Ladies and gentlemen, it's a real pleasure and honor for me to have a chance to open this scientific symposium dedicated to the next generation in immunotherapy discoveries. As a founder of Innate Pharma 25 years ago, I had the chance to witness the advent of a immuno-oncology with there at ASCO in 2011, seeing the great data that really opened a new era in cancer treatment. But so now the time has gone after thousands of trials on on targets being tried out to look for it those next generations.
As also a founder and CEO of Innate for a while, I also witnessed the importance of the academic and industry collaboration, which is really of critical importance for a company, for biotech company like us, and I would say, at all levels, all dimensions. First, when it comes to sourcing concepts, and the very first concept that we worked on actually came out from interaction with Karolinska Institute, with University of Genoa, of course, with CIML, with our scientific founder Eric Vivier, who is with us today. But this is not only about sourcing targets and concepts, it's also about putting that to development, to preclinical and clinical development.
All along the way, we raise question where we need to go back and forth to our colleagues in fundamental and basic immunology. I just to kind of have an example, which is the development of the first checkpoint inhibitor to NK cells, the anti-KIR. On there, with this Lirilumab, and that's a discussion that we had very recently. This raised fundamental questions about the regimen. So a development question, which is connected to a very fundamental question: What is the effect of having a continuous blocking of the KIR versus some pulse with respect to education, licensing, and where all the questions dealing with NK biology? Those are the typically the kind of questions that we need to address through this type of collaboration.
The next kind of interactions are, I would say, maybe even more obvious. That's about clinical development, working with key investigators. On an important component there are specifically the investigator-sponsored trial, which help us to explore potential indications for the drug candidates that we develop. So that all the dimensions that we've tried to implement in our interactions with academia and specifically with Mount Sinai. So I'm very pleased to open this seminar here today. We're extremely grateful to the team for making this scientific event possible.
And another dimension, which I forgot, and which is also of critical importance, is, of course, the training on the recruitment from our side in the biotech of MD, PhD, and post-docs. We had something specific, maybe under the guidance of Eric, again, to have the ability to welcome, train up to doctorate, up to PhD, but also, also to welcome post-docs. So you're highly welcome to apply and visit us so that we can strengthen the link between the institution on this side of the Atlantic on a small but growing, emerging biotech there in France. So I'm extremely grateful to the team to make this event possible. Of course, extremely grateful and thanks a lot, Miriam, for making that feasible. That's really great.
You attended our twentieth event in Cassis five years ago. Well, as you know, you're part of our scientific chair, our scientific advisory board, and it's very important for us to maintain this relationship and to have this really testimony of the strong connection that we have with Mount Sinai and with the institution. Beyond that, I want to thank really the team who make this event possible and organize that. So Sandy, Jessica, Eugene, hopefully, I pronounce it correctly, and on the French side, Mathilde, Hélène, and Florence who worked to make this event. Thank you to you all. I hope you'll enjoy, and this is really my pleasure to introduce the first session with Joao Monteiro. Thanks to...
Sorry, oh, before the first session, a few words from Miriam. Sorry, Miriam. I do apologize. Thank you.
I just wanted to welcome you all at the symposium. Also, I wanted to thank the team, the French team, Jessica de Peretti, Sandy Serrano, Jill Liu, and the Innate Pharma team, who have been meeting weekly for months to organize this event. I'm particularly excited to organize this meeting with Innate Pharma and with Eric Vivier, in particular, a longtime friend and the world expert in NK cell biology, and also a professor of immunology at CIML in Marseille, and founder and scientific officer of Innate Pharma. I think that this symposium takes place at a critical time in the field.
'Cause as we continue to see tremendous success in immunotherapy across cancer and inflammatory disease, including the recent breakthrough of this PD-1/VEGF bispecific antibody in lung cancer, there is growing hesitancy among some venture funds and big pharma to reduce investment in immuno- oncology. Now, this is not the first time we've seen this happen. A pattern has emerged with big pharma to invest too much too quickly, potentially, into new advancement, rush them to the clinic, and retract them when expectations are not immediately met. So this juncture, in fact, represents a very important opportunity, a moment to focus again on the basic. And this symposium aims to really broaden and continue to build our understanding of tissue and cancer immunity and inspire novel approaches to effectively modulate immune response. So I'm hoping that today's program will engage and inspire you.
Talking to the trainees here, you know, the field is still at the beginning. There is so much possibilities, and I hope that you will enjoy the day and enjoy the discussions ahead. Welcome, everyone. All right, now I am going to introduce our dear friend, Joao Monteiro, big friend of the Immunology Institute at Sinai, who is going to chair the first session. Welcome, Joao, and thank you for being here.
Thank you, Miriam. Good morning, everybody. It's a really great pleasure to be back here at Mount Sinai. As Miriam said, I'm a friend of the Immunology Department Institute, have been with you guys many times, and I'm particularly excited to be here today to talk and to learn more about what the real next generation of immunotherapy discoveries will be. There's a really great news for you. We will not have a talk about PD-1, at least in the first part of the symposium in the morning. Rather, we will have some of the most exciting people working on developments on targeting innate immune cells to treat cancer. Our first speaker is Dr. Pierluigi Porcu from Thomas Jefferson University.
He has been a leader in the development of biomarkers and therapies for T-cell lymphomas, and I would like to welcome him to the stage.
Thank you. Say again? Okay, thank you. Good morning, everyone. Thanks for having me. Let's see, where do I get the slides here? There we go. Thank you. It's a great privilege to be part of this symposium. I'm a medical oncologist with a translational research angle, particularly in T-cell malignancies. I'm also privileged to be the PI for the TELEMAC clinical trial. And what I'm gonna do today, I'm gonna describe some of the work we've done over the past ten, fifteen years in translational research in T-cell malignancy. Some of that when we were at Ohio State, now I'm at Jefferson. And we have two stories. One is about a kind of an older story about IL - 15, and then a more recent story on CD38.
And I'm gonna give a quick update on the KIR3DL2 targeting trial, TELEMAC. One of the challenges that we've had in T-cell malignancies is, believe it or not, how to make a diagnosis. And that is part due to the fact that there is no, there was no good clonality marker in T-cell malignancy. The pathology is extremely difficult, challenging to make, and therefore, some of the focus initially in the classification of these diseases was really try to help clinicians and pathologists to kind of make a diagnosis. So because of that, one of the things that came to be helpful was how the clinical presentation of certain malignancies really sort of was conducive to consolidating a clear diagnosis.
Even to this day, T-cell malignancies are really kind of classified primarily from their clinical presentation and the dominant clinical features. We distinguish nodal T-cell malignancies, which include the three entities described here. These make the bulk of the T-cell lymphomas that we see in the clinic. Another group of diseases are the so-called leukemic type of T-cell malignancies, including ATLL, Sézary syndrome, LGL leukemia, and T-cell prolymphocytic leukemia as well. Then there's a category of extranodal diseases that present primarily from mucosal and extranodal sites. These are enriched in innate cells. You know, the gamma delta T-cell lymphomas that almost all of them fall into this category. And then finally, there's a dedicated extra nodal type, that is, the cutaneous T-cell lymphomas. These have been known for a while.
So the current classification of the T-cell malignancies is listed here, which is from the ICC two years ago. And essentially, I call this kind of the phone book of lymphomas. And this is still evolving, and it identifies specific, distinct categories that are helpful from the diagnostic standpoint. The outcome for all of these lymphomas is extremely poor, as shown here, for the nodal types on the top two panels and for the extranodal types at the bottom. Essentially, a very poor overall survival and progression-free survival, with the exception of two subtypes. One is the ALCL part of ALCL, and the other one is subcutaneous panniculitis T-cell lymphoma. So there's a lot of unmet need in these diseases. So also what's clear today was not so clear yesterday.
This is a paper from Stanford in nineteen ninety-one. It essentially said that there was no difference in prognosis between the B-cell lymphomas and T-cell lymphomas. We know that that is not true now. So fundamentally, over the past several years, our group really kind of has addressed three major questions. What are the mechanism of mature T-cell transformation? And I emphasize mature because there's a lot of preclinical data in ALL, in T-cell ALL, but in terms of the mechanism and the preclinical models to leading to mature T-cell transformation, those have been very few. Also, are there any shared path to transformation in T-cell neoplasm? So, for example, all this incredible clinical heterogeneity and pathological heterogeneity, perhaps there is some underlying fundamental mechanism that are shared by some, if not all, of these entities.
Are some of these mechanisms targetable? To focus on this, we decided to focus primarily on two families. One is the leukemic or disseminated T-cell lymphomas, and the other one is the cutaneous T-cell lymphomas. The choice of these two was really kind of practical and empirical. The fact that it is easier to actually get patient samples from these diseases and also from the clinical standpoint, particularly for the cutaneous lymphomas, we had a lot of experience with cutaneous T-cell lymphomas. The first, I'm going to focus on interleukin two, which I call the Janus two-faced cytokine. No pun intended, as far as the signaling for this cytokine.
So first of all, interleukin 15 is a very potent adjuvant cytokine that is, of course, been studied significantly in immunotherapy, induces proliferation of NK cells, CD4, CD8 positive T cells, augments the function, rescues effective function in these effector cells, and has a very broad, strong, sustained effect on many aspects of T cell and NK cell function. At the same time, it is also overexpressed in T cell malignancies, as I see. One example is HTLV-1 induced transformation of T cells, where interleukin 15 and its private receptor alpha actually are induced by the tax oncoprotein of HTLV-1, and we have shown that the IL-15 induces a lot of expression of HDAC1 and 2, and other signaling regulators in T cells that are pro-oncogenic.
And then increasingly, interleukin 15 is showing an important role in the understanding and treatment of inflammation and autoimmunity. So interleukin 15 was isolated and discovered in nineteen ninety-four from two groups. One was Tom Waldmann and his group at the NCI, another was from Immunex. Interleukin15 was described as a cytokine, a new cytokine that was inducing the proliferation of interleukin two dependent cells except in the presence of IL-2 inhibiting antibodies, and so was different from interleukin two. And then a lot of groups, including Mike Caligiuri at Ohio State, showed that IL-15 was a potent stimulatory cytokine for NK cell proliferation.
If you look at the right panel, IL-15 has a number of effects, including, you know, specific effects on certain lineages of cells, both innate and adaptive, and then a lot of sort of functional effects on these cells. The primary cell that are producing IL-15 are macrophages, antigen-presenting cells, and then epithelial cells as well, and Langerhans cells in the skin. So, IL-15 can signal through a heterotrimeric receptor that is made of two common chains called beta and gamma, and then a private alpha receptor is called CD25. Interleukin fifteen can signal both in trans, as shown here, as well as in cis.
And some of the sort of signaling is in the and the outcome of the signaling can be different depending on how whether it's presented in trans or cis. And these are the canonical sort of activating pathways that are activated following stimulation with IL-15 that induce JAK3 and JAK1, STAT5 and STAT3 signaling, and then ultimately leading to activation in NF-kappa B. And throughout these pathways, IL-15 induces a number of changes in the cell function and a number of activation of transcription and also epigenetic changes, including alteration of DNA and histone methylation. Over the years what has become clear is that IL-15 is overexpressed in a number of T-cell lymphomas.
We and others have shown that IL-15 is expressed in LGL leukemia. Many others as well have shown, you know, overexpression in cutaneous T-cell lymphoma. A group has shown that IL-15 is actually very important in the development of refractory celiac disease, eventually leading to a type of intestinal T-cell lymphoma called EATL. And as I mentioned, Tom Waldmann's group at the NCI has demonstrated that IL-15 is activated by HTLV-1. So starting with the IL-15 story, which goes back all the way to 2012, actually, Mike Caligiuri had developed, with the intent of studying the function of IL-15 , had developed a transgenic mouse that overexpressed mouse IL-15 under the HLA class one promoter. So the IL-15 is expressed in all tissues.
Looking at the function of these mice and how they developed, they developed okay, but very quickly, about a third of the mice developed a very aggressive type of LGL NK cell leukemia, which is fatal. Anjali Mishra, who was a postdoc at the time in Michael Judah's lab and now is a faculty at Jefferson, essentially described in this paper kind of the mechanistic features of this LGL leukemia. But also notice that the two-thirds of the mice, so, you know, the mice that did not succumb to the leukemia, actually develop a chronic T cell lymphoproliferative disease that has some features in similarity to cutaneous T cell lymphoma. And this was then described four years later in this paper. These are the mice.
The mice develop essentially kind of alopecia, which is expected for lymphomas that kind of affect the skin. And these lymphomas are, for the most part, CD3 positive and CD4 positive, although we also see CD8 positive lymphoma. These are spontaneous mouse lymphomas. Histologically, as shown in the panel on the right, these lymphomas are epidermotropic. And when we look at gene expression, actually, they express a lot of besides the T cell markers that we expect. They also express some of the ectopic genes that we see expressed in Sezary syndrome in particular, like plastin-3, twist, Twist1, and CCR4. And this disease is transplantable in secondary mice, particularly if you take cells from the skin, not so much if you take cells from the spleen.
Anjali went on to characterize in great detail how these cutaneous T cell lymphomas essentially develop in these mice. First, here on the left, show that you know if you look at a patient, these are patient samples divided by stage of patient with CTCL. You can see that there is a kind of a stage-dependent increase in the expression of HDAC1 and HDAC6, and particularly HDAC1 is a known oncogenic HDAC. The other panel here in B shows that if we took normal T cells and we expose them to IL-15, over just a period of a few days, just two days, they overexpress HDAC1 and HDAC6.
And so when Anjali went and did a ChIP-seq on this, one of the things that she observed was the fact that if you look at both normal CD4 cells and normal CD4 cells treated with IL-15, you see the same pattern of a kind of a decreased occupancy of HDAC1 at the promoters of both HDAC1 and HDAC6. And this is shown, this is for HDAC1, and this is occupancy, and this is for the promoter of HDAC6. And so by doing this, essentially, she proposed a model where this kind of a negative feedback loop, which normally regulates the expression of HDAC1 and HDAC6, which following the lower occupancy of the promoters, leads to upregulation in HDAC1 and HDAC6.
I don't have the data here, lack of time, but what Anjali did show that part of this reduced occupancy is directly related to increased methylation of the promoter of those two genes. Something else that Anjali did was looking at the regulation of the expression of miR-21, which is an oncogenic microRNA, shown here. And once again, the pattern is that you look at the, you know, the expression in normal CD4 cells compared to a CTCL patient, there is increased, you know, expression of miR-21, which is also shown here in terms of the occupancy at the promoter. And the same happens when you actually stimulate normal CD4 cells with IL-15.
So in vitro, exposure of the cells to IL-15 reproduces a lot of the same features that you see in the T cell lymphoma patient cells from CTCL patients. So what she did as part of this study was also looking at how the inhibition of isotype-specific HDACs really sort of performed in these mice. And, you know, she had specific inhibitors for HDAC1 and 2, specific inhibitors for HDAC6. And what she observed was to actually have the optimal control of the disease, you really needed to have both inhibition. Every single isotype-specific inhibitor was insufficient to control the disease. This goes along also with the fact that many essentially almost all of the HDAC inhibitors that we have in the clinic are not isotype specific.
This also provides kind of a model for how HDAC inhibitors work in CTCL. The next step is it possible, you know, having demonstrated that IL-15 is really kind of a driver of T cell lymphomagenesis, certainly in CTCL and also in LGL leukemia, is it possible to inhibit IL-15? Ways to go about inhibiting IL-15 have been explored, primarily from the laboratory of Tom Waldmann. And essentially, this led to the development of two antibodies. One is a humanized Mik-B eta- 1 antibody that blocks essentially the beta chain and the interaction between interleukin-15 and the beta chain. This antibody was tested in phase I clinical trials in LGL leukemia, and this is more than ten years ago, and unfortunately, it was in...
not effective, although it was perfectly safe, but, not effective. So this led to the development of another, IL-15 inhibiting drug called, BNZ-1, from a company called Bioniz, which now has been acquired by another pharmaceutical company called Equillium. And, this took a while for, for this drug to be tested clinically. The mechanism is different from, MEC beta-1, because BNZ-1 actually inhibits the, interaction with, the gamma, common gamma, gamma chain....So the development of BNZ-1 has really kind of unfolded over the past several years, five, six years, and, you know, was presented-- data on CTCL were presented twice at ASH. This, the 2020 is the last presentation, and, the drug is effective.
I don't have all the efficacy data on this particular study, but clearly there were responses, and it was safe, and the dose could be escalated. But these results have not been published yet. It's not clear to me when they will be published. What got published was the sort of LGL leukemia arm of this trial, which was led by Tom Loughran and Jonathan Brammer, who is at Ohio State. Jonathan, I recruited Jonathan from MD Anderson to Ohio State many years ago. And so this paper actually was finally published just about a year ago, and I'm gonna show some of the data on this trial. This is a phase one, two. There was first a dose escalation, followed by dose expansion. There were three treatment periods.
First, a four-week initial treatment, then a three-month extension of the same weekly dose, and then a long-term extension for responders all the way to progression. The primary endpoint was MTD. There were a bunch of secondary endpoints, including a very rigorous PD analysis, which I'm gonna show, and for those levels, and this is a drug that's given intravenously on a weekly schedule. In terms of patients enrolled, there were 20 patients enrolled, 17, sorry, 13. Sorry. 13 were treated, seven were untreated. And this is the median age. 14 were enrolled in the dose-finding cohort, the 6 in the dose expansion cohort. They were pretreated. I wouldn't say that they were heavily pretreated. Well, 35% of them had STAT3 mutation, which are canonical in LGL leukemia.
All patients completed the first four weeks. Seventeen went into the three-month extension, and of these, sixteen completed it. Four patients went on to the long-term extension part of the study. There were no dose-limiting toxicities. The MTD was not reached, so this is a very safe drug, at least short term. The efficacy was kind of disappointing, only 20%, and you have to understand that the, you know, response criteria in LGL leukemia are really primarily based on, hematological parameters, besides the, the number of, leukemia cells in the blood. So, they are, you know, high bar in terms of, obtaining objective response. Nonetheless, 20% is not exactly, sort of a home run.
The median time of response was eight weeks, and the median duration of response was eight months in responders. Okay. So this is a PD analysis on this cohort, and I think that this is really what the value of this paper is based on, kind of the biology and the PD analysis. Because Jonathan and Anjali were really able to kind of analyze the apoptotic responses in vivo in these patients. And they, you know, panel A shows the flow, the Annexin V flow cytometry of both CD4 positive T cells from patients and LGL cells from the same patients, treated with BNZ-1. And you can see kind of the apoptosis in the LGL cells, but not in the CD4 cells. So this clearly...
And the same is true for, you know, cleavage of caspase-9. And then the lower panel show that the kind of the longitudinal analysis of the apoptosis, both in the CD4, normal CD4 T cells in the left, on the left, and the leukemia cells on the right. So I think that this really kind of proves, first of all, that, you know, the LGL cells are dependent on IL-15, in vivo, which is the first time that this has ever been demonstrated. Something also really interesting is the fact that, you know, there was no difference in sensitivity to BNZ-1, according to the presence or absence of STAT3 mutations. And this goes to show that even if you have activated, gain-of-function STAT3 mutations, that is not sufficient.
The cells are still dependent on, and signaling from IL-15 or possibly the T cell receptor. So our conclusions are there's strong evidence for a role of IL-15 in T-cell lymphomagenesis, particularly in LGL leukemia and CTCL. Inhibition of IL-15 signaling alone is probably not sufficient for optimal antitumor efficacy. And activating mutation in STAT3, JAK, and T-cell receptor pathways, signaling molecules are definitely not sufficient for transformation, but they do amplify the signaling from the T-cell receptor and IL-15. So I think the combination of signaling inhibitors, potentially at receptor level and then signaling cascade, may be approached. Steven Horwitz and Alison Moskowitz have a very nice trial at Memorial, looking at ruxolitinib in LGL leukemia. The response rate with ruxolitinib are significantly higher than 20%.
I think they're about 55%, with a good number of complete responses, but they're not perfect. I'm thinking that perhaps kind of a combined approach may be valuable to be investigated. The approach may also be different depending on the T-cell malignancy. The second sort of target I'm going to discuss is CD38, which is shown here, was identified many years ago. It's as a T10 marker in the workshop for the cluster of differentiation. One of the earlier markers for T-cells, and it's expressed a low level in many normal leukocytes, has a dual function as an ectoenzyme. Essentially, it's kind of a cyclic ADP-ribose hydrolase and also a surface receptor.
It's overexpressed in a number of malignancies and regulates a whole host of cellular functions, including activation, adhesion, migration, differentiation, and proliferation, seem to be very involved in the response to the immune microenvironment and has not been studied at all in T-cell lymphoma. Although as many of you know, it's kind of a standard of care drug now in multiple myeloma with multiple generations of anti-CD38 drugs. What we did was looking at the biobank of T-cell malignancies that we had at Jefferson. And this was work done by an MD-PhD student called Colleen Isabelle. And this is published, was published last year.
Essentially, we looked at this whole variety of T-cell malignancies, focusing on the abnormal T cells, and we look at what percentage of those cells were CD38 positive, and we compared it to normal T cells, and there was a kind of a trend for a higher number of T-CD38 positive cells in all of them, but statistically, it was primarily shown in peripheral T-cell lymphoma, NOS here on the left, and the T-cell prolymphocytic leukemias.
