand break
10:59repair. We found that BRCA
11:01mutant cells are very sensitive
11:02to radiation therapy. And if
11:04we look at the dynamics
11:05of repair using old technique
11:06called pulse field gel ultra
11:09pulse field gel electrophoresis,
11:11alright, in which we look
11:12at the dynamics of how
11:13DNA breaks are repaired. Normal
11:14cells have a biphasic repair.
11:16You get eighty percent of
11:17the DNA breaks are resolved
11:18in the first two to
11:19three hours, and then the
11:20rest of the of the
11:21breaks are resolved in, the
11:23next twenty four hours. It's
11:24kind of biphasic annealing of
11:26DNA double strand breaks. In
11:28BRCA mutant cells, we noticed
11:29early on that the original
11:30phase of repair looked fine,
11:32but it lagged
11:34afterwards.
11:34Alright? You had unresolved breaks.
11:36And this gave us an
11:37idea that this was involved
11:38in fundamental aspect of DNA
11:40doubles
11:41that without BRCA one, you
11:42get unrepaired DNA double strand
11:44breaks during s phase.
11:45Alright?
11:47Work by a variety of
11:48laboratories, especially Maria Jason, further
11:50refined the role of BRCA
11:51one to what we understand
11:52today, our current story, which
11:54is that BRCA one's involved
11:55in homology mediated repair.
11:57And to step back, imagine
11:59that when you have a
11:59DNA double strand break in
12:01a cell, you have two
12:02problems a cell has to
12:03face. One is the DNA
12:04strand is broken. You have
12:06physical integrity of the of
12:07the chromosome is now disrupted.
12:09The second is informational loss.
12:12Alright? So often DNA double
12:14strand breaks,
12:15or often single strand breaks.
12:16When you have a single
12:17strand break, you have little
12:18resection at the end, and
12:19you fill back in. DNA
12:20double strand breaks are often
12:21two single strand breaks that
12:22are near each other. You
12:23resect both areas and all
12:24of a sudden, oops, you
12:25have a DNA double strand
12:26break. Alright?
12:28And there, you've lost information.
12:29The nucleotide sequence there is
12:31gone.
12:32Alright? And so there's an
12:33informational challenge, which is to
12:35restore that information that's lost,
12:38at the edges of the
12:39DNA double strand break.
12:41And, so imagine the ways
12:43to repair the DNA double
12:44strand break, some which,
12:45which dominate during s phase
12:47oops. That was my pointer.
12:50There we go. Some which
12:51dominate during s phase. It's
12:53basically just slam the ends
12:54together. That's beautiful. It gets
12:56rid of the structural problem
12:57of the DNA double strand
12:58break, but doesn't solve the
13:00information problem.
13:01Alright? Because in order to
13:03restore the information problem, you
13:04have to have a template
13:06that has the missing information
13:08as a template to restore
13:09that information. The only time
13:10you have that is an
13:12s space with your sister
13:13chromatid.
13:14You also have your homologous
13:16chromosome theoretically outside of s
13:17space, but that'll that'll actually
13:19not be truly error free.
13:20You'll get loss of heterosic
13:22acid, and that occurs occasionally.
13:23Alright?
13:24The other potential source of
13:26information is RNA. And in
13:28other organisms, there is RNA
13:29templated repair, which also can
13:31restore information.
13:32The role of that in
13:33mammals is still evolving and
13:35whether that occurs or not,
13:37is not clear. Alright. But
13:39those so,
13:41and so imagine that there
13:43is, you know, all these
13:44places, all these ways,
13:46promote
13:48resolution
13:49of the structural problem. During
13:50s phase, you can use
13:51homology mediated recombination using the
13:53sister chromatid to get error
13:55free repair that also resolves
13:56the informational challenge of a
13:58DNA double strand break. Alright?
13:59And what's interesting is is
14:01that, you know, if you
14:02impair
14:04n h e j or
14:05s single strand annealing, alright,
14:07You actually don't have the
14:08cells are okay. So you
14:09you knock out fifty b
14:10p one or ATM, it
14:11caused a huge change in
14:13the entrance. You get in
14:14mice and humans, they're viable.
14:16Okay? You can have humans
14:17that live without ATM or
14:18fifty b p one and
14:19and mice that live without
14:20them. They're small sensitive radiophone.
14:22But if you mess up
14:24the genes that are involved
14:25in homogeneity of recombination, BRCM,
14:27BRCA two, you get early
14:29in mice,
14:30and your chromosomes fall apart.
14:32Alright? So that's really important
14:34for normal man management of,
14:36of s phase.
14:39The interesting thing is, of
14:40course, that,
14:42somehow, cancers that lose BRCA
14:44one in a way that
14:45we still don't really understand
14:47survive.
14:48Or if you take any
14:49normal cell, even most cancer
14:50cells, and acutely get rid
14:51of BRCA one, the cells
14:52just fall apart. Apart. Alright?
14:54But somehow, the BRCA one
14:56and BRCA mutant cancer cells
14:57survive and proliferate
14:59in a in a reason
15:00honestly, I don't think we
15:01still really understand. But, interestingly,
15:03they still they still carry
15:05this profound defect in homology
15:07mediated repair, And that makes
15:08them highly sensitive to certain
15:10forms of DNA damaging that
15:11occurs during s phase.
15:13And that includes this early
15:14work from, Thomas Helen's group
15:16and Alan Ashworth, which sound
15:17that, you know, PARP inhibitors
15:19were synthetically lethal to
15:21BRCA1 and BRCA2 immune cell
15:23lines. Important thing is this
15:24was not even found in
15:25cell lines. This is all
15:25found in primary mouse both
15:27both labs actually use primary
15:28mouse ES cells to find
15:30this. If they tried this
15:30in cell lines, it wouldn't
15:31have worked. We can talk
15:32about why that is later.
15:33Alright? They use, you know,
15:34cancer cell lines. Alright. So
15:36both these both these things
15:37are found actually in primary
15:38cells that would be our
15:39same BRCA2 division.
15:40Okay.
15:42And the original reason rationale
15:44for why this worked was
15:46that, hey. You know, part
15:47poly poly, ADP polymerase ribose
15:49is this highly abundant enzyme,
15:51and it turns on its
15:53polymerase activity whenever it sees
15:55single stranded DNA. So it's
15:56active anytime in our genome,
15:58single stranded DNA is generated.
16:00And that is basically at
16:02replication,
16:03but also at
16:06at transcription. Right? So PARP
16:08is turned on every time
16:09transcription fork is on because
16:10you have single stranded DNA.
16:11And you have also doing
16:12DNA repair. So anytime you
16:14have single stranded DNA, PARP
16:15is turned on. Alright? It
16:17causes this. And the original
16:18theory for why BRCA and
16:20BRCA may work is that,
16:21you lose PARP. Now all
16:22of a sudden you have
16:23spontaneous single strand nicks with
16:25PARP inhibitors. These are not
16:26these are not repaired. And
16:28so now, which is fine
16:29because single strand DNA doesn't
16:30cause anyone any harm. But
16:32if these go into s
16:33space, the single strand breaks
16:34become DNA double strand break
16:35equivalents, and these require BRCA
16:37on two mediated pathway to
16:38really restore the, the replication
16:40for using homology immediate repair,
16:42and you go on. But
16:43now if you don't have
16:44BRCA on BRCA two, now
16:46this is now a catastrophe.
16:48Alright? That was our original
16:49story.
16:50Alright? That it was really,
16:51you know, unrepaired single strand
16:53breaks that caused catastrophe.
16:55Now we realize, actually,
16:57it may be more complicated
16:58than that. And what PARP
16:59inhibitors really do is create
17:01bulky adducts
17:02by trapping PARP. And the
17:04lesion that BRCA1 requires
17:06is a,
17:07is a a bulky adduct,
17:09alright, that's caused by PARP
17:11trapping by the inhibitor.
