Bristol-Myers Squibb Company (BMY) Earnings Call Transcript & Summary

Hi, everyone. My name is Matt Phipps, analyst here at William Blair. Thank you for joining us for this panel on the kind of cell therapies in the next decade. We just heard some exciting work being done at MSK on AML. And now we have a nice list of panelists here to kind of show what they're doing and talk about some of the key topics. And so we'll start off with a quick introduction, a few slides for some of the people and then switch to the kind of more fireside discussion. And if you have any questions, please submit them in the chat, and we'll try to get them in. We have a lot to discuss, so let's get right into it. We're going to kick it off with Rafael Amado, the EVP and CMO at Allogene Therapeutics. Rafael, go ahead and please start.

Yes. Thanks, Matt, and thanks for the invitation and from being surrounded by great people. I always learn a lot in these meetings. These are just the legal disclaimers. What I wanted to do in this initial introduction is just show you a few slides about what we do at Allogene and what direction of travel we're taking. Essentially, we are a company that uses donor T cells allogeneic sources of CAR T cells to treat both hematologic malignancies as well as solid tumors. So we have shown both we manufacture, we can gene-edit these cells to try to prevent the [indiscernible] disease and to delay rejection. And that we can get responses in diffuse large cell lymphoma and follicular lymphoma and that those are durable. As you can see on the far left, we have a 6 months CR rating diffuse large lymphoma of 36%. We're now testing consolidated dosing, which is something that can be done. We topped the cell therapy because the cells are ready on demand. And so we're dosing patients at day 28 if they're not in disease progression to hope -- with the hope of increasing responses and durability. And of course, we can redose patients if they fall out of response. And oftentimes, we get responses that are longer than the initial response. And we are doing that in hematologic malignancies and we had just started first sort of foray into solid tumors with anti-CD70, and I'll say more about this when we go into the pipeline. Next slide, please. So this is a quick comparison of what we saw with in large lymphoma and follicular lymphoma. But particularly in large cell lymphoma, the overall response rate with single dosing in both follicular and large cell was 64% with a complete response of 46%. And then juxtapose to this are the responses that have been seen in Phase II studies. Importantly, the responses -- complete responses in the MITG and ITT are very similar. But that's not necessarily the case with other products. I know that manufacturing is getting better in the autologous setting, but there's still a lot of patients that require bridging therapy or that don't have a suitable graft or are unable for one reason or another to receive therapy. So one does the ITT, you can see a drop out in the response, which is not something that we observe as our therapies are given essentially on demand. The durability is expressed here at CRS, 6 months. And you can see, as I said before, that is 36% on par with what has been seen before. Importantly, these allogeneic therapies are very well tolerated with grade 3 events of CRS and neurotoxicity that are very low. And the concern was because we used anti-CD52 to deplete the endogenous T cells that are attempting to reject the graph that we were going to see high levels of infection. As you can see, the infection rate is not much higher than what has been seen with traditional autologous therapies. Next slide. So of course, the Holy Grail is to be able to get into solid tumors. Those are the large majority of malignancies, the ones that kill the majority of patients with cancers. And we are doing this by validating and selecting targets well, optimizing cars using multi-targeting cars and binders by increasing the diesel fitness, by using something called TurboCAR, which is the third signaling that we give to cells, improving manufacturing to get cells that are younger and less prone to exhaustion. By improving the immunosuppressive microenvironment with next-generation TurboCARs to have PD-L1, TGF beta and other molecules that are dominant negative to try to sort of sweep the tumor microenvironment from these immunosuppressive factors. And at the same time, a signal through better chain cytokines and then increasing tumor trafficking with combinations and other engineering into the CAR T cell molecule, which is one of the advantages of the shelf, which is one can do multiple engineering and genetics. Whereas in the autologous setting, one is limited to a single gene transfer, which is normally done with viral vector. Next slide. This is my last slide. I only want to refer to the pipeline that is in Phase I. Our 501 study was our first study in non-Hodgkin's lymphoma and it's the data that -- the basis of the data that I showed you before. They all had to study the same CAR, but the void of the sweep side switch. And we have data that we are accumulating in non-Hodgkin's lymphoma as a new molecular entity that in virtue of the fact that it has manipulations, not to the CAR, but to the molecule itself at removing the Swiss side switch. And we are doing studies in consolidation only, and this is really a molecule that we want to take forward this year in pivotal trials. So this will be our first pivotal trial and 2021 will be a pretty important year for us. In BCMA, we're starting in multiple myeloma with the universal study. We presented data at ASH last year. We want to update the data on both the Phase I and the expansion by the end of the year. We have a sub-study with , which is a gamma secretase inhibitor, which is currently accruing to increase BCMA density in the cell [indiscernible]. And we're pretty excited about the initiation of ALLO-605, which is our first TurboCAR. This has already started, and it's essentially the same car as 715 but containing a signal 3. We have molecules in AML. And then we've disclosed some targets in solid tumors, such as CD70 in renal cell carcinoma. This is in Phase I, and it's currently accruing and accruing very well. So we are learning a lot about the migration, the trafficking, the tumor microenvironment of CARs in solid tumors and perhaps we can chat about this during the session later. DLL3 is possibly our next target. We've seen the data with bispecifics. It looks good. And we have a CAR that we plan to move into the clinic. And then we have a number of other undisclosed targets and underpinning this is our ability to suppress the T cells to avoid rejection with an anti-CD467, which is an anti-CD52 antibody that is T-cell depleting. So with that brief introduction, I will just pass it on to the next speaker.