If we look in aggregate at the MFI, the intensity of expression CD38 in aggressive mature T cell neoplasms versus B cells, the T cells seem to have a higher expression, higher intensity, although it doesn't quite reach the same intensity as we see in multiple myeloma. So what Anjali, in collaboration with Neda Nikbakht, who is a physician scientist at Jefferson as well, did was looking at single-cell sequencing from skin of patients. And this has also been done by a group in France, just published recently, and the same, with the same results.
Essentially, we look at, you know, on the right, we look at skin from patient with CTCL, and we look then at the expression of the CD38 in the clusters. So the cluster with the red is enriched for abnormal T cells. And so the skin from patient with CTCL clearly overexpress CD38, and the CD38 is overexpressed in the malignant T cells. So we have a panel... Sorry, how much time? One minute. Okay. So we have a panel of, you know, for preclinical work here, looking at a variety of different T cells.
We started with a cell line called HuT 78, and then we also looked at a model with T-cell prolymphocytic leukemia, and we transfect these cells with luciferase. We also make PDX models from patients, and we then use the cells isolated from these mice for to transfect and do the xenogeneic transplants. So just showing quickly here, this is the in vivo experiments looking at the engraftment of the cell line and treatment with vehicle versus Daratumumab, which, by the way, also increases survival in, you know, in these mice. This is the experiment with the in vivo, with the patient-derived cells, with PLL, where the impact is less sort of strong, but still statistically significant.
And then one of the things we observed is that, in myeloma, a lot of the cells treated with Daratumumab eventually downregulate the CD38. And by looking at sort of in vivo observations in some of these mice, we had the feeling that the cells that were left with, you know, not expressing CD38 were behaving in a more aggressive way. So to test formally that hypothesis, Anjali and Colleen, what they did is they took some cell lines, and then they knocked out by CRISPR-Cas CD38. And this is just kind of the validation of the CRISPR-Cas assay, and then they looked at functional studies with these cells.
What's interesting is that when we actually use these knockout cells in vivo using the same transplant model in these mice, the knockout cells seem to behave significantly more aggressively than the wild type. This is shown here in gel with kind of an intravenous injection of these cells, but also in a subcutaneous injection model. These cells also behave much more aggressively. They grow faster, and this is just kind of the in vivo in the animals example of how these tumors grow. This work now, Colleen also did a lot of mechanistic work on these cells, looking at metabolism, looking at gene expression, and so on.
All of this is submitted and under review right now. So the conclusions for CD38 are that, you know, definitely is a target of therapy in both CTCL, including CTCL in the skin, and that the knockout cells for CD38 have a more aggressive behavior and increased metabolic activity. And so now we're looking at combinations between anti-CD38 and other drugs. Finally, thank you for giving me the extra time, appreciate it. So,
Keep going.
Thank you. So, I wanna just spend one minute on lacutamab, which I think many of you know. And it's an antibody that is retargeting anti-KIR3DL2, has been long in development with a very sort of well done preclinical work. And it's an antibody that kind of binds to KIR3DL2 and induces antibody-dependent cytotoxicity and phagocytosis as well. The phase I was completed, and now this is the TELEMAC. TELEMAC is a phase II trial, has you know two cohorts, fundamentally, one of Sezary syndrome and then a whole series of cohorts in mycosis fungoides. Patients, the study has completed its enrollment, and we presented the data on Sezary syndrome last year at ASH.
The data on mycosis fungoides, we're presenting Lugano, the year before. And we just heard that there are, you know, other presentation that's accepted at this current ASH. This is just to give you an idea of the responses, which are very exciting, and, you know, very, very strongly encouraging here in terms of efficacy, particularly for Sezary syndrome. I'd like to point out that the patients enrolled in this study, they all had to fail, mogamulizumab, which is a standard of care for Sezary syndrome right now. So this is a cohort of Moga-treated patients that is not really been tested in any other trial, before, and these are the responses. And so finally, I wanted to thank, kind of the, the group at Jefferson, showing kind of the Mishra lab here.
I'm lucky to actually have the ability to collaborate with this group, and share the lab with them, and brainstorm, and do work together. Another lab collaborators is from Dr. Neda Nikbakht, who's a dermatologist at Jefferson, has her own lab, is doing a lot of excellent collaboration. Thank you very much and again, sorry for going over time. Thank you. Questions now or later? Okay, thank you.
Can you hear me?
Yes.
Thanks, Dr. Porcu.
Thanks.
We have time for a few questions.
Yeah. Thank you so much, Amir Horowitz, Mount Sinai. Really beautiful work. I have a question about the LGL work that you were presenting. Do you have any insights as to their phenotypes and their sort of ability to engage TCR for actual TCR-dependent,
Yeah
... responses?
So, yeah. So the answer is yes, and perhaps. In other words, you know, I know that they are clearly CD3 positive. Particularly for the BNZ-1 phase 1 trial, they were very well defined. They were kind of homogeneously defined as, you know, terminally differentiated effector T cells. They all had CD3. Oh, and I don't think anybody tested whether the T cell receptor signaled or not, which, of course, would be something interesting to do. So no, we haven't done that.
Okay.
Yeah.
Thank you.
Welcome. Yes.
Hi, Li Chen from H.C. Wainwright. I have a question. Can you talk about the relevance of the trans versus cis signaling of IL-15, and whether the two drugs you mentioned, they target both mechanisms, both signaling mechanisms?
Yeah. So, of the two drugs that I mentioned, so MEK1 beta and BNZ-1, only BNZ-1 inhibits signaling in cis. Now, based on what I know about IL-15, the majority of the signaling in vivo, the conventional wisdom is that it's signaling in trans. Tom Waldmann has done a number of, you know, experiments looking at that. But I think that something I didn't show, but something that Angeli has, we never published this, but Angeli actually shows that the IL-15 receptor alpha is expressed in the same LGL cells, actually is found intracytoplasmically. And so she's working on actually figuring out how IL-15 signaling can occur in cis, in LGL cells.
We don't have any data ready to show for that, but I think that there may be more signaling in CIS, certainly in LGL leukemia, perhaps other malignancies than we think. BNZ-1 is the only drug that blocks that.
Any other questions?
I'm curious to know for the T-cell malignancies, do they express an exhausted type phenotype? Do they have PD-1? What do they have on their cell surface? And are the beta chains and the alpha chains common between the different malignancies, or does everybody have a different alpha-beta chain in their tumors?
Right. Yeah. All right. So in terms of the first part of the question, certain T-cell malignancies, they definitely have the markers of, if not necessarily terminal effector cells, but certainly the follicular helper T-cell phenotype, so PD-1, BCL6, CD10, CXCL13. There's actually one of the categories of nodal T-cell lymphomas now in the classification is called nodal T-cell lymphomas or follicular helper T cells, and they express a variety of those markers, so they seem to be also from the transcriptional profile, kind of a distinct group. The LGL leukemia, for sure, has the phenotype of terminal effector T-cell differentiation, you know, and perhaps Sézary, some Sézarys as well. Sézary syndrome expresses PD-1, but it's not something that is shared across all the different types.
In terms of the, there is a skewing of the T cell receptor beta family in some of these malignancies, particularly LGL leukemia, but that has not been studied in a lot of other T-cell subtypes.
Thank you.
Thank you.
Thank you very much.
... Thank you, Dr. Porcu. Our next speaker needs no introduction, is Dr. Eric Vivier. Has been a leader in the field of innate like lymphocytes for a very long time. I think he's the person that probably helped us the most to understand the complexity and the heterogeneity of all these different cell types, and has focused his attention more recently on using the power of NK cells to target cancer and other diseases. So welcome, Eric.
Thank you, Joao, for the nice words. I would like to start by thanking Miriam and all the Mount Sinai team, because this really is a team, it's a huge team, for their help in everything in what we are doing today. So, so thank you very much. It's really nice to have you with us. Okay, so I should do this. It's not a Mac, so I know. There we go. Oops! Mm. Okay, so, I will try indeed to tell you about what we are doing, and when I say we, it's a lot of we. First of all, I will update you on what we are doing at the Center for Immunology in Marseille-Luminy, trying to elucidate the heterogeneity of natural killer cells.
And then I will update you on collaborative work that we are doing between the CIML and Innate Pharma on trying to manipulate these NK cells. Just again, to illustrate the strength of the union. You know, in France, we have this sentence that says, "l'union fait la force." I think it can be translated as united we stand, somewhat. Maybe not, but I hope that you got the message. I think that these public-private joint ventures are very important for scientific knowledge and for the patients as well. So this is my disclosure slide.
So indeed, so since decades now, I'm not disclosing anymore, many years for some reason, but we have been trying to understand NK cells, what are they and how one can harness their function in therapies, in particular in cancer therapies, and we want to bring them to the clinic. So why are we even asking this question? Well, first of all, because in the context of cancer, it's obvious that the harnessing of T cells is working in many patients, but not all of them. So despite the beauty of the immune checkpoint inhibitors, CAR T cells, T cell engagers, there is still, unfortunately, unmet medical needs in too frequent conditions.
So, we started from there, and so considering that T cells could actually control tumors, we also try to remember what has been demonstrated by the entire immunology community since decades, meaning that actually, T cells are not autonomous in what they do. They need cells around them, such as cells of the innate immune system, to become effector cells and eventually memory cells. So with this in mind, with the unmet medical need also for primary or secondary resistance to existing treatment, and with the fact also that there are some innate cells, in particular natural killer cells, with natural properties to recognize cells in distress, such as tumor cells, we figured that maybe by harnessing these cells, one can have a kind of a double Kiss Cool effect.
The first one is that one can harness the anti-tumor function of these cells, particularly NK cells, but it could also boost the function of T cells. And because I would like to highlight the fact that NK cells are, are misnamed for some reason. I mean, natural killer cells are not really much what it means, but they are something else that lymphocytes being able to kill tumor cells. In particular, as illustrated here, they can produce an array of cytokines and chemokines that participate to a multi-cellular immune response here in the context of cancer. So in particular, NK cells can produce IL-1, CCL5, Fas ligand, that can act on DC1, which are so important for the priming of T cells in tumor conditions and not only.
And also, NK cells can produce gamma, which is interferon gamma, which is obviously of key importance in this context as well. So this is why we think that harnessing the function of NK cells can be important. Now, what are NK cells, in fact? So NK cells are part of a larger family of innate lymphocytes, which do not express RAG-dependent antigen receptors, such as, you know, T cells and B cells. So ILCs come in different flavors. I have no time to go through this. I would just like to focus on NK cells today. But keep in mind that ILC1s really look like NK cells or some NK subset, with a major difference is that most of the ILC1 are not cytolytic, but they are strong producers of interferon gamma. And again, these cells are fascinating.
They are still he described, I think, but this is not the purpose of this presentation today. Today, I will focus on NK cells, and here we've tried to recapitulate what could be the ten hallmarks of NK cell tumor immunity. I would just like to pinpoint some key features. So first of all, NK cells are equipped with a variety of activating receptors that you see here in green. And the beauty of this mechanism of recognition is that these cell surface receptors recognize molecules which are not expressed or expressed at very low density on normal cells. But they are induced or overexpressed when the cell is in distress. And this distress or this stress includes microbial infection, physical injuries, chemical injuries, but also tumor transformation.
There are two prototypical examples of such recognition process. First of all, NKG2D, which is an activating receptor expressed on NK cells, but also on some T-cell subsets, which recognize a variety of ligands, which again are not or very poorly expressed at the surface of cells in steady-state conditions. But they are overexpressed on tumor cells. The same is true for NKp30, which is another activating receptor exclusively expressed on T cells, well, on NK cells, preferentially expressed on NK cells, I should say, that recognize a member of the B7 family that we had the privilege to isolate years ago, which is B7-H6.
It is quite a known B7 family member, but again, this molecule is not expressed on most cells in normal conditions, but overexpressed on tumor cells. So that gives you the rationale, the mechanism by which NK cells can distinguish between normal cells and tumor cells. Also, I would like to highlight the fact that there are many preclinical models pointing to a role of NK cells in fighting against metastasis, and one knows that this is obviously a major issue for cancer patients. And finally, the beautiful work of Katy Rezvani, and she will be presenting after the coffee break, has shown that there is clinical efficacy in hematological malignancies when NK cells are used as cell products or as drug products by themselves.
In particular, because there is also an excellent safety profile. Beyond the efficacy of NK cells in this condition, there is an excellent safety profile due to the fact that NK cells do not mediate GvHD, which is a major differentiating factor as compared to T cells, so now what we've been doing recently is actually to put together a consortium of 20 labs across the world, trying to define at the single cell level what could be the NK cells in their different flavors, and this is a work of that I will present today, so this consortium is called MetaNK, and you recognize many of NK cell maniacs, in particular, Amir who is in the room with us, and thanks, Amir, for your collaboration on this one.
We've been generating and integrating single-cell RNA-seq data and CyTOF data for more than 200,000 primary human NK cells. It doesn't seem a lot as compared to what Miriam has published, for example, for other cells, but believe me, this is the largest collection of NK cells analyzed so far using these technologies. One of the key messages is that the heterogeneity of human NK cells can be at a very high level illustrated by these three subsets: NK1 , NK2 , NK3 . This is also an easy nomenclature to remember. When we look at what are the features of these three different subsets, you will recognize usual suspects of NK cell biology, in particular, CD16, which is represented here, or granzyme B.
We can also appreciate that there is other molecules such as XCL1 that I mentioned, which is exclusively produced by NK2s. And then there is this subset of NK3s that, as you can see here, exclusively produce IL-32. Well, are the only NK cell subset producing IL-32. So a more comprehensive view of what these subsets are is represented here, and they correspond to the old definition of NK cell subset, based on the density of surface expression of NCAM CD56. But you can appreciate here that this single cell immunologic definition is extending the precision of the definition of the diversity of NK cells.
So we've been also scrutinizing whether one can identify trajectory of of NK cell development taking in consideration this NK1 , 2, 3 as a, as a start. And to make a long story short, we could indeed, indeed identify that between the NK1 s and NK2 , there is an intermediate subset of NK cells that we call NK intermediate. And there is really many of this trajectory data that points to a trajectory from NK intermediate to NK1 A, NK1 B, and NK1 C. It might be interesting to consider this because when we look at the prediction of the function and the metabolism of this different subset, one sees some very obvious conclusions.
First of all, when you look at the different NK1 s, NK1 A, NK1 B, and NK1C , you can appreciate that their metabolism is increased, so to speak. I mean, the NK1 C being hypermetabolic, as seen or as suggested by the overexpression of many genes that you can't read here, but believe me. The second bucket is telling you that the cells, which are probably the most effective in terms of cytotoxicity, will be the s, as represented here in red.
And then when one looks at all the transcription factors and the target of these transcription factors, what we call the regulon, expressed by these different subsets, what is important to see, and probably the only thing that you can see from the back of the room, is that the first bootstrap, the first dichotomy, between these subsets is between NK2 s and the rest of the world... and actually points to a novel discovery made by us and other groups, showing that, in fact, there are two precursors to NK cells. NK ones are coming from the regular classical NK cell progenitors, whereas NK2 s are actually the progeny of a precursor which is common to the other subsets of ILCs.
I have no time to go into this, but this has been published very recently by us and others in Nature Immunology. Now, what one can do having in hand this gene signature is to look at the different proportion of NK1 , 2, and 3 in different conditions, starting with cancer. As you can see here, across 22 different tumor types, one can see that the diversity of NK cells varies. In particular, as you can see here, you have a lot of NK1 s in melanoma, but you have much less NK1 s in these conditions.
And we are studying right now whether this NK two gene signature is actually related to the dysfunction of NK cells that have been observed in many different tumor conditions. This is very important because when one wants to harness the function of endogenous NK cells in tumors, one should take into consideration the fitness of NK cells. As you can see, this fitness of NK cells is obviously very different from one cancer type to another one. It's not unexpected, but those are the data. Also, something which is not unexpected, but again, it's better to have the data, is to see the difference between what one can sample from the blood as compared to what is present on the tumor bed.
So this is also key when one wants to harness NK cells and have a precise analysis of the fitness of NK cells. The blood can lie unless you are mobilizing blood NK cells, and I'll come back to that. So just as a summary of what I've been showing you, and you know, having in mind what was organized in Paris during the summer, the Olympic Games, one can try to summarize what I've been telling you. The five Olympic rings can really illustrate the differentiation of NK cells across these different subsets, and with NK2 s being the sixth Olympic ring, which doesn't exist in you know in the Olympic Games just highlighting the fact that NK twos are very different from all the other ones.
With this in mind, how can we manipulate NK cells? How can we harness the function of NK cells? NK cells in the clinic can be harnessed using two different modalities. Here, one can use NK cells themself upon in vitro stimulation, gene editing, CAR NK cells. This is, again, the beautiful work of Katy, so I'm not gonna touch on that at all. I would tell you about the use of monoclonal antibody or monoclonal antibody fragments as assembled together with this NK cell engager. This is a cartoon that recapitulates the principle of the generation of these NK cell engagers. What we have been doing is actually to analyze the cell surface expression of activating NK cell receptor across many different tumor conditions.
And we realized that the most stable activity receptors expressed on NK cells in cancer was NKp46. So starting from this, we generated a battery, a very large panel, of antibodies against NKp46, and this work was performed at Innate Pharma. And on the other side, we were having access to many different antitumor antigen antibodies. And with this collection of antibodies, this is what we have done. So basically, we have combined these different elements to make a tri-specific NK cell engager, as represented here, which can actually engage NK cells on one side and the tumor cells on the other side.
The advantage of this NK cell engager, the tri-specific NK cell engager, is that it will harness the function of NK cells in a NK cell preferential manner due to the preferential expression of NKp46 on NK cells. We will also take advantage of the co-ligation of NKp46 and CD16, leading to a very strong activation of NK cell effector function. And at the same time, we are redirecting the function of NK cells towards a specific tumor antigen. The first molecule that was developed is actually this one targeting CD123, the alpha chain of the IL-3 receptor, which is overexpressed in acute myeloid leukemia. And this is a work of Innate Pharma in collaboration with Sanofi.
And the first molecule that was generated is this one, and put in the clinic, and the good news is that this molecule is active and got fast track designation from FDA as of last year, with almost 34% complete response in relapse or refractory AML, and this has been published last year. But now we realize that we could use this scaffold of a monoclonal antibody elements to actually incorporate other moiety. And the first molecule that we wanting to incorporate is actually a variant of IL-2, the mutant of IL-2, which is a non-alpha IL-2, for obvious reason, not activating T regs, and also preventing the binding of this molecule to endothelial cells that can express the CD25, the high-affinity receptor for IL-2.
So the idea here was not only to harness the effector function of NK cells, but also to expand NK cells due to the IL-2 signal. So this is what we have done, and those are the result obtained in preclinical model. So this is a molecule here that is called IPH6501, that actually targets CD20 to be developed against B non-Hodgkin lymphoma. And as you can appreciate here, when we are challenging a SCID mouse using a human tumor cell expressing CD20, when treating the mice using this tetra-specific IL-2 molecule, we can really control the tumor development quite well. Actually, much better than with obinutuzumab. I remind you that obinutuzumab has been approved in these conditions.
This efficacy within the limits of this preclinical model is associated with a very strong infiltration of activated NK cells at the tumor bed. This is what you can see here, looking at NKp46 and granzyme B. Now, we've been looking also at the single cell level on the effect on this ANKET on the different subset of NK cells, and as you can see here, you can find again the NK one, two, three, using the color code, and you can appreciate that upon just four hours of stimulation there is a massive change in all the subset of NK cells, and this is consistent with the fact that all NK cells express NKp46, so the expectation was that, indeed, there will be some outcome on all NK cells. This is what is happening here.
You can also see that there is a new subset, which appears upon ANKET stimulation at four hours, but also to some extent at 24 hours. This subset fits very well with the annotation or the gene signature of CIML NK cells. CIML NK cells stands for cytokine-induced memory-like cells. This is of obvious importance because it tells us that upon ANKET treatment, NK cells can be hyperactivated cells with also longer lifespan. I have no time to go through the detail, but you can see here that there is a common signature which is induced in all the subset at 24 hours, and a common signature at four hours. Again, this is going to be published very soon. A very important experiment, I think, is illustrated here.
I like this experiment very much, and you tell me whether you share that with me or not. So we took advantage here of an NK cell deficient mouse that we have generated at the Centre d'Immunologie de Marseille-Luminy. And then we challenged those mice using a human B-cell lymphoma, and we treated the mice using either NK cells infused into the mouse or ANKET, or the combination of both. And to make a long story short, what we could show is that we could mobilize the circulating NK cell that we're injecting into the mouse. I remind you that the recipient mice are NK cell deficient, so the NK cells have to come from the cells that we injected in the blood.
These cells, upon treatment with ANKET and not in the absence of ANKET treatment, are really infiltrating the tumor bed. That tells us that one can mobilize circulating NK cells to promote the infiltration in the tumor bed, which is very important because one can test, or so one can assess the NK cell fitness in the blood in a very easy way. Now, I would like to finish by telling you something, which I think might be a part of the future, because so far, I've been telling you that we've been trying to harness the effector function of NK cells, but I didn't tell you about the recognition process of NK cells.