17:12Alright?
17:14That's our second story that
17:15we had for why these
17:16work. Now there's several more
17:18stories that coming out which
17:19we're going to that it
17:20may be single strand gaps
17:21or replication,
17:24transcription interference that now coming
17:25in to figure out how
17:26PARP is. But it's showing
17:27that, actually,
17:28our story of why PARP
17:29inhibitors work, alright, is still
17:31in evolution
17:35of,
17:36in terms of how it
17:37works. The other thing is
17:38that it's clear that, at
17:40least in our clinically active
17:41drugs,
17:42the activity of the drug
17:43increases with the PARP trapping
17:44ability. So talazaparib is and
17:46and also the toxicity increases
17:48with the PARP trapping ability.
17:49So talazaparib is a much
17:50better drug than voleparib in
17:52terms of PARP trapping and
17:53is actually much better single
17:54agent in BRCA and BRCA
17:55immune cells. This tells us
17:57something very important in that,
17:58actually,
17:59other drugs
18:01that cause bulky adducts are
18:03also
18:04very toxic to BRCA on
18:06BRCA immune cells. So platinum,
18:08topoisomerase
18:09inhibitors, topo topotrappers,
18:12alkylators, especially bifunctional alkylators are
18:14also incredibly potent,
18:17to BRC on BRCA2 mutant
18:18cells.
18:19Alright? So classical chemotherapy
18:22is also in part
18:24highly, you know,
18:26we'll talk about this target
18:27therapy for BRCA and BRCA
18:28immune cells.
18:30And so the cool thing
18:31is this early,
18:34identification of the sensitivity of
18:35BRCA and BRCA immune cells
18:37to,
18:38to,
18:39to PARP inhibitors translated from
18:41a Nature paper in two
18:42thousand and, I think, five
18:44to a phase one study
18:45that was published in New
18:46York Medicine four years later
18:47in two thousand and nine.
18:49Alright? And we talked about
18:50why PARP inhibitors are actually
18:51already in development
18:52as protectants for, ischemia because
18:55they thought PARP was the
18:56final pathway of death for,
18:58perfusion injury in stroke and
19:00heart disease. And so they
19:01were already in development. But
19:02they were repurposed to this,
19:04and, they found that, hey.
19:05This highly this is kinda
19:06wild. This is a phase
19:07one study in New England
19:08Journal of Medicine in which
19:09the entry criteria was a
19:10genomic event,
19:12alright, which is BRCA o
19:13germline BRCA o one or
19:14two mutation, alright, and showed
19:16activity in a on a
19:17variety of different things. And
19:19this led
19:20to
19:21after a little bit, as
19:22you guys some of you
19:23might remember, after a little
19:24bit of a misadventure with
19:25inoperib,
19:26which was ended up not
19:27being a PARP inhibitor, alright,
19:29and which, you know, kind
19:30of sidelined the field for
19:32a while. Once we recovered
19:33from the inoperib
19:34catastrophe,
19:35finally, we got they got
19:36back on track, and now
19:37we have multiple approvals for
19:39PARP inhibitors,
19:40for BRCA and BRCA mutant
19:41disease in a variety of
19:43settings. Alright?
19:45But
19:46it's clear that none of
19:48these in the setting, especially
19:49advanced disease, are curative. And
19:51there are multiple mechanisms acquired
19:52resistant to PARP inhibitors, which
19:53are very informative in us
19:55both understanding the basic biology
19:56of BRCA one, BRCA two,
19:58and its role in tumorigenesis.
20:00So if you take cells
20:01that are BRCA two mutant
20:03and bathe them in platinum,
20:05ultimately
20:06ultimately,
20:07you, will,
20:08get cells that are resistant.
20:10Similarly, this is with Yas
20:11Jankar. If you take mice
20:13that are BRCA mutant and
20:13treat them with olaparib, initially,
20:15a beautiful responses, but then
20:16you'll have outbreaks
20:18of PARP resistant tumors that
20:19pop up over time. Alright?
20:22So
20:23the mechanisms of acquired resistance
20:24are actually quite informative
20:26and frightening,
20:27alright, and tell us a
20:28lot about the biology of
20:30of of BRCA1 as a
20:31role in repair.
20:33Alright? So the first thing
20:34that was figured out was
20:35by, both by Alan Ashworth
20:36and by Toshi Taniguchi
20:38was the concept of reversion
20:39mutations. And I'll remind you
20:40guys that, you know, most
20:41of the BRCA1, this is
20:42BRCA2 mutations, and BRCA mutations
20:45are frame shift mutations
20:46that introduce a premature stop
20:47codon. Alright? So for example,
20:49here's a classic one. There's
20:50a frame shift here, which
20:51theoretically will cause a truncated
20:53protein by putting a premature
20:54stop codon. Alright?
20:55Both of these actually, most
20:57of these mutations actually don't
20:58create a truncated protein because
21:00of nonsense media decay. Alright?
21:02So you guys remember, if
21:03there's a stop codon that's
21:05near a splice acceptor site
21:06in RNA, that RNA is
21:07tagged as being bad,
21:09and you get nonsense media
21:10decay. So most of these
21:11are functionally null.
21:13Alright? Except unless you're in
21:14the middle of a big
21:15exon or the very terminal
21:16exon where nonsense media decay
21:18is you can escape from.
21:19Alright? So these are all
21:20functionally null. Alright? Most of
21:22these.
21:23Right? So then, what Alan
21:25here Alan Ashworth did in
21:26this paper was take cell
21:28lines that had this truncating
21:29mutation, bathe them in platinum,
21:30got resistant cells, and then
21:33took the resistant cells and
21:34resequenced BRCA two. And what
21:36he found was that in
21:37the resistant cells, what happened
21:38was the resistant cells had
21:40acquired a deletion
21:41that deleted the exon
21:43that contained the frame shift.
21:45And the edges of the
21:46deletion were such that, actually,
21:48the gene was put back
21:49in frame.
21:50Alright? And so we converted
21:52a
21:53functionally null frame shift into
21:55an internal deletion
21:57that created a, you know,
21:59functional protein. So hypomorphic protein
22:01that was functional in repair.
22:02So it was a secondary
22:03mutation
22:04that actually fixed
22:06the first mutation.
22:07Alright? So reverse mutation. Frightening.
22:09You see this in yeast,
22:10but now we're seeing it
22:11in humans.
22:12Alright?
22:13And really shows that, hey.
22:14This is, the the,
22:17so called reverse mutation that
22:18really shows the mechanism required
22:19resistance. So this is in
22:20BRCA two in cell lines.
22:22This is work from,
22:23Toshi and,
22:26Susan Swisher in human cells.
22:27So this is a this
22:28is in with BRCA one.
22:29So this is initially a
22:30patient with ovarian cancer that
22:32had a truncating mutation, a
22:33classic truncating mutation in BRCA
22:35one two five nine four
22:36del c. It was highly
22:38sensitive to platinum initially. Upon
22:39platinum resistance, they sequenced
22:41the tumor, and this tumor
22:42had, in addition to the
22:44two five nine four del
22:45c, had
22:48an eleven nucleotide sorry.
22:52No.
22:53An eleven nucleotide deletion downstream.
22:55And so all of a
22:56sudden eleven plus one is
22:57twelve, twelve is divisible by
22:58three, you're back in frame.
23:00Alright? So it basically converts
23:02a,
23:07a frame shift mutation into,
23:08again, a small internal deletion.
23:11Alright? Putting the gene back
23:12in frame
23:14and having the cells,
23:15you know, produce BRCA1 again.
23:17Alright?
23:18With a small internal deletion.
23:20And if you look at
23:21the edges
23:23of the deletions,
23:24they all have microhomology.
