Chronic Lymphocytic Leukemia: Genotype to Phenotypes and Beyond

January 17, 2024

Information

Yale Cancer Center Grand Rounds: Distinguished Lecture

ID11190

To CiteDCA Citation Guide

00:00Good morning. So for those of you who
00:04either can't see me or don't know me,
00:07I'm Eric Weiner and I'm really pleased
00:10to be here to introduce Kathy Wu.
00:14This is the inaugural lecture of what we
00:19hope will be a new series and many of us
00:23in the Cancer Center spent a lot of time
00:26thinking about how we want to do conferences.
00:28And we looked at attendance and we looked
00:31at who goes to what and ultimately came
00:34to the decision that Grand runs as it was,
00:37which is now trying to be in
00:39person as much as possible,
00:42was largely attended by clinically
00:43oriented people in population,
00:45scientists and people who are otherwise
00:48looking for lunch and or breakfast.
00:50And and that there was really a need for
00:54a conference that focused a bit more
00:57on translational and basic questions.
01:01And so after some thought,
01:03a small committee of people that
01:06included Katie Politi and Megan King
01:09came up with the idea of trying a
01:12conference like this on a monthly basis.
01:14And this is the first of those.
01:17So I'm really pleased to have Kathy Wu here.
01:20I've known Kathy for many years.
01:23She was a fellow at Dana Farber and of
01:27course it's still with Dana Farber when
01:31I was a substantially younger attending.
01:35And in fact we worked together in clinic,
01:39yes, briefly.
01:40She dabbled a little bit in seeing a
01:44patient with breast cancer or one or two.
01:47And so I've known Kathy now for 20
01:51plus years and Kathy has built really
01:54a phenomenal career at at Dana Farber.
01:57Her own interests are broad.
02:00I learned last night something
02:02I didn't know before,
02:03which is that she even had an
02:05interest in sickle cell disease and
02:08therapeutic approaches to sickle
02:10cell disease way back when,
02:12but ultimately decided that some
02:14some amount of focus was needed.
02:17And her interests have really
02:19focused on immunotherapy and Col.
02:21and and beyond that,
02:24the development of vaccines and
02:27and tumor specific vaccines.
02:31She is presently a Professor of
02:33Medicine at Harvard Medical School
02:35and the chief of the Division of
02:38Let Me See If I Get This Right,
02:40Stem cell transplantation and
02:42cellular therapies at the the
02:45Dana Farber Cancer Institute.
02:47So it's really a pleasure to have you here.
02:51We're all looking forward to
02:53your talk on largely on CLL.
02:56And thanks so much for coming.
02:58We had I will,
02:59I will just say that a small group of us
03:01had a great dinner with Kathy last night.
03:04And in addition to being a great scientist,
03:08she's also just a delightful
03:09person to have dinner with.
03:19Well, it's really an honor to be here
03:22and and happy New Year, Happy Snow day.
03:26Thank you everyone in the
03:28room for trudging in this.
03:30It's really great to see you in person
03:34and and also to all the folks out in Zoom.
03:37I hope this is a successful series
03:39because I do think that the intersection
03:41between the clinical and the basic
03:43and really kind of being able to look
03:46at the translational opportunities
03:48that are afforded by the patients that
03:50we treat in the study are are are
03:52immense and so and very rewarding.
03:54So and as as Eric said I I do have many,
03:58many different different interests.
04:00I think that's a hallmark of a of a happy MD.
04:04So like we we're interested in a
04:06lot of things and and thank you for
04:08giving me the opportunity to maybe
04:10share some of the work that we've
04:11been doing in CLL Genomics. OK.
04:15So we'll start.
04:17Let's see
04:27here we go. Disclosure slide,
04:30OK, I thought I'd start here,
04:31which is you know I think just a a
04:34challenge to all of us in the cancer
04:37community whether or not we study
04:38CLL or not is really the challenge of
04:41tumor heterogeneity and evolution.
04:42This has really been kind of
04:45understood for quite some time now,
04:47made ever more clear through all the
04:49genomic studies that have been out there.
04:51But we know for sure that cancer
04:54is a heterogeneous population,
04:56for better or for worse.
04:57Unfortunately,
04:58by the time that we are diagnosing
05:00patients with cancer,
05:01we're really here at the time of
05:03escape where there's already so many
05:06different resistance mechanisms
05:07that have really come into play
05:09that make the tumor fit to expand
05:11and grow in the patient host.
05:14We also increasingly know that
05:15this is not happening in a vacuum,
05:17that there's an interaction
05:18with the host immune system.
05:20But again,
05:21by the time that we're seeing patients,
05:22there's so many different immune based
05:25escape mechanisms that are at play as well.
05:27And so a lot of the questions that
05:29I think as a field that we're
05:31really interested in asking is
05:33not only this question of tumor
05:35heterogeneity and evolution,
05:36but also how do we understand this,
05:38these heterogeneous tumor
05:40microenvironments are T cells there
05:42at the right place at the right time?
05:45How are we responding to
05:47diverse immunotherapies?
05:47And then what is the role of a tumor
05:50antigen in shaping the tumor response?
05:52I'm not going to talk today so
05:54much until the very, very end,
05:55but this is a very large area
05:57of interest in my group.
05:59And as I said,
05:59I'm going to focus on chronic
06:00lymphocytic leukemia,
06:01which honestly the questions that I'm
06:05asking could be in any sort of tumor system.
06:08But CLL really has a lot of very
06:10unique features about the disease
06:12that have made it exceptional
06:13for the study of genomics.
06:15First,
06:17in a small tube of blood you have
06:19very pure tumor that can is readily
06:21accessible directly from the patients.
06:23The other thing is for a cancer,
06:25it's quite indolent.
06:26And So what that means is that
06:29we really have really long
06:30disease histories of patients.
06:32We can really take snapshots in
06:35time and study evolution in real
06:38time along with the patient.