I told you in the introduction that NK cells are equipped by a variety of activating receptors that can sense also tumor cells. So the question is whether we could actually harness this recognition mode. So it's a very easy experiment to perform. You take NK cells, you activate them using a tetraspecific NK cell engager, and you look at whether this recognition mode now is active, meaning, again, cells which do not express the tumor antigen that would be recognized by this binder here in yellow. And the experiment is represented here. So first of all, when you activate primary human NK cells using a tetraspecific ANKET, you can see a quite nice induction of DNAM1, NKp30, NKG2D, and also TRAIL and Fas ligand.
So giving you the basis for what we are testing, NK cells are in a better position to kill other cells, and in fact, they do. So what you do here is to take these cells, which are human NK cells, stimulated with ANKET, and you use as a target a mouse tumor cell, which is not recognized by human NK cells, as you can see here. But now, when these mouse NK tumor cells are transfected with a ligand for one of these activity NK cell receptor, namely here, MICA, a ligand for NKG2D, then you can induce a very strong elimination of these cells. And I remind you that we didn't touch NKG2D at all. We just activated NK cells through NKp46 and CD16 and IL-2Rβ, and that prime the possibility of these cells to kill through NKG2D.
You can appreciate that all the killing is blocked by an anti-NKG2D antibody. I think this is key because it tells us that when we're trying to compare, the killing of tumor antigen positive tumor cells by T-cell engagers and NK cell engagers, there is something that the T-cell engagers will never do, is to kill a tumor cell which have lost the expression of the tumor antigen, which is, targeted by the T-cell engager. In contrast, NK cells have the propensity to do that, as long as the tumor antigen negative tumor cells express ligand for NKP30, DNAM1, NKG2D, and so forth. So I think it's key because it tells us that considering the multiple evasion mechanism of cancer cells to immunity, that gives another chance to NK cell engagers to be effective, in real.
So this is the portfolio of NK cell engagers that are generated by Innate Pharma in collaboration with Sanofi, or proprietary to the company. You can appreciate that we are also addressing solid tumors using B7-H3 as a tumor antigen. And if I have another minute, I would like to update you on what we have done before the NK cell engagers, meaning trying to unleash NK cells, promoting novel immune checkpoint inhibitors. And what we have been doing is actually to generate an antibody called monalizumab, which is blocking the interaction between NKG2A and HLA-E. Why? Well, because NKG2A is a prototypical ITIM-bearing cell surface receptor, which is thus an inhibitory molecule when engaged with HLA-E, its ligand.
Two important features, NKG2A is expressed on NK cells, but also on CD8 T cells, and in particular, on PD-1 plus CD8 T cells as a tumor bed. And second feature, which is very important, in contrast to classical MHC class I molecule, which has a tendency to be down-regulated at the tumor bed, HLA-E is overexpressed. So the likelihood of interaction between HLA-E and NKG2A is pretty high. So when one wants to totally unleash or more completely unleash, the immune response as compared to treatment with anti-PD-1 or anti-PD-L1, the combination with Monalizumab is an obvious possibility. So this is what has been done in collaboration with AstraZeneca in non-small cell lung carcinoma, with patients at this stage three, non-resectable, non-small cell lung carcinoma. So they receive chemo and radio, then wash out, then consolidation using durvalumab. This is in green here.
I remind you that durvalumab was approved in this setting. We can appreciate that the combination of Durva and Mona is actually much better than the use of Durva at all, leading to a phase three which is called PACIFIC -9, and which is undergoing as I speak. Okay, I'm gonna stop here. Thanks for the extra time. I see. Thanks a lot. I would just like to finish by thanking all the people involved in this work. So first of all, the people at Innate Pharma, they will recognize themselves, I hope, and also people at Centre d'Immunologie de Marseille-Luminy. Thank you very much.
Thank you, Eric. We have time for a few questions. I see a bunch of hands. I will-
Eric, really beautiful work, as always, and I feel like every time I listen to you, I learn something new. So your recent work that is in press with the... I have two questions related to that. Do you think patient NK cells, patients with active disease, would be just as activated by the ANKET as, like, normal peripheral blood NK cells? I don't know whether for your experiments you used NK cells from patients or not. And then the trafficking data is really beautiful as well. Do you think you could induce a similar trafficking into solid tumors as in a lymphoma model?
Yeah.
Why? Do you think it's just the targeting of the antigen, or are you changing the chemokine receptor profile of your NK cells?
Yeah, yeah. Very, very good question, Katie. Thanks a lot for that. So we've been starting to use NK cells from patients. And in particular, this is... Well, I don't... I can't remember if I showed it or not, but we've been using, in particular, patient cells from patients with BNHL, and we could see that indeed, these NK cells could be harnessed using ANKET. Actually, we've been comparing with T-cell engagers, and we've seen better activity with the NK cell engagers as compared to the T-cell engagers. Now we are, you know, developing this, exploring the NK cells in other tumor conditions. So obviously, the solid tumor is the grail. This is where we want to go. We don't have any definitive data so far, but again, this is what we want to do.
In particular, we want to use these tissue slice assays, in which one can, you know, make slice of tumor tissues and look at, upon treatment or upon NK cells plus treatment, in particular, autologous NK cells, whether one can see migration and activation of these cells. That's for your point. So this is in progress, in short. As for, the mobilization of NK cells, of circulating NK cells, yes, using the single-cell RNA seq data, we could see that upon ANKET treatment, there is a major change in the homing code of, expressed on NK cells. Most of the S1P receptors, S1P5 in particular, but also S1P1 to some extent, and also the chemokine, receptors are modulated.
We think this is the basis for the enhanced patrolling of NK cells from the blood to different tissues, hopefully tumor tissues as well. I think, you know, it's expected that this will be tumor dependent.
... Great data and great presentation. I work in clinical biomarkers, and we always see NK cells associated with response. I mean, almost like these are predictor biomarkers when I work on solid tumors. This, in this case, it's melanoma and with anti-PD-1 and anti-CTLA-4. Now, we have lots of trials on these two markers. Do you think we know long-term effects of these on NK cells?
Long-term effect of checkpoint inhibitors-
Correct.
-like anti-PD-1, anti-PD-1, anti-CTLA on NK cells?
Correct. Yeah.
Not that I know of.
Thank you.
Short answer. Maybe someone knows, I don't know, but I don't. Yeah.
Thanks.
Thank you, so in your construct, the tetra binding construct, you have something that binds to NK cells, and then you put an IL-2 that is skewed out of the alpha chain. But when you activate NK cells, you would express the alpha chain. Have you tested a regular IL-2, one that's not skewed away from the alpha chain, that would have the higher affinity binding?
Yes, we did, but we never wanted to use it because of the toxicity linked to IL-2. You know, IL-2 has some toxicity, so even though we would put the regular IL-2, the non-alpha IL-2, into a tetra-specific scaffold, we think that we would be less safe than with a non-alpha IL-2. So we tested it, but we never wanted to develop it in the clinic.
You had toxicity in the mice, so you got low body weight loss or-
This is expected. Actually, I'm not sure that we even tested it in the mice because we, what we tested is the toxicity of the tetraspecific once mutated on the alpha chain, in the mouse, but also in the monkey. And as a dose that we've been using, we didn't see any signs of toxicity. Obviously, if you increase the dose of this tetraspecific ANKET, even with a non-alpha mutation, you will have some toxicity. But the good thing is that we have a nice therapeutic index with no toxicity and good efficacy in both non-human primate and mouse using the non-alpha IL-2 variant. Does this answer your question?
A little bit.
We can take it offline then. Okay.
Oh, Mario? No.
I'll let somebody else ask a question because mine was somewhat similar to Tony's, so.
Okay.
I'll come back.
Uh, Amir.
Yeah. Thanks so much for that inspirational work. I'm glad you took the extra time to talk about unleashing based on the inhibition, but specific to the tetra-specific agents, when you're activating the NK cells, everything around it has a tendency to get activated as well as, you know, including tumors. When you show the activating receptors not being engaged, come up-
Yeah.
What else happens with the tumors?
NKG2A can be upregulated as well. One can think of, you know, the combination of this tetra-specific with monalizumab, but, you know, the tri-specific has already single-agent activity, so we are testing the tetra-specific for single-agent activity. That doesn't preclude us to try a, you know, different combination afterwards, at least at the preclinical level, in particular, by blocking MHC class one receptors. No, you're right. Yeah, sure. Sure. So, Mario.
So, very similar to Tony's question, when you do the co-culture of the tetravalent ANKET-
Yeah
... without antigen and with NK cells in vitro, do you see a lot of cytokine production? Because it's almost like a cis-targeted IL-2 to the NK cells.
Yes.
So theoretically, you should, and NK cells in humans are very sensitive, so I would expect you would see a lot of cytokine production without tumor antigen in an in vitro culture with NK cells.
Yeah. So we do see some, which is increased when you have, you know, the binding to the tumor antigen, but we do see some. But, if your point is related to the safety profile of this, NK cell engagers, again, when we look in the non-human primate with a molecule that we're injecting to the patient, to a serum cytokine, we see between one hundred and one thousand times less as what is seen with the T cell engager targeting the same tumor antigen. So by far, the production of so-called inflammatory cytokines induced by this tetra-specific ANKET is less than what is seen with the T cell engagers. The reason for that could be that the number of cells which are activated is much less because you have, you know, ten times less NK cells as compared to T cells.
It might be also that maybe the cytokines are not exactly the same. We've been looking at all the classical cytokine, IL-6, IL-10, and others, but this is what we see. So, this is why we think that we are on the safe side in terms of safety. Okay, I think we have to move on, right?
Let's thank Eric. Okay.
No coffee break. No, I'm kidding.
I just want to go back to this, expansion of NK, right?
Yes.
Because I think everyone focuses on that. Like, if they don't expand enough, then they won't kill enough.
Yeah.
Right. So do you think you can play with cytokine further? Like, how much can we push that expansion locally?
You mean can-
And do you think this is the limit. It's a limitation of NK cell therapy, or you think we don't need that big?
Absolutely.
I think this is the question that the field...
It's difficult to answer your question because it has to be data driven, right? So, to your point, we are not only playing with IL-2, but also with other cytokines, okay? And I can disclose that we are working with IL-15, IL-18, type I interferon. And we do see some very interesting results for that. So we are going to try to compare and exactly to your point, to try to see what is the best harnessing of NK cell possible in terms of NK cell infiltration at the tumor bed. It's not only the expansion, it's also the fact that they are going to the tissue and expand there, okay? Or expand and then go to the tissue. I mean, both ways are okay.
Trying to compare the different flavor of cytokines in terms of activity but also safety profile. This is in progress.
All right, I think we should continue this conversation over coffee.
Yeah, sure.
We'll be breaking now, and we should be back here in fifteen minutes at 11:40. Yes.
... Hi, everybody. We'll be starting shortly. If you can take your seats, please. All right, welcome back from the break. I would like to invite our next speaker, who is Dr. Lorenzo Falchi from Memorial Sloan Kettering, and he will be talking about NK engagers for the treatment of B-cell lymphoma, preclinical development and clinical applications. Welcome.
Thank you. Thank you so much. Including forgetting my name right at the first attempt, which is not common around here. Hi, everyone. I'm Lorenzo. I work at MSK, in the lymphoma service, and I'm a medical oncologist, primarily clinical investigator. I'm just very curious about the translational component of my research, so I'm very excited to be here. I dedicate most of my time to the development of multispecific immune cell engager therapies. Specifically, I've been doing a lot of work with the T-cell engagers and we were the first in Memorial to use epcoritamab and glofitamab for non-Hodgkin lymphoma, and among the first to use mosunetuzumab.
To date, we have probably the largest portfolio and the largest enrollment in clinical trials for those drugs or among those institutions. These are my disclosures. Some of these are relevant to today's talk. Really, what I want to do today is a couple of things. One is contextualize the development of tetraspecific ANKET engagers for the treatment of B-cell and Hodgkin lymphoma in the setting of other immunotherapy-based modalities, particularly T-cell engagers, and Eric alluded to that before a moment ago, before the break. Also give you some of the preclinical development data for IPH6501, which is essentially going to show you all the data that everybody was asking in the Q&A questions, in the Q&A section.
Hopefully, you'll have a visual of what we were talking about a moment ago. And then just allude to the clinical trial that's launched with this molecule. Not much, obviously, to share there because the trial just started. But, I've always made the argument that immunotherapy has revolutionized the treatment landscape of B-cell non-Hodgkin lymphoma in multiple ways. And if you call those therapeutic milestones, then the first would be monoclonal mono-specific B-cell targeted CD20 antibodies, namely rituximab, obinutuzumab, ofatumumab. The second, about a decade later, the development of autologous CAR T cells, the commercially available ones are targeted against CD19.... And then, more recently, a few years ago, the development of bispecific antibodies targeting CD3 and CD20, primarily, those are the commercially available ones.
All of these treatment modalities have certainly revolutionized the treatment landscape, and I think it, you know, it's undeniable that there are multiple, you know, patients, countless patients, I would say, that are alive today because of these treatment modalities and wouldn't have been alive today weren't for these treatments. But my job today is not so much to sort of congratulate ourselves on those successes, but to look at the flip side of that coin and essentially focus on what the limitations are for these treatments. First and foremost, for monoclonal monospecific antibodies, albeit revolutionary at the time, we have to acknowledge these drugs have quite poor single agent activity.
In fact, regardless of the clinical setting, including in indolent lymphomas, where they play a crucial part, they're just not as effective a single agent. For CAR T cell and bispecifics, the limitations are quite similar, albeit potentially to a different scale, and primarily pertain to toxicities, cytokine release syndrome and neurotoxicity, chiefly, but also efficacy. Again, multiple patients have been treated with success with these treatment modalities, but the reality is the majority of patients eventually relapse and unfortunately succumb to their disease. For CAR T cells specifically, another big issue is scalability, manufacturing, access, financial toxicity, and we could talk for hours. And so focusing, though, on the efficacy aspect and on what, the...
What the reasons could be for the therapeutic failure for these treatments, Sanders, Vardhana, and I last year put together this review, where we tried to kind of, you know, think critically about what the mechanisms of resistance to bispecific antibodies could have been. And we thought that you could group them in three categories. The tumor intrinsic ones, that's easy. You lose the antigen. Eric talked about it a moment ago. If you lose the CD20, efficacy of these drugs is fatally impaired. The second category is T cells intrinsic. For simplicity, think T cell exhaustion or T cell dysfunction.
But then there's a whole category of mechanisms of resistance that probably are at play in patients who have a CD20 positive, you know, recurrence of disease, where you could reactivate those T cells because there's nothing wrong with them, but they still have a recurrence, and so what is at play in those patients, I think, remains to be elucidated, but I think it's fair to say that that's an open field for investigation, and other immune players in the tumor microenvironment could be harnessed to potentially reverse those states of active disease progression and resistance to immune therapies, and so today, we're talking primarily about NK engagers because that's certainly one of the players, so the fact that NK cells are instrumental for therapeutic success when talking about immunotherapy, I think has been known for a while.
Here's just a little piece of work that we had done when I was occupying myself with CLL during my time at MD Anderson. It. We had a trial with lenalidomide and Ofatumumab in patients with recurrent CLL, where we saw the baseline and over time numbers and functionality of NK cells was associated with a greater likelihood, not only a complete response, but also durable response. So that's the first hint. The other is, and we've learned this, NK cells, when engaged by a single receptor, are potentially underleveraged, and the engagement of multiple receptors seems to be crucial in really getting the most out of their cytotoxic potential. And this is what the whole ANKET program really is all about, and we'll talk about it in a second.
And thirdly, as Eric alluded to, activating NK cells because the array of cytokines that are produced is simply different than when you activate T cells, so would the toxicity profile expected to be different. And then, fourth, you know, there's already signals that directly leveraging NK cells for therapeutic purposes has been very promising, and Dr. Rezvani will talk to the CAR NK setting in a moment, but there's other products as well. Enhanced NK cells in the cell therapy space particularly, I think there's already some promising results. But today, what I'm focusing on is, as I mentioned, the T cell engagers, sorry, the NK cell engagers, here on the top right, exemplified in the cartoon.
Eric's already talked about this, but this was the sort of seminal kind of molecule, so to speak, the one that first hit the clinic, the trispecific ANKET molecule that co-targets CD123 on the AML cell and CD16 and NKp46 on the NK cell. This is a molecule that has. It is proving itself in the clinic, and certainly, we'll hear more about it at future meetings. But that's sort of proof of principle that, yes, you can engage NK cells in an enhanced way and produce clinically meaningful results. As Eric mentioned, going from trispecific to tetraspecific. The construct is more complex, and it includes an interleukin-2 variant peptide that we'll see in a moment really imparts a lot of efficacy.
Why, interleukin-2? Well, the most simple reason is because there is complementarity, and I would say molecular synergism between engaging the dimeric NKp46 and CD16 receptor and the IL-2 receptor beta, sorry, IL-2 beta and gamma receptor and the NKp46 and CD16 receptors. Because the former signals through JAK STAT, it eventually inducing proliferation of the NK, and the latter through a Syk PI3 kinase and resulting in an NF-kappa B mediated NK activation, both of which then result ultimately into cytotoxicity.
But crucially, the dimeric receptor is expressed on NK cells and partly on CD8 cells, whereas a trimeric receptor that in addition to the beta and gamma has the alpha component, which is known as CD25, as you know, is expressed in T regulatory cells, which are generally speaking pro-tumor immunosuppressive cells that are often found in the cellular milieu of the tumor bed. And so by abrogating that alpha binding, the hypothesis was that this tetraspecific ANKET product could be more NK specific or NK preferential compared to just a pan NKT activator. And we'll see the data to that point. And so the first challenge was to prove that this product is active in vitro.
That it was, but not only that, what Eric and his group has really shown is that the particular tetraspecific structure is really what you need to have the maximal stimulation and activation proliferation and eventually cytotoxicity from these NK cells. You can see it here on the top in this phosphorylated STAT5 assay, where really the maximum phosphorylation. Again, this is an IL-2 sort of dependent signal, and so you can see it here. If you look at the four different constructs that they tested, where you kind of selectively take out one piece or another, none is as effective as the full tetraspecific molecule in activating the signaling cascade. Similarly, and this is differently color-coded, but same concept.
If you take out any of these components, in particular here, the Fc that would target CD16 or an isotype control that would have targeted NKp46, you don't quite get the same effect in terms of both cytokine production and direct, you know, specific NK-mediated cytolysis. On the flip side of that, not only the B cell depletion is real and is powerful when you use the full tetrameric structure, but there's not a lot of CD20 necessary for this—at least in vitro—for this molecule to be activated and deplete B cells. You can see it here at different levels of B cell expression. On the other hand, when CD20 is completely absent, the molecule is essentially inert, and this is very good to know.
This is a similar set of experiments were done in the T cell engager development. It was very good to know that if you don't have the target, these molecules just aren't active. And in the clinic, that's important. You can see, for example, in T cell engagers, that after cycle one, there is zero toxicity except for the immunosuppression, and that's because there's very little to no target to really hit and stimulate the cytokine production. Moving on to the mouse models, some of the same findings were confirmed reassuringly in this particular model. Eric's already shown this. This is a mouse that retains NK production but has no B or T cell production.
After a single injection, you can see that compared to vehicle, and you've seen it before, compared to obinutuzumab as well, the tumor control at day nine is much more pronounced, indicating actual activity in a 3D model. In a syngeneic mouse model that has more immune function, and looking at the ability to clear lung metastasis from mice engrafted with a B16 CD20 positive cell line, you can see that in a dose-dependent manner, one single injection of the IPH6501 was able to clear most if not all the lung metastasis.
Again, we're talking about one injection at doses that may or may not be mirrored in humans, but it just as a testament to the power, the immediate power of the drug. In non-human primates, blood assays, these were animals treated weekly for three times, so this is a month-long observation time. You can see that in a dose-dependent manner, again, very early on, at day three, you can observe a very substantial, almost complete, depending on the dose, B cell clearance. In parallel to that, the levels and the activation status of NK cells would increase proportionately. This is mirrored in a tumor in lymphoid tissue that were stained for NKp46 at day 32.
After the three doses were administered, you can see an enrichment in NK cells in the tumor bed. Again, speaking to the point that Eric was making before about the homing of NK cells, and this sort of like recall of those cells into the tumor bed. Ultimately, which is what we're more interested in, we're interested in everything, but we're more satisfying in seeing in lymphoid organs that were ridden with metastasis of CD20 positive cells. These are largely cleared, and this, again, is just three doses. You have to imagine this treatment will be given for a longer time. Very powerful clearance of CD20 positive tumor cells.
And then finally, in humans, obviously, these are ex vivo studies, so in healthy individuals, I think part of this was shown before, you can see that the co-culture with IPH6501 resulted in a very effective clearance of B cells gated here for CD19 , that you don't see if you don't put the CD20 binding moiety. Of course, that's reassuring, of course, a sanity check. But also you can see here in the bottom two graphs where the ability of clearing B cells from an NK engager like IPH6501, it seems to be more pronounced than what you would see with a T cell engager.
In particular, this is an Epcoritamab analog, but we know that preclinically, they're almost all equipotent. At least Epcoritamab and Glofitamab would have the same potency. And this is recapitulated here in the graph. And the same was true in patients with leukemic phase recurrent B-cell non-Hodgkin lymphoma. This was touched upon before. You can see here that the clearance of those cells in a patient's ex vivo setting, where NK cells would expect it to be potentially, at least in part, dysfunctional, yet you have a pretty nice clearance, much more so than what you would see with a T-cell engager here.