23:26And so this is the
23:27BRCA1 mutant cells inappropriately using
23:29non homologous end joining to
23:31slam together whenever they see
23:32a break they just slam
23:33together the ends. And so
23:34this is in fact the
23:35inappropriate repair mechanism that's generating
23:37these reversion mutations.
23:39Alright? So it's really wild.
23:41Here you're getting putting back
23:42in BRCA1 back in frame,
23:45under selection.
23:46Right?
23:48The second mechanism of acquired
23:49resistance, which was done by
23:51our laboratory
23:52together with,
23:54Jass Jonker's laboratory as well
23:55as found, by,
23:58Andre Nusenzweig and Sam Bunting,
24:01was compensating mutations. These were
24:03acquired mutations in other parts
24:05of the pathway that also
24:06required resistance. This is kinda
24:07wild. Let's just go do
24:08a little bit of the
24:08story,
24:09which is what we did
24:11with Yas
24:13a while back,
24:14was we said, you know,
24:15what genetic events will allow
24:17ES cells to tolerate BRCA
24:19one loss? And what he
24:20set up was an ESL
24:21in which one allele of
24:22BRCA1,
24:25one allele of BRCA1,
24:26you know, has a huge
24:27deletion from exons three to
24:28actually
24:30thirteen here. And so it's
24:31this giant deletion. And the
24:33other allele has exons five
24:34and six floxed.
24:36Alright? And more importantly, not
24:38only was it floxed,
24:39but it was triggered in
24:40such a way that there's
24:41a pure micell resistant cassette
24:42that was out of frame.
24:43So when,
24:45cui is expressed, it not
24:46only excises exons five and
24:48six, but puts the puro
24:50cassette in frame. And so
24:51the tumors the cells become
24:52BRCO deficient and puro resistant.
24:55And this is important because
24:56what what the idea is
24:57that you add on tamoxifen
24:59and then add puramycin to
25:00enforce that only the cells
25:01have undergone recombination
25:03are present. And when you
25:04do that, you basically get
25:05all the cells die. You
25:06can recover this by putting
25:08an exogenous BRCA one, showing
25:09this is dependent on BRCA
25:11one. So it's the thing
25:11where we can get acute
25:13deletion of BRCA
25:14one in,
25:15in this ESL.
25:17And then the game was
25:18then we use,
25:20piggyback retrotransposons
25:21system
25:22to cause saturating mutagenesis
25:25and say, let's turn on
25:26piggyback retrotransposons.
25:27So these are little retrotransposons
25:29that can hop. They can
25:30insert into an intron and
25:31cause capture, or they can
25:33insert in front of a
25:33promoter and turn on a
25:34gene. And the idea is
25:36let's turn on a piggyback
25:37retrotransposon,
25:39add tamoxifen,
25:40add puramycin,
25:41and see if any cells
25:42survive. If they do, what
25:44happened?
25:45Alright? What was the genetic
25:46event there that caused allowed
25:47these cells to survive?
25:49And this was done initially,
25:50Peter Baumann in in, a
25:51lab that Amal Ali was
25:53working on this. What we
25:54found was that when Peter
25:55Baumann looked at two hundred
25:56and fourteen clones that came
25:58out of this assay, Two
25:59hundred of them had duplication
26:00of the endogenous VRCA one
26:02allele.
26:03Alright? So they still had
26:04VRCA one. They just had
26:05one of the alleles had
26:06the event. Alright?
26:08And allowed pureomycin resistance.
26:10The other
26:11all had deletions and secondary
26:14loss of of a gene
26:15called fifty b p p
26:15one, which is important in
26:17non homologous
26:18recombination.
26:19And sorry. In, an in,
26:21in end joining. Alright? So
26:23this is kinda wild. Loss
26:24of fifty b p one
26:25allows cells to survive
26:27BRCA one loss.
26:29Alright?
26:30And we thought, oh, maybe
26:31this is just a checkpoint
26:32effect. Alright? So just like
26:34p fifty three, somehow p
26:35fifty b p one is
26:36involved in a checkpoint. And
26:37so cells are tremendously bad,
26:38but somehow they survive. But
26:40to our surprise,
26:42these double deficient cells actually
26:44restored Raf fifty one foci
26:45formation and restored homology dependent
26:47repair. So homology dependent repair
26:49was actually restored
26:51in these cells. Alright? It
26:53wasn't just some sort of
26:54checkpoint loss. And so this
26:55is highly, you know,
26:57unexpected and said, you know,
26:58how could two wrongs make
26:59a right? Getting rid of
27:00BRCA one, supposedly essential for
27:02homologous recombination, heading getting rid
27:03of a non homologous end
27:04joining protein,
27:06or something involved in that
27:07pathway.
27:08How did that restore?
27:09Alright? Without going into the
27:10nitty gritty details, it's worked
27:12out both by,
27:13Sam Bunting and Nusenzweig as
27:14well as, our laboratories,
27:16is this really tells a
27:17lot about how BRCA one,
27:20works in DNA repair. So
27:22the story is really you
27:23know, it's all depends on
27:24how the ends are processed
27:26after DNA break. So imagine
27:27during,
27:28fifty d p v one
27:29is expressed during all phases
27:31of cell cycle. So unlike
27:32BRCA one, which is restricted
27:33in s phase, fifty d
27:34p one is present during
27:35all parts of the cell
27:36cycle.
27:37When, you have a DNA
27:38double ramp break, what fifty
27:40b p one does is
27:41basically it prevents end resection.
27:43It prevents the ends from
27:44being processed to chew back.
27:46If you can't chew back
27:47the ends, all you can
27:49do is slam them together.
27:50Alright? So,
27:52it promotes anomalous end joining
27:53during, outside of s phase.
27:56During s phase, BRCA one
27:58is
27:59expressed. And in a mechanism
28:01that we a lot of
28:02play ups are still working
28:03on, this basically suppresses this
28:04function for the b p
28:05p one and allows end
28:06processing by the CTPI MRN
28:08complex to chew back the
28:09ends since we have a
28:10free end that can now
28:11be free for homology search.
28:13Alright?
28:14Now in cells that lack
28:15BRCA one, what we think
28:17is occurring is that fifty
28:19fifty b p p one
28:20is able to do an
28:21s phase what it does
28:22outside of s phase, which
28:23is to block end processing.
28:25And when you do that,
28:26also now you no longer,
28:28can do homology mediated repair
28:30because you don't have any
28:30ends to search for,
28:32and genomic chaos ensues.
28:34Right? If you lose fifty
28:36b p one,
28:37all of a sudden, actually,
28:38CTFP m MRN complex, which
28:40is turned on by,
28:42cyclin dependent kinases,
28:43can still do its job.
28:44You actually get hyperresection, but
28:45you allow HR to go
28:47on. And so this shows
28:48that BRCA one wasn't involved
28:49in the nitty gritty of
28:50DNA repair,
28:52but was a manager that
28:53was involved in repair choice.
28:56Alright? And, the secondary mutation
28:58in b r in fifty
28:59b one
29:01actually allowed compensation,
29:03alright,
29:04and allowed,
29:05a homology media repair to
29:07occur. Alright?
29:09And in fact, so you
29:10imagine, you know, with a
29:11lot of work from multiple
29:12laboratories, there are multiple mixes
29:13required in the cyst depart
29:14inhibitors.
29:15Alright? And we kind of,
29:16in this review, kind of
29:17characterize this as,
29:19several, as several independent potential
29:22mechanisms. And understand the mechanism
29:24resist is really important to
29:25understand how to, you know,
29:27kind of, deal with it.
29:28So one is that there
29:29are drug specific mechanisms that
29:31I've seen on top. So,
29:32for example, olaparib is a
29:33beautiful PGP substrate. So upregulation
29:35of drug pumps,
29:36alright, is just like you
29:37do for anthracyclines,
29:39will actually is a well
29:40recognized mechanism of acquired resistance.