06:40And so for some time now,
06:42our group together with colleagues
06:43in the Boston area,
06:44we've actually had this program
06:46where we've been trying to study
06:47the link from genome to phenome.
06:49How can we genomically characterize CLL,
06:52how can we understand the clinical
06:54course in response to therapy and
06:56then how can we also functionally
06:58characterize the pathway dependencies and
06:59really thinking about how we can do better.
07:02So what I'm going to talk about today
07:05is update the group on recent genomic
07:08studies and CLL driver discovery bid
07:11on our efforts in looking at tumor
07:14heterogeneity in our CLL GEM models.
07:16And then just a few perspectives
07:18of where we're going
07:20next in terms of the genomics.
07:21Again, as I said this is a
07:24very in general for cancer an
07:27indolent disease, it's typically
07:31marked initially by what
07:33we call watch and wait.
07:34So there can be a long lead time,
07:37but ultimately with treatment there can
07:40be cycles of recurrence that happen
07:43with shorter and shorter intervals,
07:45much like what we see in other tumors.
07:46I think a question that has
07:49always fascinated people in this
07:50field is how do we understand
07:53who progresses faster or slower?
07:54And what I mean by that is that
07:56there are some patients who succumb
07:58to their disease within two years.
08:00There are others that can have a
08:02little bit of therapy here and there
08:03go on for more than 1015 years.
08:05So why is that?
08:06What are the differences between
08:08the patients despite all their
08:10cells looking relatively similar
08:12under the microscope?
08:13And so for since forever there
08:15has been a long effort to try to
08:19understand those markers that we could
08:21use to distinguish amongst patients
08:24initially looking at clinical features,
08:26protein markers.
08:26But I would say over the past 10-15 years
08:28since there's been the next generation
08:30sequencing that's been available to us,
08:32there's really been an explosion of
08:34knowledge in terms of the genetic
08:36alterations later on top of that,
08:38the transcriptional alterations and
08:39even the epigenetic alterations so
08:42that we can understand what's going on.
08:44This slide really summarizes a lot
08:46of work that has been done since
08:49next generation sequencing has come upon us.
08:51I would say that the first studies
08:54in genomics arrived around 2010,
08:582011.
08:59We were among the first
09:02to describe mutated SF3B1.
09:04So a splicing factor that kind
09:05of came out of the sequencing.
09:07No one had until then kind of
09:10puts altered splicing and and
09:12lymphoid malignancies together.
09:14There's been large scale studies
09:17in looking at clonal evolution.
09:19So again CLL was one of the first
09:22places that studied really this kind of
09:25concept of clonally evolving subpopulations.
09:28And then and you can see initially our
09:32studies were about 100 patients and then
09:35around 2015 about 500 patients per cohort.
09:38What I'm going to describe for you
09:40now is our recent work trying to
09:41put together all of these different
09:43studies together so that we could
09:45get a cohort of more than 1000.
09:47I want to say that during this
09:49time that we've kind of performed
09:52these sort of genomic studies,
09:53there has been vast changes in the
09:56therapeutic landscape of CLL therapy.
09:58So whereas previously it was very
10:04standard to get chemo immunotherapy.
10:06I would say that in the in the same
10:08time that time frame that I'm speaking
10:10there has been the introduction
10:11of targeted inhibitors of BCL,
10:13two of the B cell receptor signaling
10:18and also introduction of immunotherapy.
10:20So the really big changes,
10:22you know as we start to think
10:24about the the genomic lesions.
10:26So how do we build an integrative CLL map?
10:30Well,
10:30we joined forces between our
10:33colleagues in North America but
10:35also with our colleagues in Spain
10:37and Germany and together collected
10:41cases for which there was exomes,
10:43genomes, RNA sequencing
10:45and methylation profiling.
10:46And there was a nice overlap of these
10:50different platforms in in several hundreds
10:52of patients samples that we collected.
10:55And this is a kind of a
10:59intimidating commute plot,
11:00but I think it just speaks of
11:02a number of different things.
11:04First, I want to acknowledge the young
11:06people who were the leaders of this project.
11:09It was really an international collaboration.
11:11So I had the pleasure of working with
11:14Binyamin Nisbacher and Ziao Lin and
11:17Gaddy Goetz's group computational gurus
11:19and then Cindy Hahn from Dana Farber.
11:22Awesome lymphoma oriented fellow
11:25and then Ferran,
11:27Nadeau and Marty from the group in Spain,
11:31the Spanish CLL group in Barcelona,
11:33Barcelona.
11:34And then when we looked at these
11:36more than 1000 patient samples,
11:38in fact we were able to have
11:41greater sensitivity.
11:42In the magenta are all the new
11:45drivers that we identified.
11:46So each row is a driver alteration,
11:50each column is a different case and
11:53what you can see is in fact there is
11:55a a list of recurrent alterations,
11:57but a long tail.
11:59You can see that a lot of our discovery
12:01is down here at the one 1% or less level.
12:04So many,
12:05many different sort of driver
12:08alterations that we had greater
12:11sensitivity to identify because of
12:14the increased power of our cohort.
12:16Just to make a a really beautiful
12:19Long story short,
12:21we were able to double the number
12:23of CLL drivers that we were able
12:26to identify previously.
12:27There were about 10% of patients
12:29that we couldn't account for.
12:31There wasn't any sort of driver
12:33alteration that we could point
12:34to that was this is the reason
12:36that they have CLL and we've been
12:37able to close that gap so that
12:39there's only by now 3.8% that we
12:42can't account for the two large
12:44categories of CLL that are well
12:47known in the clinical arena on the
12:51basis of their immunoglobulin locus,
12:53the mutated and unmutated IGHV.
12:56We finally had enough power to
12:58actually break those two groups
13:00apart and look and look at them
13:02separately and they really look
13:04like very different diseases.
13:05They each have distinct molecular landscapes.