And again, speaking to another point was raised before in the Q&A section, and here's a comparison with the T cell engagers again. The panel of cytokines that are produced, and these are healthy donor mononuclear cells, is just different in terms of quantity, perhaps not so much quality, at least in part. You see some IL-6, you see some TNF alpha, but not nearly as much as you would see with a T cell engager. And I think that makes sense. Those cytokines are typically what you would see with a T cell activation. You would see this in patients treated with CAR T cell, where some of these numbers go through the roof, including in clinic.
On the other hand, here, some of the other cytokines that you would want to see, interferon gamma, for example, nicely go up. Just recently, with my group, we did some work in trying to understand determinants of response to T cell engagers. In particular, these are experiments done on treatment biopsies in patients treated with Epcoritamab-based combinations. One of the things that we observed, I know for brevity, I'm just gonna focus on the scatter plots here, is that particularly when you focus on the CD4- positive compartment, this is about a month into treatment, so patients would have already gotten the first full dose of Epcoritamab.
In patients who are not going to respond or destined to eventually progress, there is an expansion of cells that are belonging to two specific categories: the T follicular helper cells and the T regulatory cells. Perhaps this is not surprising when thinking about the pro-tumor immune suppressive role of T regulatory cells. But it's sort of concerning that you're using a pan-CD3 engager. These cells are all CD3 positives, and you're sort of expanding, quote, unquote, “the wrong T cells.” This could be a limitation of T cell engagers. And remember, I showed you at the beginning, the second mechanism of resistance related to T cell intrinsic, you know, sort of failures, so to speak. This would be an exemplification of that.
So with a product like IPH6501, that is NK preferential, if not specific, then you would expect not to have this issue because you're not engaging CD25. And so here's exactly the demonstration of that. You can see that both in healthy donors and patients with leukemic phase non-Hodgkin lymphoma, the proliferation that you induce, the activation, presumably, that you induce of NK cells is more pronounced than CD4 cells or CD8 cells. There is some, but just not as pronounced as an NK cell, and this is more exemplified here in the EC50 assay. So based on the collection of this data, there's a lot more data that Eric and his group has worked on, really phenomenal science behind this.
A phase I trial was launched because there was enough confidence in the molecule, which I share fully. We obviously participated, very much engaged, in this particular study that just launched, and this study is looking at the clinical application of intravenous IPH6501 for patients with CD20-positive recurrent B-cell non-Hodgkin lymphoma. It's a classic study with a three plus three design in the dose-finding portion. And then there will be a selection of one or two candidate doses for a recommended phase II dose, and eventually an expansion in a phase II component, where you know, specific settings will be explored based on the signals that we get in the phase one. Nothing special about the study design. The study objective is very you know classic.
You know, we want to certainly look at safety tolerability, establish the maximum tolerated dose, and identify the recommended phase two dose. But then, obviously, even in the phase one, we're gonna be looking at some early key secondary objectives, including preliminary antitumor activity, and also importantly, characterize a pharmacokinetic pharmacodynamic profile of the molecule because what we've seen in the preclinical development space then has to be recapitulated in humans, especially to justify dosing, scheduling, and so forth. We're including patients with a B-cell non-Hodgkin lymphoma that has to be CD20- positive. I think it's very important to stress that the more we go into targeted therapies, the more we can just have all-comer studies, and this has been a limitation of a lot of studies.
We don't have the time to talk about it now, but you all know that when you have all-comer study, then you run into the risk of having essentially a negative study that would have been positive if you had just selected the right patients that have received at least two prior lines of therapy. So the classic relapsed refractory setting, where there would be no curative options, or at least not accepted curative options. The other exclusion, inclusion criteria are very standard. I won't dwell into that. And as I said, the study started in the USA, Australia, and France. This is where we are right now. Most sites that were intended to be activated have been activated.
The first patients have been treated in March this year, and you know, we're eager to be able to share clinical results as soon as possible. Perhaps we should have this yearly so we can update you guys on what's going on, so what are my take-homes? I think I hope I've convinced you that you know, T cell-based immunotherapy is a key addition to the treatment of B-cell non-Hodgkin lymphoma. I can't overemphasize that enough, but they do have limitations, as almost always happens, and in this case, limitation means more than half of those patients are not going to do well on these treatments, and in the case of aggressive lymphoma, unfortunately, succumb from their disease.
NK cell engagers, and in particular, IPH6501, are molecules that are first of all, off the shelf, so there's a manufacturing potentially issue, but not, certainly not scalability issues. They have a potential for high efficacy. We've seen it in the preclinical space, both in vitro and in animal models. They seem to be very NK preferential, if not entirely specific, and so that speaks to the ability of potentially preventing some of those kinda counterproductive activation of T cell of unfriendly T cell groups that are in the tumor microenvironment. And from a safety point of view, also, they have the potential to be safer than T cell engagers.
And therefore, you know, the clinical development of molecules such as this may represent for sure an alternative to classic T cell engagers, and it certainly has the potential to be incorporated in the treatment armamentarium for these patients. What I want to see in next year's meeting, though, I'm sure we're gonna have at this point here, well, first of all, I'm eager to share with you the clinical results, both in terms of safety, but obviously, as anybody wants to know, clinical activity. I'm very eager to see if the pharmacodynamics in blood and tissue are recapitulated in humans.
You know, I wish ourselves the biggest success, and I hope that everybody gets cured, but oftentimes that doesn't really quite happen in the clinic, and if patients will not respond to treatment, then we wanna know why. And so, looking at that translational work, that's so crucial in immunotherapy. And then in the long term, obviously, this holy grail of any treatment, but primary, but more prominently immunotherapies, is always to try and see if some of these patients can be cured, but also whether these drugs should continue to go as single agents or perhaps in the future could be combinable with other agents. The particular toxicity profile, combined with the safety...
Sorry, the efficacy profile of IPH6501 leads me to think that this could be potentially, in the future, a combinable drug. Lots of people to thank, but primarily Eric, Sonia, Olivier, AJ, and the teams. Wonderful collaboration. Very, very happy to be involved in a study like this, and I'll take any question you guys have.
Hey, Lorenzo, any questions? You're on the phone. Okay, I don't know who has the microphone on.
Yeah, um-
Just two quick specific questions. One is a beautiful presentation.
We got a mini microphone coming to you.
There you go.
Thanks. All right-
Thank you.
Thank you. So two quick questions. One is about the tendency of CD16 as a molecule to be proteolytically cleaved, and if something about the tumor microenvironment in patients being treated, if you guys are looking at levels of proteases. And then the second question is about off-target effects of your treatment touching other CD16-expressing cells, such as monocytic or granulocytic myeloid subsets.
Yeah, the first question is: We'll see. You know-
Okay
... we just need the, we just need the samples, but, definitely that's something to, to look at. And, when I talk about resistance mechanisms, you know, you gotta think about, those for sure. In terms of cross-activation, again, we'll see, but I can tell you my experience with, T cell engagers, for example, is, that these drugs are very - they need the trimer to really have that activity. And, in vitro, for example, in the development of T cell engagers in vitro, you can clearly see if there is no trimer, there is no activity. They have almost like an auto shut-off system, which is very reassuring clinically because, yes, there's toxicity in cycle one for those drugs, but then there's literally no toxicity before.
That's because there's no target. And so, I guess there could be cross-binding, I guess, on either side, on CD20 and on CD20. Obviously, there's not much, but on the other side. But then you need the two cell elements to interface with the drug and form a trimer. So I would expect that, yes, there could be some untoward activation, sorry, untoward binding of other cells, but I don't think there's gonna be untoward activation of those cells, at least on paper.
I will just say myeloid cells are engaged and activated very differently than T cells, so sorry.
Fair enough.
... Good presentation. Hi, I have two questions. First one, there is some depletion of B cell in the lymphoid organs to some extent. Does that, given this is only three doses of drugs, does this strike you as ideal, unexpected? What are your thoughts there? Another question is, given that there's right now, you see limited activation of T cells, do you think eventually, NK cell engager will be best partnered with NK cell therapy to achieve better efficacy in terms of T cell activation as well? Thank you.
Thank you. The first question, that's really kind of what you want to see. You want to really see a nice, deep B cell depletion. And look, the yin yang of this is that, this. And with CAR T cell is the most perfect example of this. Patients with long-sustaining CAR T cell levels, that sometimes can go on for years, have zero B cells, and they're profoundly immunosuppressed, which is a drawback of this treatment. But that's sort of an obligated on-target, you know, sort of, side effect, if you will. And you do want to see depletion of CD20-positive B cells.
Now, I would contrast engagers with CAR T cell by saying that CAR T cells is a one and done, so to speak, at least infusion, but then the effect is prolonged, and you can't take it back once you do it. Whereas, you know, engagers usually, you know, are repetitive infusions, and then once you stop it, eventually those B cells will come up. You know, there's data recently showing that slowly, but they come up. They come back up. In fact, in the model that I think Eric had shown, when you stop the third infusion, B cells come back up within a matter of a week or two to 50%. So it definitely is not an irreversible. For the combinability, I mean, that's sort of the obvious thing, you know?
But currently, there's a couple of. I would say I would put a couple of caveats there. One is, commercially available T cell engagers are CD20 targeted, and I don't know that I would want to have two CD20 targeting agents together. Both of which, by the way, work at very low levels of CD20. I showed it for NK, it's true for T cell engagers. So it, I don't know that I would want that. But the second thing would potentially be safety. And it's true that these drugs, the NK engager seems safer, but truly, we don't know until we put it in patients at the dose we think is gonna have meaningful clinical activity.
We have to first demonstrate that it's absolutely safe, and then. This is not just true for T cell engagers. This is true for if you want to combine it with immunomodulators, BTK inhibitors, you know, what have you. You want to have some phase I safety data first.
Questions? All right, I appreciate it a lot. Thanks, Lorenzo. So our final speaker this morning is Dr. Katy Rezvani. She has been a pioneer in the development of NK-based therapies for cancer, with a very exciting work with CAR-NK cells, and I'm looking forward to seeing what's the latest. Thanks, Katy.
Thank you very much, and good afternoon, everyone, and it's a pleasure to be here. Thank you, Eric. Thank you, Miriam. Thank you, the organizing committee, for inviting me and giving me the opportunity to show you some of the work we've been doing with NK cell therapies and the different platforms. All right, so these are some of my disclosures, and some of them are relevant to my presentation. So Eric and Lorenzo did a fantastic job of giving the background, so I'll go over some of these slides very quickly.
But CAR therapies, or as sometimes called, living drugs, usually take a patient, or in our case, a healthy donor's immune system, and ex vivo engineer them to then be infused back into the patient following lymphodepleting chemotherapy. And there are currently six FDA-approved CAR T cell products for multiple different indications, all in hematologic malignancies, with responses ranging anywhere from 40%-90% plus, so really amazing results. More recently, also, there's been FDA approval of TILs for melanoma, as well as TCR T-cells for sarcomas. So the field is really moving and expanding. The problem is that all of these FDA-approved agents are autologous, which means they're from the patient for the patient, and that then means there's complex manufacturing. It's obviously patient-specific, so one product, one recipient.
Because it can take a number of weeks for the product to be manufactured, sometimes patients who have really aggressive diseases cannot wait for the length of time that it takes for the treatment. And then there's also concerns about the health of the T cells of the patient that is being used for the engineering. And you already heard about some of the concerns with toxicity, such as cytokine release syndrome, neurotoxicity. And at MD Anderson, somewhere around a third to half of our patients, after they receive CAR T-cells, need to be managed on the intensive care unit. More recently, there's been concerns of secondary cancers in the T cells that are engineered and infused. Thankfully, it's rare, perhaps one in a thousand. And then you add to that the financial toxicities.
Each autologous CAR T product costs in the region of $500,000. To that, you add the cost of the inpatient management of these patients. Now we're talking about $1 million plus, which is obviously unaffordable for many, and that means that many patients who could potentially benefit from these therapies are not getting access to these potentially life-saving approaches. Which is why we thought, and other groups have also been looking at this, what if we use a healthy donor to manufacture multiple products, and then we freeze it and have it as an off-the-shelf treatment, thereby reducing costs and increasing access? You've already heard about all the characteristics of natural killer cells that really make them very attractive for cell therapy.
In the allogeneic setting, they don't cause graft-versus-host disease. They have, of course, their own endogenous innate receptors, and then you can, on top of that, introduce the CAR to give them antigen specificity. They express the CD16 receptor, which then drives them, one of the most important mediators of why many of the monoclonal antibodies in the clinic work. They're incredibly safe. There's, there's really no reports of greater than grade one cytokine release syndrome or neurotoxicity with them. Certainly, there are disadvantages in that they have limited lifespan in the absence of cytokine support, so somewhere around a week to 10 days. And it's really unclear what's the best starting population for manufacturing. If you're going to have a healthy donor to manufacture 100 products or 1,000 products, how do you pick that donor?
So at MD Anderson, we use umbilical cord blood as a source of our NK cells. Others are using peripheral blood or NK cell lines or iPSC cell lines, and we use umbilical cord blood for a number of reasons. One is because we happen to have a cord bank with more than 40,000 units of umbilical cord blood that have been collected, stored, and my good friend, Dr. Shpall, established that two decades ago. But also, umbilical cord blood is a rich source of functional NK cells. And so we developed all the SOPs to manufacture over 100 doses of CAR NK cells from one umbilical cord blood unit, where we transduce the NK cells with a retroviral vector, and we expand them using feeder cells and interleukin-2.
A few years ago, we published our first in-human trial, where we engineered these banked umbilical cord blood units by isolating them, expanding them with feeder cells. These are K562-based feeders that express membrane-bound IL-21, 4-1BB ligand, and CD48. And then we transduce them with an armored CAR. So this is a retroviral vector that encoded for CD19 CAR, linked to CD28 costim CD3 zeta signaling, IL-15 to help with the in vivo persistence of the cells, and also a safety switch that could be activated in the event of any toxicities.
And so in the phase I portion of the trial, we demonstrated that in eleven patients with highly refractory lymphoid malignancies that had a median of four prior lines of treatment, and they all had active disease when they were treated, 11 achieved complete remission, and these were HLA mismatched cord blood NK cells, and we saw no cytokine release, no neurotoxicity, no GvHD. And what was also really interesting is that despite the fact that these NK cells were HLA mismatched, we could detect them at low levels still about a year following infusion into patients. So a degree of tolerance had been established. More recently, we published a follow-up of the results of this particular trial with 37 patients treated, and for the full study, we showed a one-year progression-free survival of 32% and a one-year overall survival of 69%.
Not really that similar, dissimilar to what is being described with CD19 CAR T cells in fourth line or plus disease setting. But then the question that was interesting to us is: Why is it that a third of the patients go into remission, maintain that remission, and the remainder progress? And we looked at many factors: patient characteristic, CAR NK characteristics, but also the donor characteristics. In this particular trial, each patient had received CAR NK cells that was engineered from a different umbilical cord blood. And we thought, we found that the most important determinant of response was the quality of the donor that was used for manufacturing.
What we found was that if we used cords that were frozen within 24 hours of collection and with a nucleated red blood count of less than 80 million, a one-year overall survival was 94% and a one-year progression-free survival of 70%. Patients that received CAR NK cells from cords that were frozen longer than 24 hours post-collection and with a higher nucleated red blood count had a dismal outcome. And if you think about it, that makes sense. In the lab, we never want to use blood, for instance, that's been left around for more than a day for our T-cell or NK cell functional studies.
But the nucleated red blood count was also very interesting to us, and there was some old cord blood data that reported that higher NRBC in the umbilical cord blood is an indicator of maternal stress, hypoxia, placental insufficiency, et cetera. So then, when we went and dug deeper into that, we found that actually NK cells from optimal cords and suboptimal cords have very different transcriptomic profile and very different epigenomic profile. Those from suboptimal cords have a signature of hypoxia, stress, exhaustion, whereas those from optimal cords have a signature of effective function, expression of transcription factors important for NK effective functions such as Eomes, Tbet, the IRFs, etc.
In multiple mouse models, we showed that if we took CAR NK cells, if we manufactured CAR NK cells from optimal versus suboptimal cords, pre-selected, and then we put them into mice, just like our patients, those from suboptimal cords will not control the disease, whereas those from optimal cords will. This was very important to us because now we had an idea of how to pick our umbilical cord blood units for manufacturing of our CAR NK cells, and this is really what we did in all our subsequent clinical trials. CD19 looked great, and the next question was: Can we use, you know, is this just specific to CD19, or can we use our CAR NK cells to target other antigens? One particular antigen we were interested in was CD70.
CD70 is expressed in many cancers, hematologic malignancies, as well as solid tumors. Its expression in healthy tissues is limited just to a small subset of lymphoid cells. The natural receptor for CD70 is CD27, and then also CD70 as a target had been validated in the clinic because a antibody called vorsetuzumab had been tested in the clinic, and at least the safety had been shown. So we went and designed 34 different CAR constructs using either the single-chain variable fragments derived from the vorsetuzumab antibody or the extracellular domain of the human CD27 receptor linked to different costimulatory molecules, some more relevant to T cell biology, such as CD28, others more relevant to NK biology, such as DAP10, DAP12. IL-15, of course, we included it.
In the absence of IL-15, CAR NK cells don't really work in vivo because there's no persistence and, of course, our safety switch, so these data were recently published. I'm going to skip a lot of the in vitro data, but Sunil Acharya, Rafet Basar, and May Daher in the lab basically tested these. The most promising CAR constructs were taken into in vivo studies, and what was very interesting to us was that construct number 23, which has CD28 costimulation, resulted in the best antitumor control in our mice, and this is in multiple different mouse models. This is a lymphoma mouse model, AML mouse model, and we found the same in a breast cancer mouse model, and so the question was: Why? Because NK cells don't naturally express CD28.
In this paper, basically, we demonstrated that despite the fact that NK cells don't express CD28, they have LCK. Once you introduce CD28 in the NK cells, the CD28 can act as a docking site to then recruit LCK and drive CD3 zeta phosphorylation and basically ZAP-70 and the whole signaling cascade. It was this particular construct, number 23, that then we took to the clinic. In this case, we used an umbilical cord from the bank, and then we ex vivo expanded them with our CD27 receptor, CD70, CAR-IL-15. From two umbilical cord blood units, we manufactured 250 patient doses, and then we froze them. This was a truly off-the-shelf clinical trial where the NK cells were picked without any consideration for HLA matching, stored, and infused at patient bedside.
We've received FDA approval for two different clinical trials. One's a basket study in hematologic malignancies, led by my colleague, David Marin, and a second one in the setting of solid tumors, led by David Hong. Basically, what we do is patients receive three days of fludarabine, and then they have the cells thawed and infused at bedside. FDA were pretty tough with us in that they wanted us to start with a flat dose of just eight million NK cells. Just to put it in context, to a mouse, we give 10 million NK cells.
We had to follow their guidance, and this was the first one of the patients that we treated at dose level 1, 24-year-old male with Hodgkin lymphoma, eighth relapse of his disease, had had two different types of transplant, including a haploidentical transplant from his mother, had had all the checkpoint inhibitors, antibodies, etc. When he came to us, this was the scan that he had. He received the 8 million dose, and you can see day 30 and day 60 in his PET scan. He's gone into complete remission. All these areas of cancer are gone. All you see is the brain, the heart, the kidneys, and the bladder, which is normal, pumping the radioactive dye.
We could also follow the activity of the disease in his blood by measuring lactate dehydrogenase, which is a biomarker of lymphoma activity. You can see it was high here. The arrow shows where we infused the NK cells, and within two weeks, it's gone back into the normal range, and that's what we find. NK cells work very rapidly. Actually, patients that have palpable lymph nodes, really within a few days, we can feel the lymph nodes shrinking or disappearing. So in this particular trial, we're currently at dose level six. What's amazing is that we haven't observed any CRS neurotoxicity, no GVHD. These are totally HLA mismatched with the recipient. Not even tumor lysis syndrome, which surprises me with so much disease. We've observed responses in 12 of 15 patients treated to date.
So we're very excited about this trial and hoping very soon to try and move to a multicenter study. Then the other approach that we started looking at is, okay, well, what about targeting other antigens? And as you heard by the elegant presentation given this morning, T cell lymphomas have a dreadful, dreadful outcome, especially in the relapsed refractory setting. And the targeting with T cells is difficult with CAR T cells. And the reason for that is that the target antigen is often shared between normal T cells and cancerous T cells, so there's a risk of fratricide. Also, if you're collecting CAR T cells from the patient, there's risk of product contamination with the cancer cells, and then T cells have very long survival, so that can result in prolonged aplasia.
NK cells, on the other hand, don't express many of the T cell antigens. We were particularly interested in targeting CD5, which is expressed on many T cell malignancies, so there's no risk of fratricide. Our NK cells come from a healthy donor, so there's no risk of product contamination. They have a shorter lifespan, so the duration of aplasia we expected would be less. So again, we use the same approach of designing multiple different CAR constructs linked to different costimulatory molecules and then IL-15 and safety switch. And basically again, we demonstrated the CAR5 with the CD28 costimulation worked very well. In this setting, we also found that DAP10 was also a very good costimulatory molecule.