29:42If that's what's occurring, you
29:43can just switch agents.
29:45Alright? So these are drug
29:46specific mechanisms. Also, some some
29:48turning on a park, a
29:49park can also do that.
29:50And so if you switch
29:51drugs,
29:52you can actually still get
29:54activity.
29:55If you have reversion pathway
29:57reversion, either by genetic reversion
29:59or by,
30:02compensating mutations,
30:03now homology media repairs is
30:06is restored,
30:07and you have far fewer
30:08options to be able to,
30:10to handle that. Alright? Especially
30:11with genetic reversion, now you've
30:13lost the phenotypic difference between
30:14the tumor cells and normal
30:16cells in terms of repair,
30:17and you don't have a
30:18handle, to do it on.
30:19So understanding the mechanism of
30:21of of resistance is really
30:22important to figure out what
30:23we can do.
30:24So with this as a
30:25background,
30:26you know,
30:27these
30:28mechanisms of reversion
30:29and resistance tell us a
30:31lot about the basic biology
30:32of BRCA one and BRCA
30:33two. So what it tells
30:34us that, hey. You know,
30:35platinum,
30:36you know, anthracyclines, alkylating agents,
30:38all of those induce reversion
30:40mutations in the clinic as
30:41mexis required resistance.
30:43That shows they're all on
30:44target.
30:45Alright? That
30:47old fashioned alkylating agents are
30:49like PARP inhibitors. It is
30:51like topoisomerase inhibitors. They are
30:53all acting on the BRCA1
30:54dependent pathway because all of
30:56them exert selection pressure for
30:57reversion mutations. So they're all
30:59on target. Right?
31:01So,
31:03if that's true, alright, and
31:05you think about this, if
31:06classical chemotherapy is in fact
31:07targeted therapy for underlying defects
31:08in the repair, our entire
31:10dosing needs to be, you
31:11know, maximum tolerated dose makes
31:12no sense. It's a it's
31:13a target agent when you
31:14think about it as any
31:15other target agent find the
31:16extended exposure to the minimum
31:18effective dose.
31:19Second thing is you imagine
31:20that BRCA one tumors can
31:22tolerate reconstitution of BRCA one
31:23in HR mediated repair. Alright?
31:25So you can't cure a
31:26BRCA one division tumor by
31:28putting back in wild type
31:29BRCA one. It's not like
31:30p fifty three. If you
31:31put back in BRCA1 into
31:33a BRCA2 tumor, tumor is
31:34happier.
31:35Alright? This tells us strongly
31:36that BRCA1 loss and the
31:38repair defect is required for
31:39initiation of tumorigenesis,
31:41but not required for maintenance.
31:43Alright? So,
31:45we call this tumor suppressor
31:46tolerance.
31:47Alright? So this is not
31:48p fifty three, but back
31:49p fifty three, the cell
31:49dies. This is,
31:52you know, shows us really
31:53quite. And also tells us
31:54that second thing is, you
31:56know, what is actually causing
31:57the growth of b r
31:58these BRCA and BRCA two
31:59deficient tumors.
32:00Alright.
32:01It's not loss of BRCA.
32:03Right? And we try why
32:04this department plan is gonna
32:06correlate.
32:07With that background, what I'm
32:08gonna switch next to is
32:09that, you know, for not
32:10right now, you know, next
32:11generation sequencing of tumors has
32:13now become commonplace for many
32:14cancers. Alright? So now there's
32:15a variety of hybrid capture
32:16based methods in which now
32:18where people are constantly,
32:20doing tumor sequencing. In fact,
32:21at Rutgers, we started doing
32:22this since two thousand twelve
32:24in which all patients, adult
32:25pediatric, e malignancy,
32:27were reviewed, and we have
32:27a central molecular tumor board
32:29that came up. And we
32:29generate a lot of information
32:31from this. And one of
32:32the things I'm gonna talk
32:33to you about is, you
32:34know, how do we deal
32:35with unanticipated,
32:36unexpected
32:37mutations in BRC on BRCA
32:38two, other DNA repair tracks
32:39that we find on tumor
32:40sequencing? How do we figure
32:42out whether they're real? You
32:43know?
32:44Are they actionable? How do
32:45we figure this out? And
32:46I'll start again with this
32:47by giving you a case
32:48because all our things start
32:49with us puzzling this over
32:50in our molecular report. How
32:51do we deal with this?
32:52And so our first case
32:53is back in two thousand
32:54and twelve when we first
32:55started doing this. This is
32:56when Foundation Medicine was first,
32:58founded,
32:59by friends of ours, and
33:00so we collaborate with them.
33:02It's a fifty nine year
33:03old gentleman with a kidney
33:05cancer. He presented in the
33:07usual way, found to have
33:07this large mass, underwent resection,
33:09and unfortunately had rapid recurrence,
33:11in both the nephrectomy med
33:12and in the lung. He
33:13was treated with what was
33:14then the standard of care,
33:16at that point and had
33:17either, you know, rapid progression
33:18or intolerance.
33:19And so he was then
33:20sent to our molecular tumor
33:22sequencing protocol.
33:23And in addition to the
33:25alterations in NF two, CDK
33:26and ARRAD one a, there
33:28was a a classic BRCA
33:29two truncating mutation. In fact,
33:31this is a well known
33:32founder mutation in BRCA two.
33:34Alright? And so we sequenced
33:36his germline and found that,
33:37in fact, yeah, he's a
33:38germline carrier of this BRCA
33:40two mutation.
33:41Right? So now the question
33:43is, okay.
33:45Here's a guy with a
33:46germline BRCA two mutation. Just
33:47because you're heterozygous or BRCA
33:49two doesn't mean you're immune
33:50to normal tumorigenesis.
33:51Alright? So is this just
33:52a sporadic renal cell cancer
33:54occurring in a BRCA two
33:55mutation carrier,
33:57or is this a BRCA
33:58two deficient cancer?
34:00And with the underlying DNA
34:02repair defect and the sensitivity
34:04to agents we never treat
34:05renal cell cancer with.
34:07Alright? And can we figure
34:08this out?
34:10Right?
34:12And so, interestingly,
34:14you can.
34:16So this is his,
34:17the allele frequency of BRCA2
34:19in his tumor specimen.
34:21And then we have an
34:22estimated tumor
34:23of sixty two percent. And
34:24we have an estimated tumor
34:26purity in this specimen we
34:26talked about how that's done
34:28by genomics of about thirty
34:30percent.
34:31All right? And so can
34:32we make a model in
34:33which you figure out, you
34:34know, if you imagine how
34:35BRCA will lose a match,
34:36can we figure out what
34:36the expected allele frequency would
34:37be in a mixed tumor
34:39of wild type and normal
34:40cells which have or have
34:42not gone undergone or lost
34:43the wild type allele. Alright?
34:45And and so and the
34:46cool thing is, you know,
34:47with next, we have,
34:49news this is done through
34:50a hybrid capture based sequence.
34:51And so imagine you make
34:52a very simple model in
34:54which
34:54seventy percent of the DNA
34:56that we sequence from this
34:57tumor, it comes from stroma.
34:59And like the rest of
35:00his body, it will give
35:01you one mutant allele of
35:02BRCA two and one wild
35:03type allele.
35:04Thirty percent of the DNA
35:06is coming from his tumor.
35:07Alright? It'll certainly give you
35:08one mutant allele.
35:10And we know how BRCA
35:11two, for the most part,
35:12undergoes loss rise as a
35:13gas. So you get large
35:14deletions
35:15around the,
35:16around the locus. Interestingly, mechanism
35:18unclear. Right? So if that
35:20had occurred, you'd get no
35:21immune no allele at all.
35:23Alright?
35:24And so then the estimated
35:26tumor allele frequency in this
35:27setting is the number of
35:28mutant alleles over the total
35:30alleles, so hundred over hundred
35:31and seventy, or you'd expect
35:32an allele frequency of fifty
35:33eight percent.