13:08It highlights the diverse
13:10trajectories of clonal evolution.
13:12So maybe by virtue of where
13:13you start as AB cell,
13:15maybe there's a path of different
13:16paths of least resistance that gets
13:18you to where you're going to be.
13:19And what was super interesting is
13:21that at least for the unmutated CLLS,
13:24their their source of heterogeneity
13:26was genetic.
13:27There was a lot of lot more putative
13:29drivers in this unmutated group,
13:31but in the mutated group,
13:34relatively few drivers,
13:35but a lot of transcriptional diversity.
13:37So really a different path to
13:39achieving that type of heterogeneity.
13:42And then what I want to show you is
13:45that when we looked at the expression,
13:47you know,
13:48Benjamin was able to identify what
13:50he called E CS expression clusters.
13:53And then the nomenclature here is some
13:54of them were enriched for M for mutated,
13:56some for unmutated.
13:57And what you can see is that it
14:01actually breaks down the group's
14:02more or less based on mutated on
14:05mutator or by their epigenetics.
14:07But you can also see by the fact
14:09that there's two colors within each
14:10column that there was contribution
14:12from both mutated and unmutated to
14:14these different expression clusters.
14:15And one example in one one.
14:18One thing that was really interesting
14:19is that by the yellow asterisks we could
14:22see that certain genetic alterations
14:24actually also segregated together
14:25with these expression clusters,
14:27suggesting that they were a
14:29cohesive entity each of these
14:31different expression cluster group.
14:32So for example trisomy 12,
14:34which is a very well known cytogenetic
14:37abnormality associated with CLL,
14:39but for which there's great
14:41heterogeneity in kind of the
14:43behavior of those trisomy twelves.
14:46They actually split out into two groups,
14:48one that's in a more
14:50predominantly unmutated group,
14:51another in a predominantly mutated group.
14:53And this maybe provides us with
14:56some understanding for why some
14:58samples with the same sort of
15:00cytogenetics might behave differently.
15:02And what was super interesting is when
15:04when Benjamin started to look at these
15:06different expression cluster groups,
15:07they actually did display
15:09different clinical outcome because
15:11we had very long clinical.
15:13These were also clinically
15:15annotated samples as well.
15:17And this is just kind of the
15:18final data slide related to this,
15:19which is indeed when we kind of
15:22breakdown the samples based on
15:24their classical clinical group,
15:25based on the expression clusters,
15:27whether they were concordant or
15:29discordant to that classification,
15:31we could actually see differences in
15:34their clinical outcomes suggesting
15:36that our expression cluster system
15:38was actually increasing the accuracy
15:40of what we're trying to do in
15:43terms of prognostication.
15:44So we've been really excited to,
15:46I mean this is really,
15:47this was really a Tour de force
15:49effort to bring together not
15:50only all these different groups
15:52together and their expertise,
15:53but also to layer on all of these
15:57different genomic layers to kind of
15:59identify unique molecular subtypes.
16:01And I do want to say that this,
16:04these studies were samples that were
16:06collected in the era of chemo immunotherapy.
16:09We are actively trying to look now
16:12how these relate to the modern era
16:15of targeted inhibition and we also
16:17are interested in in trying to look
16:19at whether or not the different
16:20molecular subtypes have differences
16:22in therapeutic vulnerabilities.
16:26Now I think you know as we've gotten
16:28better with our therapies we we
16:30always have to kind of reckon what
16:31is the area of most unmet need.
16:33And I think right now clinically
16:35for the for CLL there are so many
16:38different therapies available,
16:39but we are still really faced with
16:41the conundrum of Richter syndrome.
16:43This is really it's a rare,
16:46it occurs in five to 10% of patients
16:49with CLL but it is a transformation
16:52of a small indolent histological type
16:55into a high grade lymphoid malignancy.
16:5890% have Histology similar
17:00to diffuse large B cell,
17:03large B cell lymphoma.
17:05The majority are clonally unrelated.
17:08We know that because if we
17:10follow their immunoglobulin,
17:11the clonal immunoglobulin,
17:11we could see the same in the patient.
17:14Shown here is a micrograph that shows a
17:18sample where you can see the coexistence
17:20of these two entities within the same
17:22sample and you can see the really the
17:24big kind of histological differences.
17:26These are the patients that we typically say.
17:28I'm so sorry. Please get your affairs
17:30and orders that there's really not
17:32much more that we can do for you.
17:34And it's been very difficult to
17:37understand molecularly much about this
17:39entity because there's been limitations
17:41of tissue sampling and and and it's
17:44really based on morphologic diagnosis.
17:46There's been a lack of markers and
17:49understanding of genetics and for
17:50a blood based malignancy like CLL,
17:52Richter's is really like a solid tumor.
17:54I mean, this is really so unlike
17:57what I said before where there's ease
17:59in kind of having blood draws here.
18:02We have to get biopsies often
18:04FFP specimens in order to study.
18:06And and this has not been, not been easy.
18:10But I would say that over the past couple
18:13years that because of the availability
18:15of all these nice genomic platforms,
18:18there's there's been really an
18:19explosion of new studies that have
18:22come out in the past year and a half.
18:24And at the same time there's been
18:26modeling that's been done trying
18:28to really put our attention to
18:30how we can generate mouse models,
18:32whether they're PDXS or or Gem
18:34models to try to understand this.
18:35And there's been actually a lot of
18:38progress in understanding the genome
18:39that the genetics looking at the
18:42epigenetics and the transcriptomics.
18:43And So what I'm going to demonstrate
18:45for you in the next couple slides
18:47is some of our efforts in this area.
18:49This is really work that's been that
18:51was led by Aaron Perry who is now
18:54a new junior faculty member at the
18:56Dana Farber in the lymphoma group,
18:58Roman Guiz who's part of Philo back
19:01in in France and Ignot Lechner who is
19:05now a junior faculty member at BU.
19:07And what we tried to do was assemble
19:10a nice paired matched cohort.