So we took this to the clinic, again, using the same platform of picking an banked optimal cord, expanding it using our feeders, transducing with the retroviral vector, and banking the cell frozen off the shelf. Patients receive flu/cy, the cells are thawed and infused at bedside, patient bedside. This time, we had a different FDA reviewer who allowed us to start with 40 million cryopreserved cells, flat dose. And this trial is open not only for T-cell malignancies, but also for mantle cell lymphoma and CLL because they also express CD5. And Dr. Hosing is the PI of that. The trial just started. Actually, in August, we treated our first patient, 74-year-old female with peripheral T-cell lymphoma and immunoblastic type, and she had primary refractory disease.
She'd been diagnosed in December 2023, had chemo, disease progressed, had salvage chemo, disease progressed, had checkpoint inhibitors and HDAC inhibitors, again, progressive disease. And then when she came to us, this was her PET scan. She received the four million NK cells. And then day 30, this is the first time she's gone into complete remission, and she didn't have a toxicity greater than grade one CRS or ICANS. And what was very interesting is that actually her T cells, of course, as expected, went down, but about day 24, her T cells had recovered back to normal levels. And we think that's because healthy T cells internalize CD5 when they are exposed to either CD5 CAR T cells or CD5 CAR NK cells. But we're studying this in more detail.
So now we're exploring targeting solid tumors, which of course is a completely different challenge here in that we don't have good target antigens, and then you're faced with the hostile tumor microenvironment. So in terms of target antigen, we went after Trop-2. Again, because it's an attractive pan-cancer target, it's expressed on many epithelial cancers, including breast cancer, bladder cancer, ovarian, pancreatic. Again, it's a validated target because there's already an FDA-approved antibody called sacituzumab. And so use the same platform of engineering, made different constructs, different costimulatory molecules, linked to cytokine gene and say, a suicide gene. And then we tested in vivo in this. This is a breast cancer metastatic mouse model, showed that it works.
Took our banked optimal cord, expanded them, froze them, and then again, after this trial has been FDA approved, led by my colleague, Cathy Dumbrava, in non-small cell lung cancer and breast cancer, where patients, after lymphodepletion, receive these cells. Dose escalation study, first patient infused, so far, no toxicities. Another challenge with solid tumors is the route of delivery and trafficking. So we wanted to use Trop-2 CAR NK cells for ovarian and pancreatic cancer. But when we tested them, in our mouse models, we realized that the IV delivery doesn't work, and the reason for that is not sufficient numbers of CAR NK cells traffic into the peritoneal cavity. So then we tested delivering the cells intraperitoneally to these mice with a peritoneal type of cancer, and basically, we showed that here we start getting much better control of the tumor.
And so a second trial in ovarian and pancreatic cancer, led by my colleague, Dr. Jazaeri, is testing the same Trop-2 CAR NK cells, but this time in patients with ovarian and pancreatic cancer, but in this time, delivered intraperitoneally to the patients following lymphodepletion. I mentioned the problem with many of the cancers is we don't have good targets. We don't have the equivalent of CD nineteen or BCMA for many of the solid tumors. And the majority of the tumor-specific or new antigens are actually intracellular, which means TCRs can target them because these intracellular antigens are presented in the context of HLA class one or class two to T cells, but NK cells don't express TCR, and so as a result of that, they cannot target them.
In addition, NK cells have activity of CD3 zeta signaling, but they don't have the full CD3 complex. But there is an advantage to using NK cells because many solid tumors downregulate HLA class one, which means that your TCR may not even be that active. So we thought, well, what if we engineer our NK cells to express a T cell receptor? But here, the challenge was, in addition to the TCR, we had to get the full CD3 complex to come together to have both surface expression of the TCR, but also the intracellular signaling. So basically, we engineered our NK cells, co-transduced them with two different vectors. One is a signaling complex that has a full-length CD3, including CD3 zeta, epsilon, gamma, delta, a costimulatory molecule.
Costimulation is also very important for NK cells, IL-15 , and then the TCR of interest. In this case, we use NY-ESO-1, and this was work led by Bin Liu in the lab, and then we showed that NY-ESO TCR NK cells were just as good as NY-ESO TCR T cells in controlling the tumor in mice, and then we've translated this to the clinic. We have two FDA-approved clinical trials with NY-ESO TCR NK cells, one in multiple myeloma, led by my colleague, Dr. Qazilbash, one in synovial sarcoma. The trials started, again, starting with banked umbilical cord and manufacturing these banked cryopreserved products. We are dose level one, and I'm really excited because the first patient with the myeloma is doing really well.
We also have another target, which is PRAME, and PRAME is a very attractive cancer testis antigen, which is expressed in many cancers, including AML, as well as solid tumors. So we've received, again, FDA approval for these optimal cores, manufactured the product, which is frozen off the shelf. Clinical trial in AML, led by my colleague Jeral Ramdial, and the first patient has been consented for this trial. Now I'd like to look at another approach of retargeting the antigen's specificity of NK cells. We thought, rather than putting a CAR, what if we just combine them with a bispecific engager and so redirect their specificity that way?
I mentioned to you, NK cells, through their CD16 receptor, mediate ADCC, so we collaborated with this company in Germany called Affimed, and they had this bispecific engager called AFM13, that bound CD16 on the surface of NK cells on the one hand, and CD30, which is a cancer antigen expressed in many types of lymphomas, including Hodgkin lymphoma, on the other hand. This particular bispecific engager as a single agent had been tested in the clinic in Hodgkin lymphoma with an overall response rate of 16% and no CRs. Basically, we combined it with our NK cells. The preclinical work is already published. I'm going to share with you the data of the clinical trial, which is currently under revision. Hopefully, it'll come out very soon.
Where we took again a banked optimal cord, we ex vivo expanded and activated them, and prior to infusion of NK cells, we combined them with the antibody with the AFM13. Because we know that the antibody will either get internalized or will fall off after a few days, patients also received additional infusions of just the engager. 42 patients with Hodgkin lymphoma and various types of lymphoma were recruited in this trial. They'd had a median of seven prior lines of treatment, again very refractory, up to 14 lines. I didn't even know there are 14 different options of treatment for patients with lymphoma, but at MD Anderson these are the types of patients that we see. They'd all progressed after CD30-targeting antibody Brentuximab. All had had checkpoint inhibitors. More than half of them had had a transplant.
Some had progressed after CD30 CAR T cells, and they all had refractory progressive disease when they came to us, and my colleague, Dr. Nieto, is the PI of this trial, and basically, this slide summarizes the data. Our overall response rate was more than 90%. Again, very similar group of patients that were treated with the single-agent therapy. Our complete remission rate was 67%, and this is a waterfall plot showing it. Again, we had no cases of CRS, neurotoxicity, or GVH. This is a PET scan before and after of a young man who'd had all these different types of treatment, including CAR T cells, including an allogeneic transplant, and of course, checkpoint inhibitors.
So of course, this was great, but there was a big problem in that none of these NK cells that we infused were not engineered, which means that they only persisted for a week or so, which meant that that many of these patients ended up needing additional infused cycles of therapy, and we had to infuse far larger numbers of cells, right? I showed you where we put patients into remission with just a flat dose of eight million. Here, we were infusing two to four billion NK cells, which is very expensive to manufacture.
So then we thought, "Okay, well, this to us showed proof of principle of how powerful it is to combine NK cells with engagers." And so we thought, "Well, what if we bring all of our knowledge to date together and make this new NK cell that we call plurireceptor NK cells, so that we can do multi-antigen targeting and, but also have persistence of the cells?" Basically, we already had managed to put the full CD3 complex together, and of course, you need CD3 if you want to combine your NK cells with T cell engagers. That's exactly what we did. We had CD3 vector that had the full complex, including CD3, zeta, epsilon, gamma, delta, and costimulation linked to IL-15 .
In order for CD3 to be stably expressed on the surface of any cell, be it T cells or NK cells, it needs to engage with the T cell receptor. In this case, we didn't want to target a particular antigen, so we used invariant TCR instead. Then we thought, "What if we have high-affinity CD16?" About 25% of the population have this polymorphism in the CD16 receptor that allows their NK cells to bind with very high affinity to antibodies. So the idea was that then this NK cell will express CD3, so that we can then combine it with all FDA-approved T cell engagers that are currently available.
Actually, I should include the DLL3 one here as well, and then express high-affinity CD16, which means that we can then combine them also with monoclonal antibodies and ADCs. The invariant TCR perhaps will help them also target CD1d-expressing cancer cells or APCs. Of course, they'll continue to express their endogenous receptors like NKG2D, DNAM, et cetera. And then they will secrete IL-15, which means that then now we can perhaps just give five million cells, ten million cells, forty million cells instead of giving four billion cells. So then we tested these in our mouse models, work led by Bin Liu and Rafet Basar in the lab, where we took our NK cells and combined them with a bispecific T cell engager called elranatamab.
In this mouse model of myeloma, we showed that elranatamab, which binds CD3 on the one hand, and BCMA on the other hand, doesn't have any activity in NSG mice because they don't have T cells. If we give the plurireceptor NK cells that don't have any antigen specificity alone, there is no control of the tumor. When we combine them together, then we can actually cure the mice. Now this is an approach that we're planning to take to the clinic, where the plan is to make a biobank of these plurireceptor NK cells that are frozen off the shelf. Then, basically, all we do is we give patients flucyte, we infuse the cells, the NK cells, and then the patients also will receive the engager as a separate infusion.
And then, like everybody else, we're very excited about the power of cell therapy for the treatment of autoimmunity. Since we're already gonna be making these plurireceptor NK cells, which MD Anderson now is calling them PluriLink, we also have a different protocol that's just received FDA approval in lupus and scleroderma, where we are infusing our NK cells with tafasitamab, which is a monoclonal antibody targeting CD19. So again, it's like a way of targeting and depleting B cells and plasma cells. And so the first patient for this particular trial now has been identified. Okay, so this summarizes our current pipeline at MD Anderson. I mentioned the CD70 CAR NK cells, Trop-2 CAR NK. I didn't have time to go over our glioblastoma platform, where we're using CRISPR-edited NK cells and, of course, TCR, CD5, and plurireceptor.
I think that the field in the next few years is really gonna expand, and there's going to be so much that we could potentially do. We have an increasing in understanding of the biology of the cells. We have the tools of synthetic biology. We've shown we can engineer our cells to express CARs, TCR, cytokines. 'Cause I get so excited, I just don't know what to do, really. Then we can combine them also with checkpoint inhibitors, immunomodulatory drugs, oncolytic viruses. Yes, so very exciting field that we're in. That's why science is so much fun. This is my amazing team who work day and night developing the programs.
Our GMP facility that makes all the cell products, our cord blood bank, my incredible collaborators on the clinical side, our funding agencies, MD Anderson Moonshot platform, most importantly, our patients and their families for their courage to come and really take part into our first in-human studies, and every day, I'm in total awe of them and their commitment to our program, so thank you very much for your attention.
... Hey, great. Thank you. Pretty fantastic. We have time for maybe a few questions. Right here in the middle.
Thanks. Thank you so much for the great lecture. On your first slide, you showed the cost for the CAR T cells and everything. You didn't show any comparison with the NK cells. Would you have a-
Yeah, great question. So we've calculated that our banked products. So, for instance, I showed you the CD70 CAR NK program. It costs an average of $600-$1,000 per dose. And actually, this is even without us trying to do the manufacturing scale-up, so it's significantly cheaper. And the reason it's cheaper, the manufacturing is just as expensive, is because we're getting many more doses. That's why.
Thank you for the very inspiring talk. I was wondering how long or what's the longest that you've watched the patients, after they've received NK therapy? Because even with the IL-15, what happens? Is there a high risk of relapse, when the NK cells eventually die, or does it persist?
Yeah, great question. No, we have some patients that are, like, four years or five years out in complete remission. But the CAR NK cells, after two years, we no longer find them. So they're gone. Yeah.
Impressive results. Impressive, and congratulations for an extraordinary effort. I have one question - two questions, in fact, one scientific and one non-scientific. The scientific one: So the eight million was very surprising, the data with the eight-
Yes
... million NK, right? So you have these eight million. So, do they expand tremendously? They did.
Yes.
And then you destroy all these tumors, and then you have no side effects. I mean,
It's amazing. I know.
S-
I don't understand it, you know? And we have to study it, but, you know, it's just we have so many other things to study. What we are finding, you know, with CAR T cells, a lot of the toxicity is driven by the myeloid cells. So what we find is actually CAR NK cells can kill activated myeloid cells. I don't know whether that's the reason, but in patients, we don't see peaks in IL-6 or IL-1 receptor alpha, so the cytokine profile is totally different.
Can you detect them in the blood circulation, your NK that-
We do-
Yeah
... detect the CAR NK. So we get very nice proliferation, and we find them by flow cytometry for about a month to six weeks.
Mm-hmm.
After that, really, the numbers dwindle down. By qPCR, we can still detect them for up to a year following infusion.
Wow! Okay. The non-scientific one is, there is no industry sponsors here because this is extremely expensive, right?
Yeah.
Yeah.
Yeah, great question. So our CAR 19, IL-15 NK program-
Mm-hmm
... the one that was published first, that was licensed to Takeda.
Mm-hmm.
The TCR NK program is partnered with MD Anderson and Replay have formed a company called Syena. But all the other programs are funded by grants and philanthropy. So all the CD70, Trop-2, CRISPR-edited, everything, all of it is really through grants and philanthropy. Yep.
Well, congratulations.
Thank you.
Had a great talk. Amazing talk. Two questions: The Plurilink, I saw, the addition of IL-15 to keep them alive. Is that the only means, and if, and how long do they persist?
Yeah.
The second, I don't actually know if this is my question, but the conditioning is Flu/Cy for three days. Is that the optimal conditioning for an NK product?
Yeah, great questions, Lorenzo. So we are exploring other cytokines. Actually, we just had a paper come out in Cancer Cell, where we've shown that IL-21 engineering of NK cells is probably advantageous in glioblastoma. I didn't have time to include it, but we are at the moment IL-15 seems to be the best one, at least for systemic delivery. We're also looking at combinations of, let's say, IL-15 and IL-21. IL-12 is very toxic, so we're kind of still working on that. Fludarabine, we first did just 300 milligrams per square meter. Now, we've started in some of our programs to see whether we should increase the intensity to 500 milligrams per square meter of cyclophosphamide. In our renal cell cancer trial, that's what we've done, and we see even better proliferation and persistence of the cells.
So it's quite possible that more is better. Of course, you don't want to increase toxicities. Yeah.
Last question.
Thank you. Great talk, yes, lovely data. I'm curious about two things. So, NK is developed in this way, is a beautiful way of the bulking, biologically, cancer cells in a human being. But my question is, like, with those two questions. First, in these kind of malignancies, how much is driven by the NK effect over time? I mean, when is the T cell compartment engaging? When are you generating memory, or is this treatment able to generate a long, persistent antitumor memory? And my second question will be, talking about toxicities, you are not seeing toxicities. You are thinking that, in the case of these malignancies, are related with that NK is killing myeloid compartment cells. When you are going to have to, to-...
Envision that in a solid tumor, how you are going to try to make your NK a source of new tumor antigens and engage T cells that are the only ones able to eliminate the tumor?
Yeah, great, great question. So we do have data showing that NK cells activate the endogenous immune cells of the patient. We have also really convincing preclinical data showing that NK cells and myeloid cells cross-talk, and when they're together, the antitumor activity is much stronger. We're doing some adoptive transfer studies where we are infusing NK cells and myeloid cells, monocytes together. So that synergy definitely is there. And also our cells secrete IL-15, so they're bound to also activate the endogenous myeloid cells and T cells. Solid tumor is very, very challenging, and of course, now we're getting a lot of data coming through from our patient trials. We're doing spatial transcriptomics, finding that NK cells perhaps don't traffic as well into that tumor microenvironment and trying to really decipher what are the, those mechanisms to make the next generation studies.
We're doing a lot of CRISPR screens. I think the whole field is really focusing now on solid tumors and how we can improve on the activity of both T cells and NK cells in that field.
Thank you.
Do you ever see any autoimmune manifestations?
No, we have not. We have not at all. We also do autoantibody screens on the patients, as well as alloantibody screens, because obviously, there is a mismatch between the donor and the recipient. We've seen induction of alloantibodies against the infused donor NK cells in a couple of cases, which doesn't surprise me. Yeah.
Thanks again.
Thank you.
Thanks for all the speakers for this great session. We'll be breaking now for lunch. Very nice.
Thank you.
... All right, everybody. I think for the sake of time and, like, just the amazing lineup of speakers that we have in the second half, we'll get started today. It's my great pleasure to introduce our first speaker, and I've been instructed to keep introductions near nonexistent, so Dimitrios Skokis from Regeneron. He is the Senior Director, Executive Director of Cancer Immunology Research for Regeneron. He's the Elizabeth Miller Medical Director of Clinical Sciences and Global Development. Very briefly, Dimitrios has had, like, an amazing effort targeting CD28 with Regeneron's bispecific antibody approach, and so that's what we will hear from today, so it's a great honor to introduce you and to welcome you, and let's give him a round of applause.
... so I should get the applause at the end, hopefully, not the beginning.
You can, just don't point at the audience.
All right. Thank you. All right, I would like to start by thanking the organizers, Vivian, Eric, and Miriam, and the whole team here for inviting me. It's an absolute pleasure to be here today and get the chance to share some of our exciting early clinical data on this platform, what we call the CD28 bispecific or costim bispecific platform, which is actually a tumor-targeted costimulation approach to actually turn immunotherapy cold tumors hot. I would like to emphasize that this project started a little over like fifteen years ago now, and the basic idea was to try to create a model system of plug-and-play approach, where basically we can tailor the various immunotherapeutic strategies to the immunological landscape of the tumor.
And what I call the immunological landscape of the tumor, I mean, like, the immune composition of the tumor, but also the activation state of the immune cells that composing this tumor. Oh, what is that? Yeah, this is my disclosure. So today, there are two types of immuno-oncology drugs that dominate the news. We have the non-tumor-targeted anti-inhibitory approaches that involve the checkpoint inhibitor, CTLA-4 and PD-1 or PD-L1, that did actually revolutionize the way we treat cancer today. And however, they also come with limitations, which means that not all the patients they respond, and among the responders, there are a majority of the patients that actually, they're very, very short on the tumor immunity.
Path forward for this type of issues is the combination therapies, but also patient stratification, where the use of biomarkers can predict responsiveness. On the right-hand side, we have the tumor-targeted or anti-inhibitory approaches, where we involve the CD3 bispecifics, that I will talk a little bit today, but also the CAR T-cell therapies. That can be, you know, very potent, and they have been proven to be quite efficient in certain types of hematologic malignancies. However, they also come with challenges such as neurotoxicity or due to these highly personalized manufacturing processes and preconditioning for immunotherapeutic regimens. A lot of these patients, unfortunately, they're not deemed suitable to enroll these type of treatments. We and others, we think that potentially the next big thing in the immunotherapy of cancer could be the use of costimulatory bispecific antibodies.
And why we say that? Before we go into that, I would like to take some step back and go back to some basics of how T cells are activated. So for T cell, in order to be properly activated, it requires two signals. Signal one, which means that you're gonna have tumor peptides or viral peptides that are gonna be presented on the surface of HLA molecules, and they're gonna be activate CD3 TCR complexes, providing signal one. And then the signal two situation, where costimulatory receptors, such as CD28, will be engaged by natural ligands, CD80, CD86, or B7-1, B7-2, and this is gonna provide signal two.
And that's happening only in the presence of signal one, which means that when both present signal one and signal two are present, then T cells are gonna be activated, proliferated, expanded, they're gonna start secreting cytokines, and basically, they're gonna be able to aim against the target cells and promote this target cell lysis. The immune system also has evolved to put some brakes on the system to avoid getting auto immunity by upregulating various types of checkpoint inhibitors. I mentioned here one, PD-1, PD-L1. What is happening in case of tumors, in particular, solid cold tumors? And that's a little bit the premise of the whole concept here, which basically, tumors have evolved, and they can hijack the immune system by either down-modulating signal one, which mean MHC class one molecules, HLA class one molecules, and/or signal two.
In this particular case, basically, what is happening is that in the absence of signal one, signal two, we're not gonna have a proper T cell activation. And basically, that was the whole premise of the concept, which means that now, how are we gonna be able to provide the missing signal one and signal two, specifically targeted fashion in the tumor microenvironment? Basically, in other words, we want to convert tumor cells into antigen-presenting cells. And it's exactly when these type of basic questions can be complemented by revolutionizing technology platforms. And in this particular case, we wanted to use tumor-associated antigens expressed on the surface of a tumor cell and CD3 or CD28 expressed on the surface of T cell as anchors to simply cluster, bridge tumor cells with T cells.
In this particular case, we're gonna bring a bispecific antibody, where we're gonna cluster CD3 and tumor-associated antigen, and this is gonna give signal one, the missing signal one, and a costimulus specific that is gonna provide the missing signal two. And eventually, we can add in this equation an anti-PD-1 or an anti-LAG-3, where basically we can further remove the brakes and enable T cells to be more efficiently activated. In a recent review that I wrote together with my team colleagues, Michael and Scott, that if you haven't read it, please go ahead and read it. We present a little bit the history of CD28 and the impact of CD28, a pathway in the development not only of antibodies against CD28, but also indirectly, development of checkpoint inhibitors.