35:35Alright?
35:36The null hypothesis here is
35:37fifty percent. Right? That's, l
35:39o that's just germline with
35:40no l o h. So
35:42can we possibly tell the
35:43difference between fifty and fifty
35:44eight percent?
35:45Alright.
35:46Cool thing is that with
35:48hybrid capture based sequencing, we're
35:50going to details, basically, because,
35:52the depth there with hybrid
35:53capture based sequencing, you randomly
35:55fragment the genome so that
35:57every region have a different
35:58start site. Alright? If you
35:59see a lot of reads
36:00with the same start site,
36:01it's actually a PCR artifact.
36:02You can actually get numbers.
36:04So when when they say
36:04it's a depth of four
36:05seventy one, that means four
36:06seventy one different DNA molecules
36:08from the samples are read.
36:09With that, you can do
36:10statistics,
36:11okay, and actually put in
36:13confidence intervals.
36:14And to our surprise,
36:16alright,
36:16he had a allele frequency
36:17sixty two percent. This with
36:18the confidence intervals around it
36:19for that depth. This rules
36:21in LOH and rules out
36:22lack of LOH.
36:24Alright? And I didn't believe
36:25this initially because I was
36:26like, I expected this to
36:27be carrier. And so because
36:29we had no other we
36:29had no other option
36:36at that time, we treated
36:36the patient with a platinum
36:36based regimen, which we never
36:36treat renal cell carcinoma with.
36:37And the patient had a
36:37dramatic response. Alright?
36:38And so we said, okay.
36:38Here's a way we can
36:39start figuring this out on
36:40expect you know, can we
36:42figure out a little bit
36:42of status? Well, this is
36:43a one off. We literally
36:44were doing this on the
36:45back of a napkin to
36:45figure out what this would
36:46be like. And so we
36:47say imagine that we should
36:48be doing this for our
36:48tumor sequencing. Imagine that in
36:50a germline
36:51setting, you can have, you
36:52know, you if there's a
36:53germline mutation, you can have,
36:54you know, lOH by
36:56gene deletion. You can have
36:57LOH by gene conversion. This
36:59is what happens with p
36:59t three, or you can
37:00have no LOH. And for
37:02each of these, you can
37:03write a relationship between the
37:04mutant allele frequency and the
37:05tumor purity.
37:06Alright? So for example, here,
37:08if there's no allergens in
37:08the tumor, regardless of the
37:09tumor purity, the mutant allele
37:11frequency will be just fifty
37:12five at least fifty percent.
37:13Alright? And the same thing,
37:15you do this for, you
37:16know,
37:17somatic mutations too. Somatic mutations
37:19can either be, you know,
37:22heterozygous in in the in
37:23the patient or can or
37:25for tumor suppressor can convert
37:27to, homozygosity
37:28either by gene conversion or
37:30by gene deletion. So for
37:31example, p fifty three often
37:33does this. Right? You it
37:34it undergoes LOH by gene
37:36conversion.
37:37Alright?
37:38And, again, you can write
37:39the equations. And for our
37:41for our clinicians, we develop
37:43nomograms to help them understand
37:45this and use this. Alright?
37:46So imagine here's, the correlation
37:48between true purity and the
37:49specimen and the mutant allele
37:50frequency.
37:52Alright? And, in in blue
37:54are all the lines that
37:54look at somatic alterations. In
37:56red are all the different
37:57models, you know, no LOH,
37:59LOH by,
38:01gene deletion, LOH by gene
38:02conversion
38:03for,
38:04a germline mutation.
38:06And if we just use
38:07this, we just plot where
38:08our patient was, thirty percent,
38:09tumor purity, mutant low frequency
38:11of sixty one percent, and
38:12put in the confidence intervals.
38:13You can see, ah, it
38:14already matches one of these.
38:16And if you don't understand
38:16what the tumor purity is,
38:17you can put in you
38:18can slide the tumor purity
38:19around. You can say, hey.
38:20What models does it correlate?
38:21And if we knew nothing
38:22else, we could tell already
38:24that this is gonna be
38:25a germline alteration
38:26with likely l o h,
38:27gene deletion.
38:29Alright?
38:30The other mutations that are
38:31seen here, like ARD one
38:32eight, should also fit in
38:33that true impurity in one
38:34of these models,
38:35And they do. It's actually
38:36a way to figure,
38:38this out. Alright? And in
38:40fact, we started doing this
38:41for all our tumor sequences
38:42to understand what the allelic
38:44status is. Alright? This is
38:45why p p three always
38:45has a higher allele frequency
38:46because it has undergone LOH
38:48by gene conversion. Alright? Because
38:49there's two copies in every
38:50cell. Not more cells have
38:51it. Alright?
38:54And so, you know, with
38:55this, with work from Nahed
38:57Jalul and Hussain Chibani, who's
38:59now at Genron,
39:02We actually made, you know,
39:03models, calculators online. We call
39:04logic for LOH and germline
39:06inference calculator, alright, in which
39:08you can take all the
39:09the allele frequencies and turn
39:11for a given term purity
39:12will, a, help you identify
39:14the term purity and give
39:15you an estimate of what
39:16the model is for each
39:17alteration, you know, whether it's
39:19somatic, germline,
39:20undergone LOH or not. Alright?
39:23And also depending on the
39:24local ploidy d two, which
39:25we won't, go on to.
39:26And, we validated this so
39:28that we'd this is with
39:29Judy Garber, actually, from the
39:31Brigham. We had a large
39:32dataset in which they had
39:33done, somatic tumor sequencing and
39:34germline sequencing and showed that
39:35we can, in fact, highly
39:37predict germline sequence germline status.
39:39But just by looking at
39:40tumor sequencing,
39:43for every tumor except for
39:44p for every gene except
39:45for p t three. And,
39:47I would like math versus
39:48I will I will leave
39:49that as an exercise so
39:50that we know why it
39:51fails for p fifty three.
39:52Alright. I can talk about
39:53it later. Alright.
39:56Alright.
39:58But then we said, okay.
39:59We have these kind of
40:00tools.
40:01Can we now sit there
40:02and go to a large
40:03dataset and sit there and
40:04go, how often are there
40:05pathogenic BRCA and BRCA two
40:06mutations seen in tumor sequencing
40:08across
40:11tumors.
40:12And so we collaborated with,
40:14Foundation Medicine, and this is
40:16work for Eden Sokol and
40:17Hossein Kivani in here. And
40:18we said, okay. If you
40:19look at their initial dataset
40:20of two hundred thousand cancers,
40:22how many times are there
40:23true truncating mutations in BRCA
40:24or BRCA two in different
40:25cancer types? And you can
40:27see that, hey. You know,
40:28there's a reasonable frequency in
40:29ovarian cancer, interestingly unexpected at
40:32that time in prostate cancer,
40:35squins skin squamous cells, which
40:36we'll talk about, breast and
40:38these others. Alright? And you
40:39can see how in addition
40:40to what we expected,
40:42alright,
40:43there's also things like, squamous
40:44cell skin cancer and some
40:46of these others. And, can
40:47you guys guess why squamous
40:48cell skin cancer has a
40:49high allele high percentage of
40:51bad BRC on two mutations.
40:54It's the highest mutation burden
40:55cancer we have, so every
40:56gene is mutated,
40:59in there. Alright?
41:02But now if something goes
41:03instead of this, let's now
41:05use our approach and say
41:06which ones can we call
41:07this biallelic.
41:09Alright. And what's interesting is
41:10is that all of a
41:12sudden,
41:13ovary,
41:14breast,
41:15pancreas stays well.
41:17Okay. Everything else drops to
41:19one or two percent.