19:12So in other words,
19:14not just Richter samples in isolation
19:16but antecedent CLL matched together
19:18with the Richter's where we could track
19:21evolution in time across these patients.
19:23This was about 50 patients
19:24that we collected samples on.
19:26I think the point of emphasis that
19:28I want to show you on the left
19:29side here is the CLL course,
19:31the green is the different lines of therapy.
19:33On the right side is the Richter's
19:35and I want to show you that on the
19:37left side it's years where whereas
19:38on the right side it's months.
19:40So this kind of gives you a sense of kind
19:42of the the time course of these patients.
19:44The black dots are the different samples
19:46that we collected on the CLL course.
19:48The yellow here is the Richter
19:50diagnostic sample.
19:51Unfortunately,
19:51there's a lot of red here,
19:53which is that the patients
19:55did succumb to their disease.
19:56There's a number here with
19:58black arrows that are living.
19:59For the most part,
20:01these are patients who.
20:02We received therapy and then went
20:03on to stem cell transplant and
20:05really did a complete overhaul.
20:08So we we obtained eggsomes on most of
20:11these patients also had some genomes,
20:13RNA sequencing and single
20:15cell sequencing data.
20:16But I want to point out to you
20:17that you know a lot of these
20:19studies are really quite different.
20:20I think that the the conundrum that
20:22we've met with Richter's is that it's
20:25really two malignancies in the same sample.
20:27So how do you pull apart the genomic
20:30contributions of one versus the other.
20:33And for that we had a come up
20:36with a computational approach
20:37that was quite challenging,
20:39but we were able to succeed where
20:42we really optimize the copy number
20:44analysis to deal with FFPE artifact.
20:47We had a number of different filters
20:50that allowed us to kind of increase
20:53the sensitivity of our analysis
20:54and deal with contamination of
20:56whether tumor in the normal or
21:02the reverse.
21:03As I said the artifact from FFPE.
21:05And then we were able to put in our
21:08algorithms that allow us to identify
21:09clones and then also establish phylogeny.
21:12So at the end of the day,
21:14we were able to separate out the
21:16contributions of the CLL clones
21:18compared to the Richter's clones.
21:19And in doing so then we could look
21:21at start to look at phylogeny and
21:23understand which branches were CLL
21:25versus Richter's and look across time.
21:27So again, Long story short,
21:29I think one of the questions that has been
21:31asked in the field is it is Richter's,
21:33is it a distinct entity,
21:35is it similar or is it different
21:38from the Novo DLBCL?
21:40And here we had the advantage of being
21:42able to access older data of more
21:45than 300 samples of lymphoma that our
21:47colleague market ship had collected.
21:50And then using those data we
21:55performed unbiased NMF clustering.
21:57And you can see across the purple
21:59on the top that the Richter's
22:02really stand different.
22:03They're you know separately
22:06from DLBCL and so the the,
22:11so this is clonally unrelated Richter.
22:14So these are the few samples here
22:17do appear to be like de Novo DLBCL,
22:21but the vast majority,
22:23the clonally related stand separately
22:28among the Richter's itself.
22:29We were also because of all the
22:31genomic alterations that we found we
22:33were able to also perform unbiased
22:36clustering and discern that there's
22:38actually it appears to be molecular
22:41subtypes within Richter's itself
22:44and these TP 53 has long been
22:47associated with Richter's but we can
22:49see that there's different flavors.
22:50So this one here has enrichment in
22:54whole genome doubling this group.
22:57Here RS3 has Co occurrence with
23:00Notch one also deletion 15 Q which
23:04covers MGA which is effects Mick
23:09and then RS5 also has Notch one
23:13as well wild type Notch one and a
23:17lot of copy number alterations.
23:18There were also two other
23:19subtypes that did not have
23:23TP53K Ras S Pen, Notch one together with
23:26Trisomy 12 and also SF3B1 with EGR Two.
23:29And again these different subgroups
23:31appear to have different clinical
23:33behavior where the ones that have
23:35TP 53 seem to have worse prognosis.
23:38Now what is the meaning of kind
23:39of trying to look at all these
23:41different genomic alterations?
23:42Well one thing we realized is that
23:45maybe we could harness all of this
23:47and actually look to see this,
23:49whether this could help us devise a non
23:52invasive approach to identifying Richter's
23:54and getting us to earlier detection.
23:58And it turns out that with simply ultra
24:01low pass genome sequencing $150.00 a pop,
24:05you can focus on these different alterations
24:08that we identified and start to look.
24:11And in fact we were able
24:12to see in this example,
24:14this is a patient where we could
24:16identify the Richter's alterations
24:18even close to five to six months
24:20before the actual diagnosis.
24:22So if you follow this in the blood,
24:24the blood cells have CLL at this
24:26time early on and it's a very,
24:28very quiet genomic profile.
24:32Whereas the plasma shows all of
24:34these different alterations that
24:36match very similarly to what was
24:39detected much later when the actual
24:41the the tissue diagnosis was made.
24:43We've been able to see that in a
24:45number of different other cases.
24:47This is a nut.
24:48Whoopsie, this is another case.
24:51Well anyway,
24:52let's see
24:55where the in the plasma we were able
25:00to again follow find those kind of
25:03Richter's genomic alterations that
25:04was not evident in the blood cells.
25:08And finally, this is a case of a
25:10patient who went through transplant and
25:15we were able to identify post transplant
25:18relapse months before the actual diagnosis
25:21and then institute therapy and you
25:23see those alterations go away again.
25:26So I think just to summarize
25:27this part of the talk,
25:28I I would say that we've been able to
25:31actually find that the majority of
25:34Richter's does evolve from CLS subclones
25:36through acquisition of additional drivers.
25:38Clonally related Richter's is
25:40distinct from de Novo DLBCL.
25:43There are molecular subtypes
25:46of Richter's that have and and
25:48these different subcategories do
25:50have prognostic significance.