Also, the development of CD28 costimulated CARs, and also the development of CD28 or costim bispecific antibodies. But I want to draw your attention to a significant event that happened in two thousand and six, which means that how actually specific ways of designing experiments may have a long-term impact in the way we approach science, and also in the way we design clinical trials. So in two thousand and six, Regeneron developed a CD28 superagonist antibody, and by superagonist, it means that it binds to the proximal domain of CD28, so it can activate T cells in the absence of Signal one. So six healthy individuals enrolled in this trial, and all six individuals experienced cytokine release syndrome, followed by multi-organ failure. After this incident, a lot have changed, which means that how first-in-man trials are approved by regulatory authorities have changed.
The design of cytokine CRS mitigation strategy did change, but also all the activities related to the CD28 pathway were also completely shut down. So we have this question that we need to ask back in two thousand and seven, two thousand and eight, when we first got interested in this concept: How are we gonna move forward? So we had a lot of clinical data that demonstrated that CD28 pathways is extremely powerful pathway. So the whole point is that how we can grab this tiger by its tail and control it. That's the whole concept here. So we want to develop antibodies, costim bispecific antibodies, that they are not superagonist. And our costimulatory bispecific antibodies were designed with several molecular controls in place, which means that our bispecific antibodies will work only in the presence of Signal one, and only in the presence of a tumor-associated antigen.
If one of these two or both are missing, then these antibodies, they're gonna have limited antitumor activity, limited activity, and, or no activity, basically. So in addition, we work very closely with the FDA to develop multiple CD28 by TAA bispecific phase 1 trial designs in order to ensure safety before even we start measuring activity. So in this cartoon, I summarize a little over ten years of research, you know, based on Charles Dickens, A Tale of Two Signals, a little bit influence on there or philosophical. On the left, on the left-hand hand, we have actually the first modality, which means that we can combine a CD28 bispecific together with another PD-1. In this case, we rely on the ability of tumor cells to present tumor-mutated peptides on their surface. So we have an endogenous TCR triggering.
On the right-hand side, we combine a costim bispecific antibody together with a CD3 bispecific antibody. In this case, we providing exogenously Signal one. So therefore, you can imagine that we need to actually make sure that the type of tumor-associated antigen we are pairing with the CD3, they have to be, you know, not necessarily highly expressed in normal tissues to avoid superactivating or activating T cells a little bit all over the place. So today I'll be talking a little more, focusing here, but there is a lot of activity on the right side here as well. So today I'll be talking about a combination of PSMA CD28, with cemiplimab, our anti-PD-1 antibody, in metastatic castration-resistant prostate cancer, and in our EGFR CD28 bispecific antibody, again, in combination with cemiplimab in several advanced solid tumors.
For the people that want to dive more into the preclinical data, all this work has been published back in 2020. One of the very best experiments that we did was actually to go back and expand the data that came from Jim Allison and Liping Chen, where basically we overexpressed CD28 natural ligands, in this particular case, CD86, on the surface of tumor cell lines, in this case, mouse colorectal MC38. Actually, as you can see, although overexpressing CD86 on the surface of MC38 does drive some sort of antitumor efficacy, adding a PD-1 does a much better job. This is associated with long-term survival benefit in this tumor model. This efficacy is driven by CD8+ T cells, which means that when we deplete CD8, actually we lose the survival benefit.
Actually, this combination does also drive strong memory, which means that when we rechallenge these tumor-free mice, basically, you know, with a tumor cell lines that not necessarily express PSMA, parental, actually, all these mice, they do survive, which means that they do develop this long-term antitumor immunity. Now, in this experiment, actually, was designed originally to better understand the localization of PD-1 and CD28 within the immune synapse, but also in parallel, Ira Mellman published a beautiful paper, we're demonstrating a little bit the contribution of, you know, of this pathway and PD-1 CK2 phosphorylation. Al is here in the audience. So basically, in this experiment, we actually use engineered cell lines that express different targets of interest, engineered T cells and B cells in this particular case.
And then we use Amnis technology, which means that we use a technology that combines FACS, flow cytometry, and confocal microscopy. On the top line, we're using an anti-PD-1 antibody that is non-blocking, so it does not block the natural interaction of PD-1 with the PD-L1. As you can see, when we add CD20, CD3, between you create this synapse interaction between the B cells and T cells in this particular case, and this antibody colocalized at the intersection of these cells. Now, if you don't block PD-1, PD-L1. PD-1, upon engagement by its natural ligand, PD-L1, is going to accumulate in the immune synapse. And interestingly enough, CD28 is going to be more or less excluded from the immune synapse, and this is going to be quantified on the right side panel here.
However, when you use a PD-1 blocking antibody, in this case, Cemiplimab, which is the approved drug, you block PD-1, PD-L1. PD-1 is going to be more uniformly expressed around the cell membrane, and then CD28 comes back to the synapse. Basically, we conclude that PD-1 inhibition does decrease localization of PD-1 and enhances the accumulation of CD28. Now, to do this experiment, we use actually engineered mice, where actually we humanized. This is five types of humanizations. We humanize the extracellular portion of the target of interest, and we use intracellular domain for signaling purposes. Intact, where basically here we humanize CD3 gamma, delta, epsilon chain, CD28, and PSMA. And basically, what we do in these mice, we implant it with MC38, in this case, and as you can see, the two monotherapies, they have limited antitumor efficacy.
However, we put together these two reagents that drive a long-term survival benefit, and you do not detect any peripheral cytokine secretion when you compare with actually the CD28 superagonist antibody. Now, this is a prophylactic treatment. When actually you delay and you start treating later on, by day nine, where the tumor is approximately around 50-100 mm³, then actually, again, the combination effect is strong. And interestingly enough, when you take, isolate tumor-infiltrating lymphocytes from the mice that have been treated with a combo, these are the guys that actually they have been activated, and as here, you detect interferon gamma secretion, which means that the original hypothesis that these type of therapies can actually specifically target T cells intratumorally has been validated. Now, in a monkey tox study, again, we use a superagonist antibody TGN1412, in this case.
As you can see, TeGenero antibody, actually, we could detect 5 hours post-administration of the TeGenero antibody, strong IL-6 and interferon-gamma secretion. Unfortunately, the trial that was run back in 2006, they looked 24-48 hours after the administration of the antibody. So matter, when you look cytokine secretion, does matter. However, when we add PSMA-CD28 and alone or in combination with PD-1 at high doses, you see no detection of peripheral cytokine secretion. Similarly, in an FDA-recommended plate-bound assay, again, where we use PBMCs, again, the TeGenero antibody does drive PBMCs proliferation, and this is not the case with our bispecific or CD28 parental bivalent antibodies.
So some of the preliminary results now from our phase I study that basically the primary objective here was to assess safety, tolerability, PK of the combination of our REGN5678, our PSMA CD28, in combination with Cemiplimab in metastatic castration-resistant prostate cancer. This is the study design, where basically we start with a lead-in of your bispecific for three weeks, followed by the fourth week addition of Cemiplimab. And you continue for another three weeks, where you add again Cemiplimab, and this represents one cycle. As you can see here from the dose schema, we started below MABEL, right? Again, I'm referring to the way a lot of these trials have been designed due to this superagonist effect of the CD28 back in two thousand and six. So we start at 0.1 milligram IV all the way up to DL 8, 300 milligram dose.
To get from 0.1 to 300 took us a little over one and a half, almost to two years. These are some of the efficacy data we see. There are minimal signs of efficacy at low doses of costim bispecific between 0.1 and 10 milligrams, okay? We have only one out of 17 patients showing PSA decline. However, when we start dosing at higher dose levels, between 30 and 300 milligram, we have over 40% of the patients that show very steep and sustained PSA declines. Actually, if you look DL 8 at 300 milligram, you have three out of four patients that they show really severe PSA declines. Interestingly enough, this phenomenon is also associated with immune adverse effects, which means that a lot of these patients, they do develop different types of autoimmune toxicities.
Currently, we're working very closely with the FDA to uncouple this autoimmune toxicity with this very phenomenal and strong antitumor efficacy that we observe in this tumor type. In addition to that, we are also replacing PD-1 by a PSMA CD3 because we believe that Signal one, the source, the type, and the quality of Signal one is an important driver in this type of results that we're getting. I will also to share some, you know, preliminary clinical data from another phase I study, where we're using EGFR CD28 in combination with cemiplimab in patients with advanced solid tumors. Again, 70/75 EGFR CD28 is again, a novel approach. It's a first-in-class EGFR CD28 antibody that binds EGFR expressed on the surface of the tumor cells and CD28 on the surface of the T cells that reduce proliferation, survival, and memory. This is the study schema.
Primarily, we have metastatic, a metastatic or locally advanced solid tumors, validated for EGFR expression, that in the patients who are on this trial, they must have exhausted therapeutic options that are expected to provide meaningful clinical benefit. Again, the same similar study schema, we have a lead-in of our costim, followed by addition of Cemiplimab at 350 milligrams. And again, we started below MABEL at 0.03, and we're going over, all the way up to 900 milligrams. Primary objectives: preliminary efficacy, PK, and immunogenicity. This is the patient characteristics. Primarily, the type of tumors and patients that are on this trial, we have 61 patients of, MMR proficient and MSS stable CRC. Again, this data are coming our way, and we're about to analyze.
This is a subset of patients that actually we do see responses in CRC, and these are the patients that they don't have active liver mets. We have, again, preliminary data, 20% responses with one CR and two PRs. Actually, after data cutoff, we have an additional PR response. Just to remind everybody, this is a very difficult tumor to treat, and you have zero responses to any immunotherapies out there. I just want to emphasize the, this—which is the same data that I showed before, which is basically, I want to emphasize a little bit the durability, the, how a lot of these patients, they are still all the way up to eight or nine months, and you still have these, decreased target lesions from baseline.
So in addition to that, we can detect interferon gamma over time in these patients. We have this level of interferon gamma in the serum, and it is associated with the combo effect in comparison to the monotherapy and the baseline. But also, we have association of increasing interferon gamma secretion in the serum with complete responses and PR responses in these patients. So I would like to conclude here that this is a novel agent with early efficacy and encouraging PK evidence, suggesting that seven thousand seven hundred and seventy-five can enhance immune responses and antitumor immunity to cold tumors. So far, it seems to be safe and tolerable up to nine hundred milligrams, and this is gonna be the first costim biospecific immunotherapy to demonstrate clinical activity in a historically immunotherapy-refractory tumor.
Actually, I didn't mention that before, but including a patient with liver mets after data cut off. Currently, we're initiating these expansion trials, including non-small cell lung cancer, head and neck, CSCC, and CRC. I just wanted to finish with this slide, sharing that actually, we are enrolling within our prostate cancer patient, PSMA CD28, as I mentioned, and coupling autoimmune toxicities with efficacy with our PSMA CD28 with PD-1. Enrolling currently PSMA CD28 with PSMA CD3. Then we are enrolling in multiple expansion cohorts with our EGFR by CD28. We have ongoing effort with a MUC16 by CD28 in advanced ovarian cancer, and again, we're using other combination with PD-1 and/or with our MUC16 by CD3.
And finally, but not least, we have a big effort with hematologic malignancies, where we combine our CD2, CD28 with a recently approved Odronextamab with a CD20, CD3 in diffuse large B-cell lymphoma. But also, we're combining with our CD38 by CD28 in multiple myeloma. So an additional costim biospecifics were, they're gonna be entering the clinic in twenty twenty-five and beyond. I have to update this slide. So most important slide, again, it takes a village to do all these experiments and effort. This is an effort that spans for over fifteen years and primarily was initiated by Janelle Waite, a very great scientist and director in my group. Bei Wang. Then we have great help from Immuno-Oncology biospecific department, headed by John Lin, previously headed by Gavin Thurston. Clinical oncology team, Izzy Lowy, you know, and his team.
They're doing great things. And of course, special mention for George Yancopoulos and Drew Murphy for constant support. But above all, we'd like to thank a lot of these investigators, the patients and their family that are enrolling these trials, and all of you for paying attention to this talk. Thank you very much.
Dmitri, how's it going? Great, great work. So I heard- How's it going?
Good to see you.
So you— I've heard George say this before, but you, you said that you saw autoimmunity. So why are you saying autoimmunity and not, you know, CRS or inflammation or... You know, is it, is it— What exactly are you seeing that you're-
Yeah, so basically, we have. We do not see CRS, by the way.
Okay.
In comparison to what has been described in the past for different types of CD28. When I say this tight correlation between efficacy and autoimmune toxicities, is because, for instance, we have infiltration of, you know, liver toxicities, that we see like hepatitis, right? So this could be one of the type of case. So we have smaller type of like, you know, side effects around the mouth, you know, that-
But is it like anti-antigen?
Need to discontinue. So the hypothesis here is basically, we believe what is happening. We said what exactly we're supposed to do, basically. We are targeting T cells within the tumor microenvironment. We are activating T cells that actually may end up being cross-reactive with self. These T cells, if they travel and they go to different sides of the body, they may recognize self-antigens and then may drive this type of phenomenon. That's why I say it's very important to understand, first of all, to demonstrate that you have these shared TCRs between tumor and tox tissue.
Yeah
but also demonstrate that actually you can uncouple. If you understand the, where this is coming from, you can find ways to uncouple. This is not the first time that autoimmune toxicity has been described. They have happened in CAR T cells as well, and you have approved drugs with this type of like-
Interferon alpha.
NK profile. So I feel that we are working very hard to really uncouple this phenomenon. That's the reason that I'm talking about immune toxicity, because this, I don't think that this is driven by activation and by ability of these T cells to recognize self. In a way, you can break tolerance here.
Great.
That's kind of what it comes down to that.
Dimitri? It's a great talk. The costimulatory bispecific space is very hot, as you mentioned. What are your thoughts regarding a comparison between separate costimulatory, like in your model, for example, lymphoma, guys, so the Odronextamab model, combined with a CD22, CD28, versus a molecule that has a costimulatory moiety in it, as a single drug?
Oh.
Do you think that-
That's a different question.
Yeah.
Okay.
Do you think that it's important to have them in form, like a trimer only?
Mm
... or it's important to have a tetramer with the two drugs? Because sometimes schedule can come in the way of that. Like-
Yes, absolutely
... the timing when you administer the two drugs.
Absolutely. Yeah, I had a lot of conversations with different folks in the field about that. This is a very important question. We are developing a lot of these multispecific antibodies. We're developing a bi-specific, tri-specific, tetraspecific, and yeah, different types of formats, and there are, like, hundreds of different types of formats out there. Conceptually, I like the bispecific for this type of approaches to simply because you can dial up or down separately, signal one with signal two. I'm not in a fixed geometry, where basically, if I have a CD3, CD28 with TAA, I'm gonna be forced with one of those to give signal one and signal two at the same time. So if I want to dial up or down differently, my signal one versus signal two, I'm gonna be very limited. So for this situation, I'll prefer that.
There are different situations that I may prefer something else, but for that, I'll go with that. And I had this conversation with Frank Nestle, actually. When they are developing their tri-specific antibodies, they have a CD38, CD3, CD28 that they put in the market, and they're having another one. Yeah, unfortunately, you know, they discontinued right now. I don't know what the exact reasons are, but I believe that the fixed geometry can be problematic for both for efficacy and adverse effects as well. I don't know. I don't-
Engage.
I think the side effects are driven by the type of signal one that you're providing. I think the signal one can be the driver here. Here, we're relying on endogenous signal one. So you have an environment that the tumor cells are presenting the antigen, especially when you're providing exogenous signal two. So you convert them actually to antigen-presenting cells. However, if you come with an exogenous CD3 here, then actually you're gonna generate deeper, a different type of repertoire, where you're gonna be a little more driven to a specific TAA. You may get difference in the memory. You may get stronger memory when you're relying on endogenous signal one versus on an exogenous signal one.
But we are trying to unravel all these rules right now in the clinic because we're visiting a lot of these things with a state of mind and knowledge that we have generated over the last thirty years. But I just feel that often we need to change certain of these rules to understand whatever it is we're seeing in the clinic, the real data. And a lot of these data, they do actually correlate, associated with preclinical. A lot of these preclinical models were quite predictive of what we see in humans, with the exception of toxicity.
Yep. Thanks so much, Dimitri, for a fantastic talk, and congrats on all the wonderful drugs you guys are developing. I have a question about the patients with liver Mets. So we know we have seen lack of responses in melanoma and checkpoint inhibitors. Here, the mechanism is obviously different, so which is a bit surprising because it's the targeting is different. So do you see responses in the non-liver targets, or is this really liver exerting a systemic immunosuppression and making your T cells not being able to target the tumors that are not in the liver? Yeah.
Great question. Thanks for your comment.
Yeah.
It's very early. This is literally very early data. I cannot comment because it's literally early. I'm gonna. I mean, I can speculate, but, you know, it's a different mechanism to begin with.
Yeah.
I don't know what to anticipate. What I was saying to Mario before is that each of these cohorts, it's a different solar system. I take them separately, and I'm trying to take the face value, all the data we're getting. I don't know how, especially with liver mets, because this is a particularly cold, freezing tumor. Let's see, this is very hopeful that it's gonna translate to more and more patients, and we're gonna start to get some patterns. But so far, we see more higher response, meaning whatever I saw here, higher response with no patient with no liver mets. But I'm not going to speculate more because it's too early. I don't want to do that.
Sounds great. I think for sake of time, we can carry on questions for the next one. Thank you so much, Dimitrios.
All right, well, so it's my great honor to introduce Dr. Miriam Merad. Miriam is not only my mentor, but a mentor to many of us, both at Sinai and abroad. We're all exceptionally grateful for the vision and the leadership. Miriam, very briefly, 'cause I'm, I know we're not supposed to do major introductions. A huge pioneer in the space of myeloid cells, early work on defining ontogeny of macrophages, dendritic cells. And for some of you that most well know Miriam for work on cancer studies, it's really not limited to that. It's work on tissue homeostasis, inflammation, cancers, infectious diseases. So thank you so much, Miriam. I'm just trying to figure out-
I'm going to...
Oh, okay. Cool.
I'm going to speak up. Executive privilege. I was the only one able to use my computer.
Yeah.
It's inside my bag.
Oh, there we go.
Ah! All right, so, I'm going to switch gear and talk about, you know, a small lineage, small hematopoietic lineage called myeloid cells. Hope you've heard of them. So I'm going to talk about myelopoiesis. But before, I just want to remind you that while we spent, you know, the whole time talking about lymphocyte, important effector cells, most tumor, in fact, are dominated by myeloid cells, and particularly by macrophages. And this is what you see here, this is a human non-small cell lung cancer. It's full of macrophages. This is the tumor-free adjacent tissue. You can see the number of mac that you should see in a normal tissue, abundant macrophage, and we know that these macrophage are usually pro-tumorigenic because they dampen tumor immunity, because they remodel the stroma, and because they promote angiogenesis.
And yet there are no FDA-approved treatment to target these cells in cancer. So this is what my lab has been studying now for years. So first things we've done is to understand the origin of these cells. You know, many of us now have identified these two lineage of macrophages, tissue-resident macrophages that are part of the tissue. They renew locally, they are heavily imprinted by tissue cues. Then there is the monocyte-derived macrophages that accumulate in response to injury. So in this study that we published now a few years ago, what we wanted to do is to know whether those macrophages that accumulate in tumor are those the resident macrophage, or are they coming from the periphery? And we did. We used gene fate mapping of adult stem cells and in fact, several fate mapping models.
And to our surprise, I have to say, we saw that most of the macrophages in tumor lesions are in fact recruited from the blood circulation, from the bone marrow. So these are the tissue-resident macrophages that are mostly alveolar macs in the case of lung cancer. You can see that, in fact, they are excluded from the tumor lesions. And they are excluded, they don't die. They are just excluded from the tumor lesions. And when we look at the monocyte-derived macrophages, this was done in unbiased manner using the gene fate mapping system, we see a large accumulation of blood-derived macrophages.
Using some type of cross-species analysis and now a lot of datasets that have been generated in human cells, we know in human tumors we know how to recognize when in patients we don't do the same type of fate mapping, but we know how to recognize the tissue-resident from the monocyte-derived macrophage, and we see this is very similar in human. There is a reduction of tissue-resident macrophages that are excluded from the tumor, and there is a big recruitment of monocytes that differentiate into macrophages at the tumor site. So we've been looking at the molecular program of this monocyte-derived macrophage for a long time now, and we see that. Looking and searching, in fact, for commonality of program across tumors, because these will be the first cells that we want to potentially focus on.
We have identified, for example, the Trem2 macrophage program, first in non-small cell lung cancer. In fact, the first person to do this was Yonit Lavin, collaboration with the Ido Amit's lab. Now we found this program almost everywhere. These Trem2 macrophage we now show are quite suppressive. In fact, strongly dampen NK cell function at the tumor site. But this realization that in fact there is this big recruitment of myeloid cells and mobilization, in fact, myeloid cells remind us that cancer is, in fact, an inflammatory lesions. The definition of inflammation is the increased production, release, and recruitment of myeloid cells at the tissue site.