41:20Alright. So most of the
41:21alterations we see in other
41:22tumors are somatic and heterozygous,
41:24often the setting of high
41:25mutation burden. So for example,
41:26mismatch repair, endometrial cancer. You'll
41:28see lots of truncating mutations,
41:29BRCA one or PALB two,
41:31other things, and they're bystander.
41:32If you look at them,
41:33they're somatic and heterozygous.
41:35Alright? And if you look
41:35at where the mutation is,
41:36they're in a polynucleotide
41:38repeat in the middle of
41:38BRCA one because that's where
41:39mismatch repair causes mutations.
41:41Alright?
41:42So those are bystanders they
41:43should be activated on. Right?
41:46And,
41:47but interestingly, there's still a
41:48low level of cancers that
41:50really have bialylic
41:51LH, like one or two
41:52percent. But, you know, one
41:53or two percent of lung
41:53cancer is a lot of
41:54cancer.
41:55You know, it's almost like
41:56more than fifteen percent of
41:57ovarian cancer. Alright?
42:00That's there. Alright. And so
42:02so if you look at
42:03kind of, you know, when
42:04you see germline or somatic
42:05mutations, how often do we
42:06see LOH or BRCA one
42:07or BRCA two? So interestingly,
42:09in high grade serous ovarian
42:10cancer,
42:11if you see a predicted
42:13germline or predicted somatic alteration,
42:14either BRCA or BRCA two,
42:16over ninety five percent of
42:17tumors have clear genomic evidence
42:19of loss fedoras egosy, loss
42:20of the wild type
42:23allele. Alright? In breast, it's
42:25about eighty five percent, alright,
42:27have undergone LOH. And we
42:28can talk about why that
42:29is, you know, because, obviously,
42:31as you get older, there
42:32are in fact,
42:34you know, wild type tumors
42:35that that progress in through.
42:36You know? You know, not
42:38every tumor that are a
42:39woman who has a germline
42:40BRCA one, two mutation is
42:41not immune to normal tumorigenesis.
42:43It's an additive effect. Right?
42:46Prostate is very interesting. If
42:47you have a germline BRCA
42:49two mutation, the cancer that
42:50arise almost always have undergone
42:51LOH.
42:52BRCA one, not so much.
42:54Even if your germline BRCA
42:56one, most of those are
42:56not under LOH. So those
42:58are so BRCA two is
42:59a prostate cancer gene. BRCA
43:01one, not so much.
43:04Alright.
43:05From this analysis. And you
43:06can see pancreas cancer, you
43:07also see, again, BRCA germline
43:09there, somatic alterations are harder
43:11to pick out. And if
43:11you're outside of those four
43:12genes, even if you're germline,
43:14most have not undergone l
43:15o h. Alright? So they're
43:17bystanders. There's a way to
43:18pick up, hey. Is these
43:19bystander mutations? There's really are
43:20there are are there really,
43:22alterations going on? Next question
43:24you can ask is, okay.
43:25We we show this, you
43:26know, biolytic alterations in a
43:28small subset of cancers outside
43:29of the ones we expect.
43:31Alright?
43:32Do they,
43:33really have an underlying DNA
43:34repair defect, and will it
43:35be sensitive to these agents?
43:37Is it just some sort
43:37of statistical anomaly? Is there
43:39a way we can find
43:39out? Right? So one way
43:41you can look at that
43:41is actually look at you
43:42know, there are these markers
43:43which aren't great, but okay
43:45in the setting of BRCA
43:45and BRCA two, which says,
43:46hey. You know, when you
43:47have BRCA and BRCA loss,
43:48the whole genome falls apart,
43:49and then you have clonal
43:50selection. And so you can
43:51see the pattern of genomic
43:52changes, gross chromosomal alterations
43:55that you can you can
43:56impute by,
43:58by targeted sequencing. And so
43:59there are these features like
44:00c m ary choice. We
44:01have, you know, global LOH,
44:04telomerelic
44:05allelic imbalance, large scale transitions,
44:07which are all different ways
44:08that these are chromosomal abnormalities,
44:09kinda like that original karyotype
44:11I showed you of cells
44:11falling apart, without BRCA one,
44:13BRCA two. We can actually
44:15measure global LOH,
44:17using
44:18high,
44:19quantitative that, especially in the
44:21high-depth sequencing that's done for
44:22foundation medicine, but also in
44:23certain other high depth
44:27hybrid capture based sequencing approaches.
44:29And so the question we
44:30ask is, okay, What is
44:31the level of global LOH
44:33in the tumors that we
44:33say are monoelelic
44:35versus biallelic for BRCA one?
44:37Okay. And what we can
44:39see is that in multiple
44:40tumor types,
44:41when there is biallelic
44:43status, the global LOH is
44:45elevated.
44:46Alright? But it's either monolelelic
44:48or wild type or low,
44:50suggesting that in these rare
44:51cancers
44:52that when we impute biolelic
44:53loss, there's in fact, gene
44:56evidence of a global global,
44:58you know, genomic instability pattern
45:00underlying this, suggesting that these
45:02are really, you know,
45:04affecting the genomic instability of
45:05the term and maybe there.
45:06So imagine that, you know,
45:08biolig loss is associated with
45:10evidence of an HRD signature
45:12in multiple cancer types, and
45:13maybe a better biomarker
45:15or even chemotherapy sensitivity,
45:18than just the presence of
45:19the mutation alone. Alright. You
45:20have to see what's the
45:21allelic status, and is it
45:22really is it really an
45:23early clone of it?
45:25Alright? And it's important because,
45:26you know, is we actually,
45:28Ethan published this with, others,
45:30that are here that, you
45:30know, for example, we see
45:32this in a classic things.
45:33We see lots of monolithic
45:34somatic mutations, BRCA and BRCA
45:36two, microsatellite unstable tumors. And
45:38these are a side effect
45:40of the of the underlying
45:41mismatch repair and not a
45:42driver.
45:43Alright? And so, we won't
45:44go to this data, but
45:45that's just, you know, it's
45:46important because you wanna be
45:47distracted by BRC and BRC
45:48mutations if they're not biallelic.
45:51Alright? And in fact, in
45:53many high mutation burden cancers,
45:55you will see point mutation
45:56burden cancers, you will see
45:57this. Now the question is,
45:58hey. You know, can PARP
45:59PARP inhibitors work
46:01for mutations in other DNA
46:02repair genes? Now we developed
46:03a method to look at
46:04allelic loss and a a
46:06way to look at,
46:09a global Oh, at least
46:11in this kind of data.
46:12Imagine that there are a
46:13lot of genes which have
46:13been imputed to be associated
46:15with sensitivity to,
46:17pop inhibitors. They're core HR
46:19genes, which clearly are probably
46:20in the process. BRCA one
46:22is partner BARD one, BRCA
46:23two is partner PALB two,
46:25and the RAFF two on
46:26paralogs. These are core. Lose
46:28any of these genes as
46:29lethal. It's s these are
46:30all s phase dependent. These
46:31are all the same biology.
46:32There are others which are
46:33kinda HR ish,
46:35I say. You know, ATR
46:37kind of is Birch and
46:38HR and CHK one, the
46:40zemia genes, which really involve
46:41more in interest rate and
46:42cross link repair than really,
46:44homology media repair. BRIP one,
46:46which is kind of funny
46:46in of itself, CDK twelve,
46:48possibly,
46:49in here. Then there are
46:50others which really have no
46:51no role in HR, but
46:52are, you know, DNA repair
46:53genes itself like ATM, CHEC
46:55two,
46:56MBN, and others which have
46:58been imputed but not clear
46:59that they're in the in
47:00this pathway.
47:01And so what we could
47:02do is we actually looked
47:04at,
47:05together with, with colleagues of
47:07Foundation Medicine and a bunch
47:08of colleagues in in Europe.