25:52And then the we're very excited about
25:54the self free DNA as a way to get us
25:57to non invasive earlier diagnosis
25:58because I think this could be really
26:01quite impactful for our patients.
26:05I think we're always trying to.
26:06So I'm going to transition now in
26:10terms of talking about the immune
26:13microenvironment for Richter's.
26:14You know,
26:15I think we're always trying to gain
26:17a bird's eye view of the landscape
26:19and really the advent of single
26:21cell analysis has really been
26:23so impactful all around.
26:25This is something I put together with
26:27one of my postdoctoral fellows where
26:29we tried to look at across the field.
26:31You know single cell sequencing was
26:34named the method of the year in 2013
26:36and then subsequently 2019 in multi
26:38ohmic analysis was the method of the year.
26:41CLL has had a bit of a lag time in
26:43terms of the the rest of the field,
26:45but again the easy access to material
26:48has really kind of stimulated us to
26:50start to look a little bit more closely.
26:53We've been able to apply this approach.
26:55Again I mentioned that Richter's
26:57is this area where the therapeutic
27:01opportunities are not great,
27:05but what has caught the attention of
27:08many is that it turns out that there is
27:11a response to immune checkpoint blockade.
27:13So fit 42 to 65% responses to
27:16PD1 blockade in Richter's.
27:18This is really quite remarkable because
27:20a lot of blood B cell malignancies
27:23do not have a great response to
27:25to PD anti PD one and so this sort
27:30of across these many studies.
27:32This raises the question are
27:33there determinants of response
27:35and resistance to PD1 blockade.
27:36We were able to partner together
27:38with our colleagues at MD Anderson.
27:41Again this is the work of Aaron
27:43Perry where they had already started
27:45a trial where they had patients
27:48initially on nivolumab and then then
27:50after the first cycle then ibrutinib
27:52was started and then response
27:54assessment happened at three months.
27:57And so we were able to collect bone
27:59marrow samples from these patients,
28:01a number in the green that had either
28:03a partial or complete response to
28:06patients that had progression even
28:07at the three month time point.
28:09And then just for comparison to CLL,
28:12patients were treated on the same
28:14trial and what Erin did was she was
28:16able to take marrow samples from
28:18these patients and through flow
28:20cytometry you can see that the
28:24the small cells were the CLL cells,
28:25the large cells were the Richter's
28:27and then there was another
28:29population here which was neither
28:31and this was the immune cells
28:33that were in the bone marrow.
28:34And then she was able to
28:36perform a single cell
28:39characterization.
28:40And again to summarize a large body of work,
28:43what was really clear is that the
28:45responders compared to the non
28:47responders when you started to look
28:48at all of those single cell transcriptomes,
28:51those there was a kind of a cluster
28:53of cells that kind of segregated
28:54with a unique phenotype and
28:56we called this cluster one.
28:58It turns out it was high
29:01expression for a transcriptional
29:03factor called Hobbit ZNF 683.
29:05And as she started to look
29:07at this population,
29:09she was able to perform some functional
29:11studies and demonstrate through cut
29:13and cut and run and various various
29:16different sort of over expression and
29:18knockout kind of analysis that ZNF
29:21683 does appear to regulate T cell
29:24pathways with activation cytotoxicity.
29:26When we started to look at the
29:29trajectories the ZNF 683 high
29:30seemed to be a divergent pathway
29:32from terminal exhaustion.
29:34We also looked across other different
29:37solid tumor till settings and it turns
29:40out that the ZNF 683 high does mark
29:43a population that's of patients that
29:45have better response to PD one therapy.
29:48Notably we looked at Melanoma
29:50across and other settings and also
29:53in she was also able to see that
29:56you know we did our analysis in
30:00the marrow but to make it more
30:03clinically facile could could this
30:05be actually detected in the blood.
30:07And so she was able to look at independent
30:09patients who are responders or non
30:11responders on the MD Anderson trial.
30:13And in fact the responders have a
30:16very distinct profile in the blood T
30:18cells compared to the non responders
30:20where there is high expression of
30:22Z and F683 and and other cluster
30:25one genes as well and this is we.
30:31So we were very proud of Aaron and
30:34Camila to get this into cancer cell.
30:36We actually tried to for a cover.
30:41It did not work.
30:42So you will never see this published
30:44only here in the seminar series.
30:47But we were trying to make a play on
30:50ZNF 683 and The Hobbit and the idea
30:53that if those of you were Middle
30:56Earth aficionados or token lovers,
30:59you know,
31:00the idea that you can either take
31:04a path and get to the valley of
31:06death with exhaustion or you can
31:08take a divergent pathway and end
31:10up back in the Shire happy.
31:12So that was our idea. Didn't work.
31:15Whatever.
31:16So, so that.
31:18I'm going to move on to
31:20the second set of study,
31:22second chapter shall we say
31:24in trying to look at function.
31:27And here you know in the same
31:29way that in the,
31:30in the genetic realm we've been able
31:33to study heterogeneity in patients.
31:36Well,
31:37can we not actually generate mice
31:41that are actually faithful to the
31:44disease through the by mimicking
31:46some of these genetic alterations
31:48that we've identified And then
31:50that provides us a platform with
31:52studying mechanism of disease
31:54and testing novel therapies.
31:55And I just want to point out that
31:57there are different flavors of models.
32:00I I don't need to tell this audience
32:02or folks that yelled at, but
32:06the GEM models in general in,
32:07in particular I just want to point
32:09out have the advantage that this is
32:11kind of in a physiologic setting.
32:12It does allow us to look at tumor evolution
32:16and also immune micro environment analysis.
32:20And so for the past period of time,
32:23my group has really been
32:24interested in this question, well,
32:25how do you get from AB cell,
32:27what are the kind of pathway hits
32:29that happen that gets you to CLL?