So the bone marrow is sensing something in the periphery that is leading them to expand this myeloid compartment and then releasing it in the blood. So we became quite interested in really understanding where the driver of pathogenic myelopoiesis in patients and in mice, and this work was done by Samaksh Hegde. I don't know whether he's here. He's actively looking for a job, so he's probably applying, but really remarkable fellow that was building, in fact, on recent work from Nelson Lamarche, who is now a faculty at CRCL, who showed that, in fact, in lung cancer cues were inducing the release of, in this case, IL-4, by inducing the production of IL-4 or potentially the expansion of cells producing IL-4.
These cells were mostly type two granulocyte in the bone marrow, and this bone marrow, the IL-4 production in the bone marrow, was in fact inducing a suppressive program early in the myeloid lineage. In fact, based on this data, Tom has designed a clinical trial to block IL-4 receptor in patients that were either relapsed or refractory in non-small cell lung cancer, and he'll talk about it later today. We started to really focus the lab. There is big effort now in my lab to really understand what was driving myelopoiesis in tumor-bearing mice and also in cancer patients. Here, Samaksh quantify this expansion of myeloid progenitor at the progenitor level. Different. These are hematopoietic stem cells. You can see the increase.
Here, the mice are injected with KRAS mutant p53 deleted lung cancer cells that Tyler Jacks developed. We've done this now with many other tumors, and you can see that there is a steady expansion of hematopoietic cell stem cells once with tumor growth, but we see also this big expansion of myeloid progenitors. These are early myeloid progenitors. These are granulocyte monocyte progenitors. This is granulocyte progenitors, and these are monocyte progenitors. So the whole myeloid lineage expand quite significantly, and Samaksh has also looked at patients with lung cancer, these patients with very small lesions, and he already see, in fact, release, excess release of hematopoietic stem cells in the blood circulation.
Hematopoietic stem cells are always released in the blood circulation, even in homeostasis, but usually they go back to the bone marrow, and you can capture some of them. But here, where you can see there is dramatic expansion of hematopoietic stem cells, which is a surrogate for the expansion that's happening in the bone marrow. So, what Samaksh then decided to do is to really understand the epigenetic cues that were potentially driving this myeloid expansion. So what he did is he used this combined single-cell ATAC and RNA sequencing along the myeloid lineage, starting from myeloid progenitor, monocyte, and tumor macrophages. And then he also did that in patients. But here in patients, he used mostly monocyte and macrophage at the tumor site.
He was looking for, in fact, transcription factor binding motif that were enriched along the myeloid lineage and shared between mice and humans so that we can probe causality in mice. This is how we've been working for a long time. What he saw, very early, in fact, is that there was an enrichment of this stress module in monocyte, from monocyte to macrophages. Here, I'm not showing you the bone marrow progenitors. It's a mistake because this is what I needed for this figure. Same in patients, we saw the stress module from monocyte to macrophage in the tumor. We became interested in the stress module because it was shared and because it was dominant. We dive into this.
So these are all the different modules that are enriched in myeloid progenitors in tumor-bearing patients or mice compared to naive. And we saw this dramatic upregulation of the stress module, but there are also many other modules that are expressed. And we decided to focus on stress because it was present in both mice and human and because it was quite dominant. And particularly, we became interested in this molecule, Nrf2, which was known to be an antioxidant, quite well described, in fact, in lung cancer lesions and in lung cancer cells, and known also to promote the survival of cells in a stressed environment. But not much was known in the role of Nrf2 in the myeloid lineage.
Nrf2 is a well-known antioxidant, promote the cytoprotective response to stress, and but also is known to significantly dampen inflammation, including dampening of interferon response. We became quite interested in this. The first things we did is we then first measure this motif accessibility of this transcription factor, Nrf2, along the lineage. What we saw is there was increased motif accessibility. You can see this along the lineage. Here it was post, not so active, and then we saw that motif accessibility increased along the differentiation, and that was associated with increased activation of this downstream gene module activated by Nrf2. You can see that there is increased activation, which is associated with increased accessibility of this Nrf2 transcription factor.
While Nrf2 was being increased, what we saw, there was a dampening of interferon response genes along the lineage, right? Which is something we anticipated from the literature. Okay, so then we decided to then flux or remove, delete Nrf2 specifically from the lineage, and we used this deleter mouse that was made by Florent Ginhoux, and that targets specifically the GMP, so granulocyte monocyte progenitor in the bone marrow. We remove Nrf2 only from early myeloid progenitors, and then we injected this same lung cancer lesion that I described earlier. What we saw is there was a dramatic, well, substantial reduction of tumor growth in these mice. We did the converse experiments, where we now deleted Keap1.
Keap1 is a negative regulator, promote the ubiquitination of Nrf2, especially in homeostasis. And when we delete Keap1, what we saw is increased tumor progression. So, we wanted to know whether we affected, in fact, this myelopoiesis. In fact, we did. What we saw is that just by removing Nrf2, which is one transcription factor, there are many others that regulate the stress response. We saw a reduction of this stress monocyte, CD88 positive monocyte, that supposed to be stress monocyte described by Stefan Jung a few years ago. There was also strong reduction of the stress monocyte in the blood, a reduction also, and a strong reduction of macrophages, including this TREM2 positive macrophage, which we know is quite suppressive in the tumor microenvironment. So, that's exciting to us.
And then we went back and looked whether this had an effect on the effector lymphocyte compartment, and we saw here a significant rescue of NK cell activity and a significant rescue of T cell activity. But we saw that at least in this model, when we depleted NK, we completely abrogated the beneficial effect of this Nrf2 deletion, suggesting that somehow the largest effect is probably mediated by reactivation of NK cell. I had to have these slides, you know, to fit in this session, so. Right, but then then the referee really wanted to know whether that early Nrf2 expression that we observed or induction had any effect on this phenotype. Because, you know, there is ROS everywhere.
You know, you have ROS at the tumor site and probably in the circulation and in the bone marrow, but I, is it playing a role? Samart has done this difficult experiment. Here, this is not so difficult, but useful. Here he co-transplanted Nrf2-deleted progenitors and wild type progenitors, same mouse, right, and then looked at how they differentiate at the macrophage output and the progenitor output in the same mouse, so same tumor microenvironment, just comparing Nrf2 cell intrinsic capacity to somehow affect the myeloid lineage. And just by deleting Nrf2, so we are in the same environment, he saw that he reduced myeloid expansion here. This is not always working, so reducing myeloid expansion and also reducing macrophage accumulation.
So then he did another experiment, where now he purified bone marrow progenitors from either Nrf2 no, sorry, he purified bone marrow from a naïve or from a tumor-bearing animals. I'm blanking on something. Okay, so now I remember. So he I thought we had an Nrf2 deletion experiment in this slide, but apparently not. So here, what we did is we take bone marrow from tumor-bearing animals or from non-tumor-bearing animals, and we inject them at the tumor site. And what we see is that in few days later, the bone marrow that has been somehow receiving tumor cues is giving rise to much more TREM2 positive macrophages, much more Nrf2 positive macrophages.
And here we also measure heme oxygenase, which is a downstream target Nrf2, which is significantly increased. So suggesting that bone marrow that receive tumor cues is going to produce macrophages that are much more suppressive and potentially also going to have the stronger cytotoxic response, which is in fact cytotoxic stress response, which is dampening further their inflammatory potential. And he showed the same thing in vitro. I think I'm missing the Nrf2 deletion, so I apologize to Samaksh here. But he did the same thing and showed that this is when he delete early versus delete late. This is what he wanted to show. So I think what he showed also is that when he delete Nrf2 early and versus late, early had the strongest effect in rescuing, in fact, macrophage suppressive activity.
So, the hypothesis that we have now is that, in fact, tumor cues induce two hits. You know, there is this big hit in the bone marrow that is inducing a suppressive program. This is mediated potentially by circulating oxidative species that are being sent by bone marrow progenitors, and this is going to imprint the suppressive program and contribute to further dysregulation that's happening in the tumor microenvironment. So it's really promote disengagement into a dysregulated differentiation. And we think that by targeting, right, early in, in myeloid differentiation, we could facilitate the reprogramming of macrophage at the tumor site. So this has been posted, but he's still working on identifying this first hit, which I didn't describe so well, but I hope I was able to articulate.
Okay, so this realization that now this myeloid cell expansion is contributing potentially to cancer progression, made us really think about aging, right? So aging is known to be associated with enhanced myelopoiesis for numerous reason, and cancer is a disease of aging. Now, it's been suggested that the main reason why age is the number one risk factor of cancer is because epithelial cell have this accumulation of mutation that is contributing to increased stemness and transformation potential. Okay, and in fact, there was a recent study posted in bioRxiv that surprised us. You know, here what they did is they took an old versus young tumor line isolated from GEM, so genetically engineering mice. Very similar to what I described to you, same line, but these were genetically engineered, so they take old-...
or young tumor, and then inject them into old or young recipient. And what they saw in this paper is that, in fact, the aged tumor cells have, in fact, limited stemness and tumorigenesis in the lung. Right? So if you put them in a young recipient, they, in fact, have no competitive advantage compared to an old tumor cells. And this is maybe because they don't have all this big metabolic fitness that we thought, you know, they will have because of this accumulation of all these mutations, suggesting that, in fact, the host is really potentially one of the main contributor of the cancer risk. Okay, so we like this study because we wanted to explore the contribution of aged myelopoiesis to cancer risk. So the first thing Matthew Park did, Matthew, I think, is in the audience.
He's not looking for a job yet. He just went back to med school. He's an MD/PhD student, graduated two weeks ago - no, two months ago, and now he's back in med school. And he recapitulated a lot of data in the literature showing that with age, you have this expansion of myeloid progenitor. These are the same myeloid progenitor that I described earlier, granulocyte, monocyte, progenitors, et cetera. You see them expansion in bone marrow. You see also expansion in the blood. And then he looked and also recapitulated some data in the literature showing that there is increased cancer progression in old animal. We are always working with the same lung tumor, but he's done that now also with - I'll show it, I think, later in different tumor lesions.
There is increased progression of tumor in old animal compared to young animal. Here, we work with seventy-two weeks or seven weeks, black six mice, and this was associated with also significant accumulation of myeloid cells. Here I want to point again to this reduction of tissue-resident macrophage. The tissue-resident macrophage always reduce in the tumor. We see an attrition of tissue-resident macrophage in aged tissue, and this is what he shows here, Matt. But really, the bulk of myeloid compartment is coming from the periphery. Then he did another experiment, really just measure the tumorigenicity of the immune system. He wanted to exclude the role now of the stroma of the tumor.
He takes animals, young or old, animals, and injects them with either young or old bone marrow. Okay, now the stroma and the tumor are either young or old, right? The only thing that's going to change is the bone marrow. Bone marrow is mostly hematopoietic cells. What he shows is that when he puts old bone marrow into young, tumor progressed much faster than young into young. Okay, so here he separates. He's separating. We're not saying he's separating the bone marrow compartment from the rest, from the stroma and from the tumor. He did the converse experiment, which is reassuring to me, that young bone marrow potentially can rescue or reduce lung cancer progression in old animals.
So if you have a mouse or something, maybe you can. Well, I'm not suggesting this, but this suggests that somehow a young bone marrow or an old bone marrow is contributing to tumor progression. You can either reduce it by giving just young hematopoietic system or aggravate it when you give an old hematopoietic system. Of course, this could be due to many things, including defective T cells, defective NK cells, but we became quite interested in looking at the myeloid compartment. Here, so we are going to try to understand why there is this aggravation. Are myeloid cells different in an old bone marrow or young bone marrow that accumulate in these young tumors, right?
What we saw is that, well, first, there is much more myeloid cells in an old, when your bone marrow is old, so these old myeloid cells are going to accumulate more. When you look at the expression profile, we saw that there was a significant accumulation of first myeloid progenitor. There is always some type of myeloid progenitors that accumulate at the inflammatory site, whether it's tumor or inflammatory disease or infection disease. This is what we call emergency myelopoiesis. The bone marrow sense the peripheral cue. They start producing more myeloid progenitors that are released, and these progenitors are going to accumulate at the tissue site.
So we always see progenitors, but here there are more myeloid progenitors, and they seem to produce very high level of interleukin one alpha, so IL-1 alpha, but also IL-1 beta, but IL-1 beta was much more ubiquitous. But in these old mice with an old bone marrow, you know, what stood out, it was this IL-1 producing, IL-1 alpha-producing myeloid progenitors. We verified that at the protein level, where we see, you know, increased production of IL-1 alpha by old myeloid progenitor in these mice.
Then we looked at the capacity of old monocyte to produce IL-1 alpha in response to challenges, in response either to tumor debris that they will encounter in the tumor microenvironment or to LPS. And here we see that these old monocyte produce much more IL-1 alpha, in fact, and beta in response to these challenges. So something is happening that they produce more inflammatory cytokine. And I have three slides here, Amir, if possible, 'cause Matthew is going to be very sad if I don't go through them, unless I. You want me to sum up? I can sum up. Can I sum up? It's okay. All right, I'm going to sum up here. I'm just going to say one thing. This is a super cool experiment.
So we try to see, okay, so what is driving this IL-1 production? And, and we started to profile, right, myeloid progenitors and, tumor-associated macrophages. And what we saw is that with age, even in the absence of CHIP mutations, so you know that CHIP mutation contributes to this biased myelopoiesis that happen after, happen in old age. And these, there are several, of them, Tet2 and DNMT3 being, you know, the most dominant. And what we stumbled into is that there was a reduction of expression of DNMT3A. Here, what Matthew shows is that there is, a reduced expression. I didn't show it, but also protein expression of DNMT3A in circulating hematopoietic stem cells in aged patients with lung cancer. So it's a cool experiment, okay?
I just want to point this, and I'm going to sum up. Here, what we show is that in patients with lung cancer, in older patients that are more than 65 years old, if you look at their hematopoietic first, they have more hematopoietic stem cells circulating in the blood. This is what we are showing you here. And then when you look at the expression profile of this CHIP mutation, you see they are reduced. This is what I show you here in green, compared to the younger counterpart. So CHIP, there is reduction, even maybe there is mutation marks.
Even in the absence of mutation, we see this downregulation of this molecule, which and the mutation is a loss-of-function mutation, which is so loss of function of this methyltransferase, which leads to reorganization of the chromatin, which may contribute to this increased production of IL-one. We see that DNMT3A, in fact, deficiency lead to increased expression of IL-one. Oh, no, I'm going to skip this. I'm going to skip this. This is what I wanted to show, that DNMT3A inhibition lead to increased production of IL-one by both mice and human monocytes. And then we blocked IL-one receptor signaling, either in mice that were injected with this KP line, and we see significant reduction of hematopoietic stem cell, reduction of this GMP, and reduction of myeloid cells and myeloid progenitor in the lung tissue.
Here we saw a dramatic also reduction of, in fact, lung cancer progression. I have to say here that IL-1 alpha blockade was also able to reduce lung cancer progression. IL-1 beta blockade did well, but not as well as IL-1 alpha blockade, but did well. And then when we combined the two blockade, we significantly reduced tumor progression. When we blocked signaling, we reduced tumor progression. So we are very excited about this. We are starting to expand this data in other tumor. Meant to say that we started the IL-1 blockade very early, almost as a prevention setting. And this is where it works very well. It works particularly well in colorectal cancer, progresses much faster in 72-week animal, and then progression is strongly reduced by IL-1 receptor signaling blockade.
This is the sum up that I wanted to do, but I think you got it. That we think that during aging, there is a reduction of DNMT3A. There is potential mutation in some patients. This reduction is going to lead to increased production of IL-one in response to injury. We are working in molecular mechanisms that are leading to this in collaboration with Nicolas Vabret . This increased IL-one production is contributing potentially to stromal remodeling. But what Matthew described in paper that has been published is that he thinks it's acting also in early myeloid progenitors and potentially in hematopoietic stem cell to further the myeloid dysregulation and contribute to myeloid cell expansion and dysfunction at the tissue site.
Based on this data, Tom Marron has now designed a clinical trial where he's going to test a combination of IL-1 receptor and IL-4 receptor blockade in patients, hopefully in aged patients mostly, that are relapsed refractory, and we are trying to do it in aged patients. But what we are really mostly excited about is to start thinking about immunoprevention. As you remember, you know, with this big CANTOS study that gave IL-1 beta antibody blockade to 10,000 patients with cardiovascular event, there was significant reduction of diagnosis of lung cancer incidence. Here, what we are hoping to do is to identify a group, hopefully with Laurence Zitvogel and Gustave Roussy, at risk for lung cancer.
Age would be one of them, DNMT3A, could be one of them, inflammatory myelopoiesis, the site of myelopoiesis is something that is being in discussion. And then they will be treated not only with IL-1 beta blockade, but hopefully with IL-1 alpha and beta blockade. And we will. I forgot to say that we will select patients that have already pre-cancer lesions or nodules, so that we also will select a group that is at risk. So smokers with nodules cannot go to surgery or radiation, and then we will look at the progression of these nodules. Problem is to find funding for this type of study, because not many people. Well, in pharma, if you want to fund us, we'll be very happy of funding. It's very difficult to obtain funding for prevention studies.
With this, I'm going to end, and, I'm sorry, I, it was much longer than I thought. My lab, Sam Martin and, all the people that contributed, Tom, Robbie, fantastic lab.
... Okay, fantastic talk, Miriam, as always. Just one question. You're talking about this kind of crosstalk between the tumor and the bone marrow. Do you have any idea of
What is inducing it?
the tumor cues
Mm-hmm.
-as you call them, that impact on hematopoiesis-
Yeah.
-and myelopoiesis?
Yeah.
You talked about tumor debris, but-
Yeah
How do they circulate?
Okay. Well, first, so we are looking at that, of course, and, but I think it's going to be a multifactorial effect, right? It's not going to be one factor inducing. It's not possible, right? So we look at these. So if you take the supernatant of the tumor, or you take the sera of the tumor, and you just put it on bone marrow, you see some of these program. So we know that there is something in the blood circulation that is inducing these. What Samart is doing is that he's fractionating, looking at whether it's metabolite or protein. So we thought it would be metabolite. Sure. Of course, it was not. It seems that some type of protein is driving the Nrf2 program. The, for IL-4, program, so, you know, we are looking at many of these program, right?
So we saw this IL-4 program, which we think is inducing this repair program early on. We now are very focusing on the Nrf2. The Nrf2 program is interesting because we see it also in inflammatory bowel disease, big Nrf2 program in circulating monocyte in IBD patients. The same type of response, right? So the bone marrow is sensing, and they sense with different cues, right? So I think it's not going to be very easy to find one target. What's clear is that sensing happens, otherwise you will not be releasing these myeloid cells, and there is a whole group, you know, in IBD and cancer, there's the Nrf2, but probably many other things that are going to drive myeloid differentiation. Laurence. She came from Paris just to attend this symposium, so
Yeah, so, what is the hierarchy between IL-4 and the Nrf2? Because I had the understanding from Thomas Marron, that basophil secreting IL-4-
Yeah
... would be upstream to trigger the progenitor myeloid IL-1 program. So IL-4 is upstream of Nrf2 or not?
At all. Are you talking about IL-1 or the Nrf2? Yeah.
Nrf, Nrf2-
Yeah
... is driven by IL-4 or not?
It contributes to it. What, when we delete IL-4 receptor from the Ms4a3, we see a reduction of that program.
Okay.
Whether it is gone, guys? No, it's not gone.
It's not gone.
So I would like to have confirmation from the team here, but they're not here. It's not gone.
It's not gone.
Yeah.
No.
But we see a reduction. TREM2 is the same. You know, TREM2 is not really quite induced by IL-4. Marco had shown it many years ago, but we see the same thing. TREM2 is quite driven by Nrf2. When you dampen IL-4, we see a clear reduction of this program. There are potentially other cues.
Why do you say that the supernatant of tumor cells is a prime stress here? Because when the ileum is sensing the tumor, and so the gut-bone marrow axis is also very dominant. Actually, we are doing the experiment with the MS4A3, and we can see the exodus of bone marrow progenitors following permeabilization of the gut when in tumor bearers. Why do you say obligatorily that the molecular cue or the cellular cue is coming from the tumor?
Mm-hmm.
Would it be possible that it comes from-
No, no. Of course, it's possible that,
It comes from somewhere else.
Yeah, the gut is. Yeah, we could look at these. But that gut also, it could be inducing something that is perfusing the tumor, you know, and entertaining, right?
Right.
It's possible-
Right
... that your gut response is somehow also perfusing your tumor and, and creating even more damage.
Yeah
... but it's testable, I suppose, 'cause we take both, and we could compare.
Take one more question.
Oh, thank you. Miriam, what a beautiful presentation. Really very elegant-
Thank you.
Congratulations to your team.
Matthew agrees, but,
Yeah
... he would have done a much better job.
Congrats to your team. I mean, incredible. Very difficult experiments. I have a very naive question because, of course, you know, myeloid cells can be stimulatory versus suppressive. So when you make these shifts and you prevent, for instance, the trafficking of these myeloid cells with IL-4 or IL-1 inhibition, do you think that there is any shift in the in a suppressive activating phenotype, or it's just a block in the trafficking? And then if you're losing myeloid cells, that could potentially be behaving like APCs. Would that have an impact on the ability of the-
Yeah
... endogenous T cell for response?
Yeah. So I don't think it's only. So there is a reduction of myelopoiesis, production of myeloid cells, right? Don't know whether we are really reducing recruitment, and because we see that reduction already in the bone marrow, and we see also a reduction of blood monocyte. So there is a reduction, somehow, myelopoiesis, which we'd like to explore further. So now, the program are also changed because, you know, when I talked about TREM2, that TREM2 is really a molecular program. We call it TREM2 high, but it's like we define it based on 15 genes. Those is really strongly reduced, very clearly.