47:10We did a pan cancer
47:11analysis using the tools that
47:13we developed,
47:14and asked, you know,
47:16when
47:17when
47:18is by biolig loss of
47:20which genes is associated with
47:21evidence of genomic instability?
47:23Alright? And we found that
47:25there was incredibly strong associates
47:26in the biolytic status and
47:28evidence of global o h,
47:29interestingly,
47:30in BRCA one, BRCA two,
47:32RAF fifty one, PAL b
47:33two, and RAF fifty one
47:34c, the core HR genes.
47:36Alright?
47:37There was weak associations from
47:39other genes, and some of
47:40the weak associations like in
47:41RAF fifty one b,
47:43and, HR is that the
47:44numbers were low, so the
47:45error bars are high.
47:47Alright?
47:48And there was no association
47:50with things like ATM,
47:52BAP1, check two. And here
47:53we have lots of good
47:53numbers.
47:54Alright? So even in bialylic
47:56ATM loss does not cause
47:57the kind of genomic stability
47:59pattern that's associated with BRCA
48:00on BRCA two. And so
48:01we predict that ATM, as
48:02we well know, is not
48:03an HRG. ATM mutants shouldn't
48:05be sensitive to the department
48:06editors. Alright? I'll I'll understand.
48:10And in fact, if you
48:11look across
48:12the,
48:13the genome, you know, kinda
48:14look at, you know, where
48:15are the biologics, what kind
48:16of cancers have them. In
48:17addition to ovary, prostate, breast,
48:19and pancreas, where, again, prostate
48:20is driven mostly by BRCA
48:22two, alright, Not BRCA
48:25one.
48:26There is a low level
48:27of bioelectrical alteration of variety
48:29of the core HR genes
48:30in other cancers.
48:31Small percentages, but these may
48:33be appropriate to think about
48:35those are the ones that
48:35need to target with, the
48:37HR pathway.
48:38Alright?
48:40And then I will
48:41end with you know? So
48:42imagine that, you know, HR
48:44is present with all disease
48:45exam, more common BRCA one
48:46syndromes, but they are present
48:47at low level and other
48:48other types. So imagine that
48:49we really understand the allelic
48:51status of BRCA1 and BRCA2,
48:53all these other core HR
48:54repair genes to really understand
48:55when they're potentially actual,
48:57alright, and to understand the
48:58biology.
48:59Alright?
49:01Last thing I'll do is
49:02this really important because, you
49:03know, recently, there was a
49:05yeah. Not not so recently
49:06anymore, but, you know, there
49:07was this, you know, large
49:09approval for PARP inhibitors for
49:11a set of fifteen
49:12genes involved in DNA repair
49:14based on, the study. And,
49:16yeah, I'll just take you
49:16through it just to understand.
49:17It's kinda wild that this
49:18was done. Alright?
49:20So this is a labra
49:21for metastatic castration matrix americans,
49:23and they analyzed two cohorts.
49:24Cohort one was BRCA one
49:26sorry. It's just BRCA two
49:27and then ATM. These are
49:28grouped into was one cohort.
49:30And you can see that
49:31those three of the labra
49:32did better than standard care.
49:33Care. Then there was a
49:34combined cohort a and b,
49:35which is BRCA one, BRCA
49:37two ATM, and then twelve
49:38other genes. And this is
49:39the,
49:40the result of this combined
49:41cohort. And based on this,
49:42FDA approved elaborate for castaways
49:44and prostate cancer with mutation
49:46any one of fifteen, quote,
49:47unquote, HR genes.
49:50Alright? Alright. This is the
49:51current FDA approval. If you
49:53look at the data, alright,
49:54this
49:55cohort a and cohort b,
49:57there's no effect in cohort
49:58b. Just twelve other genes.
50:00Alright? And this is this
50:01is just pulled from the
50:02supplemental data. Alright? That's there.
50:04This is cohort a, BRCA
50:06one two plus ATM.
50:07Let's look at that. This
50:08is BRCA one two versus
50:10ATM.
50:12Alright. No effect in ATM.
50:14It's a BRCA and BRCA
50:15two. And in fact, if
50:16you look what's here, it's
50:17all BRCA two.
50:19Alright.
50:21There's very few BRCA one
50:23in there. Alright. So the
50:24entire cohort was driven by
50:25the BRCA two mutant population
50:27causing the effect. Alright?
50:29And so it's really important
50:30to understand this. So if
50:31you actually this is the
50:32relative increase in survival. You
50:34can see the only signals
50:35seen in BRCA two, RAFT,
50:36EVO one b, and fifty
50:37four d. It might be
50:37a little thing for CDK
50:38twelve, but the errors are
50:40error bars are overlapped. And
50:41BRCA one, of course, there's
50:43no effect.
50:44And you guys now know
50:45why because even when there's
50:46a BRCA one mutation in
50:47prostate cancer, most of the
50:48time, there's no LOH.
50:50So it's a bystander.
50:51Alright?
50:52There's a subset that may
50:53respond, but not all. Just
50:54better.
50:55Alright? And so with that,
50:57you imagine that sequencing data
50:58is a lot more information
50:58than just a list of
50:59mutations. You have to really
51:00make a model to figure
51:01out what's going on.
51:02Alright?
51:03And that they're,
51:05understanding true impurity and the
51:06variant allele frequency. You can
51:07build models that sit and
51:08go, what's going on in
51:09the tumor? What's going on
51:10in the stroma? What's going
51:11on, in the allelic state?
51:13So especially for tumor suppressors.
51:14Alright?
51:15Oncogenes is reasonable. There it's
51:17there. But tumor suppressor, many
51:19of them, you know, the
51:20most don't have as a
51:21clear haplo infusion, phenotype, which
51:23we would talk about. You
51:24need to understand the locus
51:25test. And not all DNA
51:26repair genes are the same.
51:27You can't lump an ATM
51:28with BRCA one, BRCA two.
51:30Alright?
51:32ATM mutants will respond to,
51:34like, for example, ATR inhibitors.
51:35They're synthetic lethal with a
51:36variety of things, but different
51:37than what the BRCA and
51:38BRCA two mutants will will
51:39respond to. Right?
51:41And reversion events, I didn't
51:42go into subclonal events, even
51:44clonal hematopoiesis can be picked
51:45up by this kind of
51:46analysis. So we can pick
51:46up what genes but things
51:48that fall off the model,
51:48we can pick up clonal
51:49hematopoiesis in solid tumors and
51:51other things which we published.
51:53And other things to leave
51:55you about, which I wanna
51:55is classical chemotherapy really may
51:57act as targeted therapy aimed
51:58at underlying defects and repair.
52:00And so we really need
52:00to think about how we
52:01approach chemotherapy.
52:02Alright? We could I think
52:04PARP inhibitors is nothing but
52:05chemotherapy, but given as an
52:06oral agent,
52:08side effects of PARP inhibitors
52:10are nausea, vomiting, hair loss,
52:11and neutropenia,
52:12you know, just like etoposide.
52:14Alright?
52:16And I think, you know,
52:17we need but we need
52:17to rethink how we, how
52:19we dose that if we
52:19really understand the DNA repair
52:21effects. Alright? With that, I
52:23will, stop. I'm happy to
52:24take questions.
52:28Are these some funding sources?
52:30While people ask questions, I
52:32will
52:33formulate questions. I'll ask you
52:34two quick somewhat related questions.
52:37Okay.
52:37First,
52:39are we sure that mono
52:41monorhemic
52:42loss does not increase risk
52:44at all? And
52:46I'll give you the second
52:47one too, which is when
52:48we pick up somatic
52:51BRCA one and two mutations,
52:52are they always biolabel?
52:55Yeah. So, so when we
52:57I think the germline you
52:58know, there's a lot of
52:58thing. Hey. This, you know,
53:01epigenetic silencing of BRCA one,
53:03etcetera. That may play a
53:04role in a very small
53:04subset,
53:05of cancer. So, you know,
53:06obviously, the LH events will
53:07pick up if there's an
53:08inversion or there's epigenetic silencing,
53:09it won't pick that up.