32:31And can we study some of these alterations
32:34that we spent a lot of time genomically
32:36identifying such as SF3B1 or IK,
32:41CF3 or DMT3A and so and so forth and
32:45can we start to look at these things.
32:47So I won't go over these past studies
32:50only to say that it has in fact been
32:53very gratifying to generate these
32:55mouse models and to demonstrate that,
32:57yes, these putative drivers that
33:00we've identified through sequencing
33:03actually generate CLL in mice.
33:05Most recently we had a very nice
33:08study ELISA 10 Hacken generated
33:13the setting where using CRISPR she
33:15was able to introduce combinations
33:16of different alterations and release
33:19combinatorial study the different models
33:21of CLL and Richter's that we identified.
33:23But for today, I'm going to talk
33:25about new unpublished data where
33:27we've been focused on one of the
33:29newer drivers that we identified,
33:30RPS 15 and some of the insights
33:33that we've identified there.
33:35So RPS 15, what is it?
33:38It is a ribosomal protein.
33:42It's identified in 5% of CLL patients.
33:45It's enriched in patients who
33:48are relapsed following therapy.
33:51It's associated with shorter
33:53progression free survival and it
33:56commonly Co expresses with TP53.
33:58One of the things that we found
34:00interesting about RPS 15 is that there
34:03does seem to be a hotspot region
34:05where a lot of the alterations happen.
34:07And so this kind of piqued our interest
34:10in trying to learn more about RPS 15.
34:12I do want to put this in the context
34:14that they're across different cancers.
34:16There's been a lot of different ribosomal
34:19mutations that have been found for CLLR.
34:22PS15 is the only ribosomal
34:25mutation that's been identified.
34:27But certainly across other
34:29cancers including breast cancer,
34:30Melanoma, myeloma,
34:32you see that this biology seems to be there.
34:35And carbosomopathies have been
34:37associated with a variety of
34:39different altered functions,
34:41so including DNA damage,
34:45proteasomal alteration and metabolic
34:47rewiring.
34:47So we were interested in trying to dig
34:50a little bit deeper about this in CLL.
34:54So we used our,
34:56we used this in a similar fashion to
34:58the other mice that we've generated.
34:59We introduced one of these hotspot mutations
35:04that was intercross with CD19 cream mice.
35:07So this alteration is only present in B
35:09cells in the context of CD19 expression.
35:12So in B cells we were able to generate both
35:16heterozygous and homozygous mutated mice.
35:18We also intercross also with deletion 15,
35:22sorry TP 53,
35:24so that they were also mice that
35:27had double mutations as well.
35:31And so this is just a bit of
35:32the targeting strategy.
35:33This was really studies led by
35:35an MDPHD student and currently
35:38at MGH as a as an intern.
35:41And then Marwan Kwok is a awesome postdoc
35:44in my group right now who's leading
35:45up on some of the functional studies.
35:47Neil Ruthin is in grad Graduate
35:50School in the New York area
35:52for computational biology.
35:54So RPS 15 mutations,
35:55we we're very able through our mouse
35:58models to confirm that it does have
36:02oncogenic potential because certainly
36:04over time we're able to identify that
36:08there is a population of RPS 15 mice
36:12that are do have expanded B cells.
36:15You can see this also in
36:18screen sizes over time.
36:20It does take quite a bit of time
36:22consistent with the human disease.
36:23It does take about 15,
36:28about 818 months,
36:2918 to 218 months to two years
36:31in order to see disease.
36:33So this is really a labor of love.
36:36But I would say that for sure with
36:39the RPS 15 mutations mutant mice we
36:42do see onset of disease less so with
36:46just the TP single mutant TP 53 but
36:49with a double mutant we also see not
36:52only CLL but evidence of Richter's.
36:55But what was interesting is in the
36:57setting of hypo hyper proliferation
36:58when we look early on it seems to
37:01there seems to be hypoproliferation.
37:03So if we measure the B cell percentages
37:06in the homozygous mice in the
37:08setting of pre leukemia it's actually
37:10depressed compared to wild type.
37:12So what is going on?
37:14How is this kind of hypoproliferation
37:17turning into hyper?
37:18And so to kind of gain some clues,
37:20we really focused on these pre
37:22leukemic mice for which we collected
37:24B cells and started off by just
37:27looking at gene expression profiling.
37:29And it became quite evident that there
37:31was quite a few different altered
37:33pathways including cell cycle checkpoints,
37:36MIC targets, DNA repair.
37:38And looking close more closely,
37:40we could see that this was related
37:42to either reduction in proliferative
37:44capacity as well as there was increased
37:48G1 checkpoint activity after mitogenic
37:51stimulation and increased apoptosis.
37:53Now these alterations in in cell
37:58cycle could be due to cell stress.
38:01So we started to look at the question
38:04of whether or not there was changes in
38:07oxidative stress and in fact using a
38:10Mitosox assay in our homozygous mice,
38:13we do see evidence both at baseline
38:16and with stimulation that there
38:18is increased enhanced oxidative
38:20stress which is supported by the
38:22fact that when we use the inhibitor,
38:24so that pro oxidant we actually see
38:27that the RPS 15 mice are more sensitive
38:30to this inhibitor than the wild type.
38:34Now because of the cellular stress,
38:36does this actually can this actually
38:39support acquisition of genotoxic injury?
38:44And in this case,
38:46we were able to use gamma H2 AX and
38:48we see in the homozygous mice that
38:50there is increase in gamma H2AX.
38:53And as we started to,
38:55there's a lot of westerns that
38:57I could have shown you.
38:58But suffice it to say that through
39:01examination of the mutant mice,
39:03we do see impaired cell cycle
39:05checkpoint response to DNA damage,
39:07impaired response signaling,
39:08abrogation of ATM and CHECK 2
39:11signaling and heightened intrinsic
39:13aberrant DNA damage response.
39:19And Despite that, there's also
39:22increased proliferation signaling.