So we are not only changing the number of cells or their recruitment, but also the ability to respond to some of these cues and then do and have some of these suppressive program.
Thanks so much, Miriam. Thank you. All right, everyone, so it's 3:09, so we're just gonna add nine minutes onto the clock. But it's my great pleasure to introduce Dr. Tony Ribas. Please. So, Tony is visiting us from UCLA. He is a professor there, director of the Tumor Immunology Program at the Jonsson Comprehensive Cancer Center at UCLA. And very briefly, Tony has been one of the very biggest pioneers of cancer immunotherapy to date, dating back to the very earliest trials focused heavily on melanoma. Since then, he's advanced our knowledge, both in terms of understanding anti-tumor T cell functions in this context, but also trying to understand why things don't work for some patients and really starting to dive into further pioneering the combination space.
So I don't wanna take up too much more of your time. Thank you so much.
The laser pointer, we need more batteries.
Yeah. So I'll become-
Yeah. Thank you very much, Amir, for this very kind introduction. I looked at the program, and I saw the people who were here, so I said, "Well, I'm not going to show up with my regular PD-1, blah, blah, blah. This works this way or that." So I'm going to present data that most has been generated by an MD-PhD student, and most of it it's very recent and unpublished, so it may not be my smoothest presentation I've given. I have a series of interactions with industry. I think none of these are relevant to this presentation, but please take them into account.
There's something that really gives me pause, which is there's some patients who have a tumor that has a genetic mutation that leads to no B2M expression, and B2M is the light chain of MHC class one, and then MHC class one is not stable on the surface of cancer cells, and some of these patients still respond. Here I put a series of references of papers, including one of us, where there's well-documented homozygous loss-of-function mutation or copy number neutral loss of heterozygosity of B2M, leading to no B2M, no MHC class one expression, and still those patients respond to therapy. I know for sure that B2M is really important for responses to anti-PD-1.
Here I'll show you a data that I've presented several times, but this is a case of a patient with a metastatic melanoma in the pelvis that responded and had an objective response for three quarters of a year, and then the tumor progressed and progressed somewhere else. We compared these two tumors with the original one and see what was different. Here it's different at the cancer cell intrinsic level. We sequence the genes that we can try to look for mutations that are new in the tumor. This is work from Jesse Zaretsky from years ago, and this is what we call a circos plot. This is chromosome one, two, three, four, five.
You go up to twenty-two, and here's heterozygous mutations are all around here. The shared mutations between this tumor and this tumor are in gray. The lost mutations are in green. And then we're going to focus on red dots, which is new mutation. We're going to focus on a new mutation that's homozygous. And I'm going to come here, and Amir, you're going to be my volunteer. Let's see. This stopped working.
I can be your pointer as well.
Okay. You're not color blind, are you?
I'm not.
Okay, good. Thank you. I'm sorry. That's too much personal information. So biallelic frequency of one would be that both alleles are the same, and if there's a red dot in this track out here, it would be a de novo homozygous mutation. So I'll go around, and then you stop me if you see any red dot. It's really difficult for a cancer to make a de novo homozygous mutation. Here's this one dot. Is there another one? Is there ano-
I don't see another one.
Okay.
Yeah.
So you take Amir's word, not my word. There's only one new and homozygous mutation in the resistant tumor compared to the baseline tumor out of twenty-two chromosomes and twenty-two thousand genes. This one new and homozygous mutation was a loss-of-function mutation in B2M when one allele had been lost and the other one was mutated and amplified, so we have two copies, so it's more stable. Thank you very much, Amir. So,
I'm available.
Amir demonstrated that if you lose MHC class I expression, then CD8 T cells do not recognize the cancer, and that leads to resistance. We can do it experimentally when we take a tumor like MC38 that responds to anti-PD-1. We knock out B2M, and then we give anti-PD-1, and the tumor no longer responds, which is the logical thing. Why would that be? Because if we have the T cells here that recognize tumor antigen, and once they recognize, they make interferon gamma and lead to the 2,000 interferon gamma response genes. If there's no B2M, there's no way that CD8 T cells recognize the cancer and do any of these.
But we just said on the background that occasionally there's cases of patients who have tumors that have lost B2M at baseline before they received anti-PD-1, that responded to PD-1 blockade therapy. So we wanted to model this, and our thinking is that in steady state, CD8s are the main cells that are inducing an anti-tumor response. They recognize antigen presented by MHC class I. That leads to the production of interferon gamma, that leads to the cancer cells making 2,000 interferon response genes that change the tumor microenvironment, and everything that was bad in the tumor microenvironment then turns into good. The macrophages that were bad now are making good genes.
And that from Don Schumacher's data goes up to forty layers into the tumor from where the T cells are making interferon gamma. But if the tumor expresses no MHC class I, then the CD8 T cells cannot recognize and cannot do anything. But we thought that what if we stimulate other effector cells, NK cells and CD4s, and could we reverse this resistance induced by the loss of MHC class I? So to do this, we turned, at that time, was a promising agent which is called Bempeg, which is an IL-2 that was pegylated, where it binds to a IL-2 alpha and then loses the pegylation after administration to patients.
And then with time, you can get an NK and a CD8 and a T cell response. So in work that we presented a while ago, we take the B2M knockout MC38, that this no longer responds to anti-PD-1 in red. If we give Bempeg, these tumors respond, and we give anti-PD-1, but with this variant IL-2, then we get even better response. So even in cases where there's no MHC class II expression, we can get anti-PD-1 working in a mouse model. And then you just deplete the different cells that could be inducing this response. And here, if we deplete CD8s, nothing happens. But if we deplete CD4 or NK cells, there's partial loss of the anti-tumor activity.
If we deplete both CD8s and NK cells and CD4s and NK cells, then there's a complete loss of the benefit of adding this IL-2 variant to induce an anti-tumor response and B2M in a B2M knockout tumor. Which takes us into thinking, well, in responses to anti-PD-1, in occasional situations where there's the CD8 do not encounter the target because there's no MHC class one expression, then NK cells and CD4 cells may be may be effectors. So that's when Mildred Galvez, who's an was an MD PhD student, actually finished her PhD, so she's Dr. Galvez. I started working on these systems and mostly looking at data from patients to try to see what happens in these in these tumors that lost MHC class one expression.
She took a data set that was curated by Katie Campbell, who's now a junior faculty at UCLA, was a postdoc in my lab at that time. Katie had harmonized the clinical data sets of exome sequencing and RNA sequencing from three Checkmate trials, Checkmate 38, 64, and 67, and then three series that were published in the literature, the Gide, the Lee, and the Van Allen. Then she set it up so anybody can go to this paper and then just click on things. I don't know how to do any of this, but she tells me that, and now that all the data is harmonized, then you can merge these data sets and then get answers if you have a question.
So Mildred asked the question: If we look for the B2M, look at B2M mutations, either somatic single point mutations or copy number alterations, that we could understand what happens in those, and then look at the expression, look at the RNA seq and see which cells are changing in these, in these tumor biopsies. So by doing that, she looked at all of these biopsies, and here we have color-coded in green, the patients who have a response to anti-PD-1 therapy, stable disease or black progressive disease. And here's the B2M unaltered, and there's a distribution. There's more gains of B2M than losses of B2M. Who knows what that means?
But that's what happens in this data set, and these are all melanomas. The distribution of complete objective response is stable and progressive disease the same. But if we look at loss of heterozygosity, so loss of at least one of the two copies of B2M, we see that there's an enrichment in progressive tumors and less stable disease or complete or partial response. There were only two tumors that had homozygous loss of B2M. Those two were patients who had, at best, a stable disease response.
So if we then distribute these based on B2M loss of heterozygosity versus a response assessment, that there's a p-value suggesting that the more the loss of B2M, I'm sorry, patients. Loss of B2M in the tumor leads to patients with less responses to anti-PD-1 therapy. The B2M, if we look at the expression of the gene, the B2M loss of heterozygosity tumors have lower transcripts of B2M, which makes sense. So this suggests to us that in these tumors that have lower copy numbers or at least loss of one of the copies of B2M, then we may have an alteration of the immune response because they have less MHC class one expression.
So we wanted to see if we look at those tumors, is there differential, is there a change on the immune cell infiltrations in those tumors? That's done by looking at the RNA seq and looking at doing CIBERSORT, which is a way of, from the expressed genes, inferring which cells are in those tumors. So here's an unreadable slide. Here's a whole bunch of different immune cells that you can get from the CIBERSORT deconvolution of the data. In blue is B2M unaltered, in red is B2M loss of function, a loss of heterozygosity. Then up here is non-significant, and there's only one that's significant.
And the one that significantly change, it's NK cells that have activated phenotype, which are increased in the B2M loss of heterozygosity cases. The next two, without being significant but close to significant, are CD4 memory resting cells and macrophages, M1 macrophages, both of them with increased numbers in the B2M LOH tumors compared to the unaltered ones. Does this have anything to do with responses to immune checkpoint blockade therapy? Here we have a similar graph where all of these CIBERSORT immune cells, and then and then I now separate by CR or PR objective responses versus disease progression.
That's focusing only in the cases that have B2M loss of heterozygosity, and we can see that, the one-- the only one that's significant is CD4 memory activated cells. They're increased in these, in these patients with disease progression. And the next ones are activated NK cells and M1 macrophages, all with the same trend, that these three cell types are enriched in non-responding tumors in patients who have B2M loss of heterozygosity. So what else is there in these tumors? Well, if we continue with the RNA seq, we can ask the question, which genes are more expressed in the B2M loss of heterozygosity, LOH tumors, compared to the ones that are unaltered?
And if we look at the ones that are unaltered, the highest gene, that's an outlier, it's a non-coding RNA, so it makes no protein, so I have no idea what that does. There's no literature on it, but it's clearly way out there. But if you look at the other side, which is the increased expression in the B2M LOH tumors, there's actually a gene that may resonate to the NK biologists in the audience, which is HLA-G. That is by far the most highly expressed transcript in these B2M loss of heterozygosity tumors. And if it wasn't for this one, this would be way up there.
We look at the expression of this HLA-G, and actually it's a marked increase in B2M LOH tumors, and this is a non-classical MHC class I molecule that has been attributed many roles. It's expressed in the placenta as the main tissue type, and then some cancers express it, and when it's expressed by cancer cells, it leads to a worse prognosis. Its ligand is KIR, so it turns off KIR NK cells by the expression of this non-classical MHC class I. Does this have anything to do with responses or resistance to anti-PD-1 therapy?
Actually, we can continue doing bioinformatic analyses and separating tumors based on the expression of B2M or not, and loss of heterozygosity of B2M. Then we could see that the HLA-G expression is higher in the LOH B2M LOH tumors, and is significantly higher in the PD-1s, where here there's a spread, but in biopsies of patients with disease progression, there's much more HLA-G expression.
And if we plot this in two ways, and we look for the expression of HLA-G with the presence of activated NK cells, which would be their ligand, we can see that it's flat in all of the comparisons in the B2M unaltered, but in the B2M LOH tumors, there's a linear correlation between the expression of HLA-G and the presence of activated NK cells, and that's only in tumors of patients who have disease progression. So what the thing is happening is that in the majority of cases, we get CD8 T cells that recognize cancers by MHC class one. We block PD-1, and we get responses.
But with selective pressure, the cancer does something that usually does not want to do, which is lose MHC class one expression, because if it loses MHC class one expression, now becomes a target of innate immune responses. And that leads to resistance, but NK cells can respond to anti-PD-1 therapy, it seems like, and then it can lead to a response in these few, these mostly anecdotal cases where there is, there's 15q21, B2M loss, exemplified by the bioinformatic analysis showing that activated NK cells are increasingly expressed in there. But since the tumors that progress express HLA-G, our hypothesis would be a logical one, which is HLA-G is turning off these NK cells that were trying to control the tumor.
That was just looking at one of the genes involved in MHC in antigen presentation, in MHC class one presentation. What if we look at all of the genes that have been described to be involved in controlling MHC class one? Could we find some genes that when they're not expressed, they lead to decreased responses to PD-1, to the immune checkpoint blockade therapy? So to do that, Mildred curated a list of 164 genes, which were all of the genes that are described as being involved in the MHC class one antigen-presenting machinery, all of the interferon response genes, and for some reason, the MHC class one transactivator, NLRC5, is not on either one of these lists.
So much for the gene list that people publish because this is the master regulator of MHC class I expression, so we had to manually add it. And then, I don't know what this is, but we'll turn it off. Cancel. Cancel is a good thing. So then we will look at mutations on any of these genes, and then next I'll look for the changes in the expression of proteins. Oh, so now I've really canceled myself. Okay.
Of these 164 genes, we looked at some mutations in the genes, and we looked at the number of mutations on each one of these genes, and then we need a waterfall plot, where on top is the gene with the most mutations, on the bottom is the gene with the least mutations, at least the ones that we can plot. And here's the percent of mutations. NLRC5 is a long gene, so a long gene, if you have a tumor that leads to a lot of mutations, may be by chance having more mutations. But if you look at the top two, it was NLRC5 and CIITA. And these are the MHC class I transactivator and the class II transactivator. And we can put a box through them because they're mutually exclusive.
They're mostly, there's very few that have both, and there's very few that have both here. It was the top mutated gene, two genes in this long list of MHC class one and class two control genes. And if we look at silent mutations and non-silent mutations, silent mutations do not change the amino acid listing, so they don't really matter. Non-silent mutations change the sequence of the protein, and those would be the ones that matter, but both of them lead to a P value that suggests significance of being mutually exclusive, see NLRC5 or CIITA mutations. So NLRC5, that's, as I said, a long gene, and it has this NOD-like receptor sequence that binds to nucleotides and does it by a domain called NACHT, N-A-C-H-T.
We wanted to see, are the non-silent mutations and the silent mutations equally distributed or distributed differently in this long gene? The red part here is the NACHT domain. The rest is the NLRC5 gene across chromosome 16. On the bottom in red are the silent mutations. Those are the mutations we do not care about because they do not change the protein sequence. On top are the non-silent mutations, and we can see that they're clearly clustered in this NACHT domain, suggesting that there has to be a selective advantage to mutating this part of the gene.
Then we look at the expression of the gene. We see that the non-silent mutations lead to lower expression of NLRC5. So what is the result of this? Well, we want to see if this happens. Are these patients who have these tumors with these NACHT mutations on NLRC5 having more or less responses to PD-1 blockade therapy? First, we wanted to see what does it control? I'm sorry, I forgot about this. So that was the logical next question. So if you have NLRC5 mutations that are non-silent, they should control HLA A, B, and C, beta two microglobulin, and TAP1 . All of these are NACHT-controlled genes.
ERAP1 is another antigen-presenting machinery gene, which is not controlled by NLRC5. So if we look at the expression level of these different genes, we can see that the five that are controlled by NLRC5 have lower expression in NLRC5 mutant tumors compared to a wild type, while the one that is not controlled by NLRC5 has the same level of expression, suggesting to us that these mutations have a functional role leading to lower expression of MHC class one genes. Does that lead to changes in responses to immune checkpoint blockade therapy?
So here we have in green objective responses, CRPR, the few that have stable disease as best response on therapy, and that, the biopsies of patients who have disease progression, we can see that NLRC5 mutant, even though there's very, very few of them, we can see that, that the, there's lower expression of NLRC5 in NLRC5 mutant tumors, and we do not see that in the NLRC5 wild type. If we look at the expression of B2M, we also see decreased expression in NLRC5 mutant in patients who do not have a response to therapy. And if we look at the gene that's not controlled by NLRC5, ERAB1, we do not see this trend.
Suggesting to us that the control of NLRC5 express, NLRC5 controls the this gene expression and leads to lower responses. The other, if we look at the patients who have disease progression, and we look at the Notch domain mutations, they're all clustered here. There's a lot less cases, so it tells us that non-responding tumors with these mutations in the NACHT domain are more likely to have disease progression. So this is what it has the same meaning. Just to finish up, the CIITA, which has a similar structure, is the class two transactivator. It also has this Notch domain.
If we look at the mutations that are non-silent, that change amino acid, they're clustered in the Notch domain, as opposed to non-silent ones that are distributed along the gene. So, in conclusion, this data suggests to us that B2M is important. It leads to increased disease progression and acquired resistance to anti-PD-1, but the loss of MHC class I expression attracts activated NK cells that could lead to clinical responses in those few cases, but that is limited by the reactive expression of HLA-G by cancer cells. NLRC5 and CIITA, among the MHC class I controlling genes, are the most frequently mutated genes. They are, and their mutations are mutually exclusive, telling us biologically that that has to be important, because otherwise they would be equally distributed.
They are mostly in this Notch domain, which is the active binding to nucleotides of these transactivators. So this is mostly the work of Mildred Galvez, some of it with David Torrejon and Katie Campbell. And here are the sources of funding. And I'll be happy to take questions if I can answer them.
Yeah, thank you so much, Tony. That was amazing. I'm just gonna get one question in quickly, one question and a comment. So the question is about loss of HLA class I. Only certain alleles encode ligands for NK receptors, and it's actually a form of education when they're developing. So my question for you is, would you expect those differences, the delta, to increase if you were to actually reappraise the germline first for educating alleles and then see which alleles are lost? And then just one comment, the HLA-G, while it does indeed encode a ligand for KIR2DL4, it very specifically regulates HLA-E expression in the absence of HLA-ABC. And so you could only engage HLA-E at the protein level at that point, but it's worthwhile seeing if it's truly KIR2DL4 or if it's actually a secondary effect on HLA-E.
Amir, it's clearly that you know a lot more about all of this. That's why I presented it here.
I am interested, though, to know if you looked at the alleles.
Yeah. No, no, that's great.
It's beautiful. Like, it's going in the right direction that you would expect outcome wise. Have you guys considered appraising the alleles, though, around educating alleles? That's my question.
The thing is that all of this is bioinformatics, and it's based on the biopsies and analyzing the biopsies. We were surprised that we didn't find mutations in HLA genes. They were very uncommon, but copy number alterations in them. Mostly, they were amplifications of one or the paternal or maternal allele, but very few losses, at least in this data set. B2M losses were more frequent, and at least one of the two alleles at a loss. I don't think we can. We analyzed how that would match into different HLA-A, B, or C alleles, and if there's-
Okay
... Cw7 was up or down in any of them. I don't know. Lauren?
So, Tony, it's a trivial question. So, I miss the point whether it's a primary resistance to immune checkpoint blockade or secondary resistance to immune checkpoint blockade.
Yeah. The only secondary resistance I showed was the case with the, where we had the Circos plot that Amir called-
Okay
... the loss of-
The check-
homozygous loss of function mutation of B2M. All of the rest were patients who were treated with mostly nivolumab or nivo-ipi, in this series, and those were baseline biopsies.
So do you think that primary melanoma without immunoediting would also exhibit this type of MHC class I loss?
It probably depends on how immunogenic was that primary melanoma. These were all metastatic.
Mm-hmm.
- melanomas. If we look at the data, for example, the recent paper in Nature with 2000 whole genome sequencing of patients with colorectal cancer, we can see that B2M loss is more common in MSI-high tumors.
Yeah.
Which makes sense.
Yeah.
If it's more immunogenic, some of them may want to get rid of the HLA expression. Yeah?
Have you looked for mutations that affect the transcription either in the promoter or in the enhancer? Because in the autoimmune context, many of the GWAS hits are actually hits like that.
No, we looked at that. In these studies, we looked at the part of the gene that encodes the protein. Yeah, the exons. This was whole exome sequencing.
I have-
I think they'll need a mic. Who has the microphone? Oh.
Yeah. Yeah. I have a question.
Uh, oh.
Definitely here.
Oh, there. There you are.
Yeah. So, is there any underlying condition like other than cancer, for example, autoimmune, where the regulatory T cells become cytotoxic effector cells? Have you looked at that?
No, but there's going to be a Treg talk at the end, so maybe that will answer your question. I don't know. I don't know. I trust-
Great talk. I have a question. So there is now increasing evidence showing that the CD4 T cells in the tumors, in human tumors, are, like, making all these killer molecules in interferon gamma and all that, making granzymes, perforins, and in fact, more, they're more potent in making these molecules. Have you seen something like that in your datasets?
Yeah. So these tumors that have loss of MHC class I expression because of B2M LOH or decreased expression, they have more CD4 cells. They are the same as activated NK cells. We have done no mechanistic studies to see what they do. Maybe the last question?
Yeah, sure.
Thank you for the nice talk. I had a question on the tendency of increased expression of HLA-G in tumors that has loss of heterozygosity in B2M gene. So, you mentioned that HLA-G is highly expressed when these tumors lose B2M gene. So I was wondering whether this is just a consequence of selective pressure of NK cell activity, or is there, like, a mechanism that leads to activation of HLA-G expression when B2M is lost?
I wish I could answer your question. I cannot. This is all descriptive data from bioinformatic analysis of biopsies. The hypothesis will need to be tested. The problem is that it has to be human systems because the mouse is so, so different.
In pregnancy, GM-CSF is very sensitive at driving HLA-E, HLA-G on trophoblasts. It hasn't been studied well in cancer, but that might be a corridor.
Thank you.