53:10It look like there's no
53:11LH. But interestingly, in ovarian
53:12cancer,
53:13alright,
53:14we don't need to that.
53:15That explains ninety five percent
53:16of the events we can
53:17see by genomic. We don't
53:18have to pick another explanation
53:19for that. That makes me
53:20think that is the main
53:21mechanism of LH. Interestingly, the
53:23few that have undergone like,
53:25in BRCA1 carriers,
53:27there's never epigenetic silencing of
53:28the other allele as a
53:29mechanism
53:30of of LOH.
53:31There's always deletion, which is
53:33interesting. There are patients who
53:34have upfront dilute epigenetic silencing
53:36in which they undergo LOH
53:37by deletion of the of
53:38the other allele, alright, which
53:39I haven't talked about. So
53:40if you have I think
53:41you either have germline mutation,
53:43somatic mutation, or epigenetic silencing
53:45first event. The second event
53:46is always deletion, which is
53:47intriguing.
53:48That suggests there's some sort
53:49of unstable area in that
53:50locus,
53:51that's there. Somatic alterations. I
53:54think most somatic alterations, VRCA
53:55or VRCA two, outside of
53:58ovarian high grade serous ovarian
53:59cancer are mostly somatic
54:01are are are mostly I'm
54:02sorry. Are mostly,
54:03a monolipid.
54:05And yet in breast cancer,
54:07they respond those patients on
54:08metastatic breast gets respond to
54:10part of the No. No.
54:11So so in sorry. Breast
54:12and ovary Right. And okay.
54:14Breast and ovary. That's right.
54:15Ovary is hundred percent. And
54:16breast two, you can saw
54:17we saw that even when
54:18you have somatic alteration. So
54:19breast, ovary, prostate for BRCA
54:21two,
54:22pancreas,
54:23the there is some sort
54:24of selection for, bilevel. So
54:26we can see bilevel alterations
54:27absolutely in the somatic setting.
54:29That's there.
54:30Questions?
54:31So, for guacamutin
54:34patients that are treated with
54:35platinum invasive therapy, often do
54:38you see,
54:39reversing event? Oh, that's the
54:41most common mechanism to resist
54:42this. It's what it's when
54:43we look for them. Often,
54:44we don't look.
54:45Alright?
54:47Alright. When you look, it's
54:48about fifty percent of the
54:49time.
54:50Alright? And the other ones
54:51we miss is probably because
54:52we're missing new versions or
54:53other mechanisms.
54:56Alright? And, I'll say I'll
54:58I'll say it somewhere. I
54:59don't believe in. I think
55:00there's BRCA one and BRCA
55:01two. They're very different genes
55:02with very different underlying things.
55:04We shouldn't my thing is
55:05we shouldn't con we shouldn't
55:06conflate them together. They just
55:07happen to be interesting in
55:08the same way.
55:11You know?
55:13That's
55:14my personal lives.
55:21Do you have
55:22a preference of sequencing then?
55:25What? In? When you have
55:26a broken mutation, how do
55:27you sequence them with what
55:28the perception part inhibitors in
55:30other eight? Yeah. I think,
55:31you know, it's hard. Right
55:32now, our standard are NCCN
55:35guidelines.
55:37Well, we treat them the
55:38same as anyone else with
55:40the addition of, of olaparib
55:41in the in the in
55:43the early stage set. Right?
55:45In the advanced setting, we
55:46have no ideas. Alright?
55:48Because they respond well to
55:49a variety of agents,
55:51because I think they're all
55:52hitting the same possible, pathway.
55:54I think the PARP inhibitors,
55:55especially,
55:56are are well tolerated. It's
55:58and some women are are
56:00well tolerated in oral dosing.
56:02Some PI patients who prefer
56:04platinum to
56:05to all you know, couldn't
56:06tolerate all elaborate. You know?
56:08And I give them either
56:09oral or liposide to load
56:10this platinum and respond.
56:11You know? But I think
56:12it's very complicated, and I
56:14think in some patients, there's
56:15audio version even by the
56:16time you see them, you
56:17know, depending on when
56:20I
56:21see it. And I think
56:22in some ways, there are
56:24many cancers in which I
56:25think a repair effect
56:26occurred early in the form
56:27of tumorigenesis
56:28and then reverts
56:30by the time you get
56:30a well established cancer. BRCA
56:32was an unusual setting where
56:33we we don't see we
56:34see that history occurring in
56:35front of us through selection.
56:37You know?
56:38But it's there. But I
56:39think, you know, we have
56:40to, another thing we don't
56:41understand is what the what's
56:42the commutation profile? What are
56:43the rearrangements that are present
56:45that are driving the biology?
56:47You know? And that's something
56:48we know nothing about.
56:53Yeah. I mean, it's just.
56:54I just have a question
56:56following up on what you
56:57just said,
56:58which I noticed you had
56:59a lot of I mean,
57:01really, they're great.
57:02You
57:03know, you're talking about mechanisms
57:05where they have a select
57:07it has a selective impact
57:08on this. It has a,
57:11an effect on the pharmacogenic,
57:14process of the cell that
57:15you the cell
57:16because
57:18But another thing that these
57:20DNA repair mutation do is
57:21just talk mutations. Right? So
57:23this is a completely separate
57:25kind of phenomenon that you
57:26haven't mentioned it. So I'm
57:27just wondering what context you
57:29might put for separating out
57:30the new cache out of
57:31sex. So for instance, early
57:33on in the generation of
57:34cancer, maybe they're just playing
57:35a role in creating a
57:36suite of mutations that then
57:38can be able to care.
57:39Absolutely. I think I think
57:40it's clear that BRCA1 loss
57:42is required for the initiating
57:44event of tumorigenesis, which is
57:45probably the two populations.
57:47And the big question to
57:48new was I call it
57:49the missing drivers. What the
57:51hell is causing the growth
57:52for
57:52a BRCA1 or BRCA2 mutations?
57:55Okay?
57:56You know, what is actually
57:57causing,
57:59growth? Because you sequence them,
58:00you're showing sequence, and you
58:00get fifty three and up
58:01in house, and then, quote,
58:02unquote, copy to go all
58:03places. You know? Ten percent
58:05of the genome. Whatever gene
58:06whatever gene you find will
58:07be, quote, unquote, amplified in
58:08ten percent of high rates
58:09of.
58:10Alright?
58:13And so, sure, there's lots
58:14of genome gaps. But underlying
58:15them, there's probably real drivers.
58:17And we actually have a
58:17large effort trying to find
58:19this, and what we see
58:20is there's not lots of
58:21infraim regions. There's this entire
58:22story we did with about
58:24out of frame reagents,
58:25the tiresome kinases
58:27that are present that are
58:28actually drivers.
58:29Alright?
58:30And I think that's what's
58:31gonna happen. It's gonna be
58:32a classical drivers,
58:33but altered in ways that
58:34we don't that we don't
58:35currently understand this DNA driving
58:37multiples,
58:39by the mechanism mutagenesis.
58:41Alright? Because the mutagen mechanism
58:43in BRCM and BRCM mutant
58:44cancers is not.
58:46It's genomic rearrangements and deletions.
58:49And we're able to prepare
58:50to actually interrogate that without
58:52currency.
58:55Right? I think hidden there
58:56is the bias. This is
58:57my hope. Alright? And I
58:59think that these will be,
59:00like, non small cell of
59:01cancer. Like, two percent of
59:02this, four percent of that,
59:03five percent of this, it
59:04gets you caused mutagenesis
59:06and then selection,
59:08okay, or some driver events
59:10that are if we believe
59:11one.
59:16Thanks for a great talk.
59:18Thanks for being here. I
59:19believe.