39:24So one of our highest hits in
39:27our gene expression was ZAP 70,
39:28which has relevance to CLL.
39:30So we see that here.
39:32And there's also enhanced ABCR signaling.
39:34So definitely a balance between
39:36different forces at play.
39:38Going on, our next question was that is
39:42seeing these different sort of phenotypes,
39:44since this is a ribosomal protein,
39:47is there actually alteration?
39:49Is there effects of mutant
39:52RPS 15 on translation?
39:54And so we asked could RPS 15
39:57mutation cause ribosomes to
39:58preferentially translate certain genes?
40:01Could the mutation cause ribosomes
40:03for example to stall at specific
40:06protein coding sequence motifs
40:08interrupting translation of certain
40:10genes or could it read through
40:12stop codons and then result in
40:14misfolded and degraded proteins?
40:16And so for this we performed
40:19A ribosomal profiling.
40:20And when we started to look at
40:23whether or not there was evidence of
40:25differential translation efficiency,
40:26there were certainly many genes that
40:28were appeared to be have enhanced or
40:32depressed translational efficiency.
40:34And as we started to look at the
40:37pathways that were impacted,
40:39these included many of those pathways
40:41that I already talked to you about
40:43in the pre leukemic setting.
40:44So cell cycle, MC target,
40:45cell cycle checkpoints and DNA replication.
40:49And specific examples that we could
40:51see were genes that are have very well
40:53known roles in these different pathways.
40:58We were able to support this this
41:01kind of ribosome Riboseek analysis
41:03by looking at protein expression
41:06and we can confirm that what we saw
41:10as as having depressed translation.
41:12So the GPX one we could actually
41:15confirm at the protein level for
41:19GPX 1 and O2O2 four and increase
41:24expression in PTP 4A2.
41:28So that was actually very nice to see
41:31that linkages between translation and
41:33the the pathways that we were impacting.
41:36When we started to look at,
41:41we were also able to see evidence
41:44not only in a in a murine cells
41:47but also in a human cell line.
41:50We saw evidence of stop codon stalling.
41:54So you can see kind of a pile
41:56up here in terms of the relative
41:59position to the stop codon,
42:01but we also saw evidence of
42:03stop codon read through.
42:05And so we do see that there's enrichment of
42:09certain codons in that kind of stop site,
42:13suggesting that this is not a random process,
42:16but there's actually motifs that
42:17are kind of guiding this process.
42:20And finally,
42:21as we started to look at the
42:23leukemic B cells,
42:24we could see up regulation of Mick targets.
42:28And I'm going to just skip over this,
42:30but only to say that as we
42:31start to go through our model
42:33of what we think is going on,
42:35we do see that in this mutated
42:38ribosomal protein that there is
42:41evidence of altered translation
42:43through a couple of different
42:45mechanisms that these do initially
42:47lead to hypoproliferation.
42:49There is elevated ZAP 70 and BCR
42:54signaling as well as make activation.
42:56But in initially there's P53
42:58mediated apoptosis and cell cycle
43:01checkpoint changes that are leading
43:03to that hyper proliferation,
43:05but that over time there's acquisition
43:07of DNA damage and genomic instability
43:09that tip the balance and get us to
43:11the state of hyper proliferation.
43:13So just to conclude this part of the talk,
43:16I'll just say that again our our new
43:19work suggests that RPS 15 mutation
43:22is ACL driver and reinforces the
43:24notion that CLL has these core
43:27pathways that are affected.
43:28So I didn't go into this,
43:30but across our different mouse models
43:34we are seeing common pathways through
43:37different mechanisms that appear to be
43:40involved and current ongoing work is
43:44starting to look at the immune micro
43:46environment so that we can start to
43:48link the genotype with whether or not
43:52they're related to distinct changes
43:54in the immune micro environment.
43:58In the final slides,
43:59I just want to say that you know I
44:03think that where we're going next
44:04in sort of sort of our studies,
44:06a lot of the CLO work until now I
44:09think across the field has been really
44:11focused on the blood easy access,
44:14lots of tumor there.
44:15But I think increasingly we do
44:17need to look at these specialized
44:19hematolymphoid organs where there is
44:21a specialized immune microenvironment
44:24that we can understand better.
44:26I think that there is a
44:29priority and interest in trying
44:31to go earlier in disease.
44:32So how can we understand those early events?
44:36How can we intervene early?
44:38How can we change Natural History?
44:40We're only going to get there by
44:42understanding a little bit more about this
44:45earlier time Multiomic profiling for sure.
44:47There's so much data out there and how
44:50can we link them all together and kind
44:52of not have them as separate entities,
44:55but really trying to coalesce
44:57into kind of archetypes that we
45:01can understand spatial analysis.
45:03So our group is actively working
45:05on efforts to try to look at the
45:08architecture of lymph nodes and
45:10bone marrow to see how malignant
45:14cells are organized and also in
45:17relationship to their genotype.
45:19So their mutations and do specific
45:22clones segregate with specific types
45:25of niches and and are they organized
45:29in certain type of neighborhoods.
45:31And finally I I touched upon with our
45:33self free DNA work some of the early
45:35detection I'm going to end with the
45:37last couple slides speaking about early
45:39intervention we hope in the future.
45:41But another big part of the work
45:44that my group does is think about
45:47cancer neo antigens.
45:48And from all the genomic studies
45:51that we've been doing,
45:53we've realized that there there
45:55is the opportunity for these
45:57mutations to generate epitopes that
45:59can be recognized by by T cells.
46:01I'm not going to go into this in
46:04great length only to say that there's
46:06straightforward algorithms by now
46:07that allow us to take start with the
46:10sequencing data and identify for us
46:13what those new antigens might be.
46:16I want to say that some of our earliest
46:18work in the new antigen field and
46:20kind of setting up these pipelines
46:21were in CLL because that is where
46:23we had the data and all the tools to
46:28help us construct some of the these
46:31first pipelines that were available.
46:33And certainly our vaccine neo antigen
46:38work that Doctor Weiner alluded to has
46:41taken our group very far afield from CLL.
46:44We've gone into the solid tumors and
46:47we've been able to conduct some early
46:49proof of concept studies that such
46:51an approach of starting with tumor
46:53looking for genomic alterations and
46:55generating a personal vaccine is feasible.
46:58But I've always been super interested
47:00in trying to bring it back to CLL.
47:02And so I'm happy to say that right
47:04now we have a phase one study for
47:08patients with unmutated IGHV led by
47:11ine on and supported by Matt Davids
47:14and Jennifer Brown to study and
47:18look at the impact of this vaccine
47:21alone vaccine together with low dose
47:24cyclophosphamide as a way to kind of
47:26alter the immune micro environment
47:28and maybe address T regs.
47:30And then also a third cohort to actually
47:33add immune checkpoint blockade together and
47:39we already have enrolled in a number
47:42the first three patients we're already
47:44seeing interesting immune responses
47:46compared to our solid tumor settings.
47:49These are patients who actively
47:50have circulating disease.
47:52So is it possible to even vaccinate and
47:55generate meaningful responses when there's
47:58leukemia that's that's in circulation?
48:00And the the short answer,
48:01it seems like yes.
48:02So we're we we are actually seeing
48:04very nice brisk immune responses
48:06to actually some of our patients.
48:08So I hope you stay tuned and
48:10hopefully we'll have more to say
48:12about that in the time to come.
48:14I've tried to acknowledge
48:16folks along the way,
48:17but here's a more extensive
48:20list and I really appreciate
48:22your attention and thank you.
48:33Yes. So how do you think
48:35about driver mutations,
48:36specific driver mutations
48:38related to transformation,
48:41related to potential for
48:48these differentiation B cells
48:49leading to the clinical outcome?
48:51You listed a whole list of
48:53potential driver mutations.
48:54It's not clearly what the individual
48:56driver mutations are doing.
48:58And how you think about getting
48:59the answer to that question,
49:00if it is an important question,
49:04yeah, no, I think I skipped
49:05over a lot of stuff.
49:06And so I think one of the things that
49:08we can do when we have these driver
49:11lists because we can see whether they
49:14segregate into particular pathways.
49:16And by virtue of kind of separating
49:20out the CLL versus Richter clones,
49:22we were able to kind of identify which
49:25of those drivers seem to be CLL,
49:27which were Richter's and which were which
49:30were in a path on the way to transformation.
49:34And so some of those pathways that
49:38we see affected are related to Mick,
49:44for example, they're related to cell cycles.
49:47So this it's not a surprise,
49:50but it and metabolic rewiring as well.
49:54So there's many.
49:57So I think the drivers do help us think
50:01about the biology of what is going on,
50:04but I think that I hope that we can also
50:07use them as ways to for early detection.
50:11I don't know if this is answers
50:13your question,
50:13but
50:15yeah, I don't want the questions online,
50:17but what what do you think about the
50:19role of RGS 15 in normal CD5B cells?
50:23So it's there, yes.
50:25So the question is what is its
50:28function in thinking about what
50:29CD5B cells are doing in terms of
50:32maintenance and tolerance for
50:33example and their potential product
50:35activity and where they are,
50:36right. So we haven't looked into that.
50:38I mean I think we we have the tools and
50:44so we've we've really been focused on the,
50:47the mutant setting. Yeah.
50:49But I I think it's a really interesting
50:53question and I think that it would
50:58be a separate question where it
51:00could be like all of these different
51:04mutations that we're finding. Yes.
51:07Yes the the genes and and what are
51:08their roles in in normal business.
51:12I think I I I think you are
51:15absolutely correct. Yeah. Yes.
51:18Yeah I'm I'm getting discredited I think
51:21you said that the unmutated CLLS have
51:24a re urgent headed with the nursery
51:27so yeah so the quest so the unmutated
51:31has there are there's a far longer
51:34list of mutated drivers in unmutated
51:39CLLI see. So I guess the question then
51:42is do you think that the mechanism
51:45that's leading to the mutations of the
51:48IGH locus is unrelated to the genetic
51:52diversity that we're getting or is there
51:54a relations to them and how does that I
51:56I I think that's a great question.
51:57So the the question is whether or not
52:01how the immunoglobulin mutational status
52:07relates to kind of the genetic diversity.
52:11So. So yeah, it's been understood
52:13that whether or not the CL LS have a
52:17mutated or unmutated immunoglobulin
52:19relates to their cell of origin,
52:22kind of where are they in kind of B
52:25cell development and and whether or
52:28not those kind of normal physiological
52:30mutational processes are are present.
52:33So I think it does speak to the
52:36underlying biology of that cell
52:39of origin and probably it helps us
52:42understand why there there could be
52:45more mutations in in these different
52:50genes compared to the unmutated.
52:54So that that that would be a way
52:55to put it together.
52:59I have some questions. OK, yes.
53:03So Marcus Bosenberg asks are
53:06there any recurrent genetic or
53:08epigenetic changes in CLL arising
53:11at later time points in RPS 15?
53:16Marcus, hello, great question.
53:18We haven't actually looked at that.
53:21I think that's a great question and
53:23and probably something I should take
53:25back to the group and we should look,
53:27but we we haven't,
53:28we haven't looked at that.
53:30So thank you.
53:35One last question,
53:37there's George Miller asks,
53:39can you comment on the role of Epstein
53:42Barr virus in conversion of CLL to
53:47PLVCL?
53:51I really can't maybe yes,
53:54we we have not looked at that.
53:56It's a great question and certainly
54:02EBB does is does play a role in
54:06immortalization of B cell lines.
54:08But but I I don't have much
54:11deep thoughts about that.
54:13So my regrets. Thank you.
54:15Well, thank you very much for
54:17visiting us. Yes, thank you.