Thanks. Next, we have Christian Itin, CEO of Autolus Therapeutics. Christian, go ahead.

All right, Matt. Well, thanks a lot for having us. So I'd like just to focus very briefly on 2 of our programs to give you a sense of kind of how we're moving with the overall pipeline. I think there's opportunity to touch on some of the other programs as we go through the conversation. So I'd like to just briefly introduce to you a program we call [indiscernible] or AUTO1 and what we're trying to do with this program, which is targeting acute lymphoblastic leukemia and particularly focused on the population is to combine a product or create a product that combines a very high level of clinical activity with very long persistence to match the stem cell-like nature of the driving cells behind the disease as well as combining that in addition with a well-tolerable profile, the very manageable profile given that these patients tend to be rather beaten up from the prior lines of therapy. But also as a consequence of the underlying disease tend to actually have significant challenges, tolerating high levels of toxicity. So what I'm basically what we're showing here is that we obviously have a very active program that gives us a high level of molecular CRs in the range of 85%. What was very encouraging is we have no patient experience high-grade cytokine release syndrome in this disease setting, which is a very challenging one where you can have as much as 90%, 95% leukemic blast in the marrow and a limited number of neurotoxicity. When we look at the event-free survival, it's been very encouraging to see was that, in fact, the curve starts to go horizontally at around 12 months, which obviously is one of the hallmarks that we're all driving up in this disease setting where we're aiming for long-term remissions in these patients. And frankly, you might our life who were not able to do that with a bispecific T cell redirecting antibody, Blincyto, blinatumomab. Now what's obviously underpinning all of that is the exquisite persistence that we're seeing with the product. And as you can see at the bottom, this is a time line going to 24 months is that we have very nice persistence across that entire time period, which gives us an ability to put pressure on these leukemic driver cells that are very hard to actually come by and to keep in check. Building on the experience and this set of profile, we also obviously looked at pediatric patients on the next slide. And one of the key things that we did find where we're testing auto or we sell in pediatric patients, is that while we have very similar levels of CR rates and molecular CR rates and very similar event-free survival. Actually, we do have a significant proportion of the patients relapsing due to antigen loss. They lose literally the CD19 antigen on the surface, which has also already reported for Kymriah. If anything, we see we're almost selecting more patients with this highly potent approach. Now similarly to what we had seen in the adult population, no high-grade CRS, very limited neurotoxicity. And so what we basically went in with building on this particular backbone of the AUTO1 program, actually adding a CD22 CAR that we highly optimized to address the complex nature of the structure that the CD22 antigen has, and actually develop a highly active product against CD22 as well. And with that, looking to minimize the relapses that could occur due to single-antigen loss in this population. This program is currently being explored in a Phase I trial. We expect data at the end of the year. On cell itself, that I mean, obviously, we just looked at before that program is a pivotal stage, and we're expecting to have the first data cut, middle of next year and full data by the end of next year in the adult population. When we look further beyond these 2 programs, there's quite a lot of activity that we have ongoing also towards the treatment of solid tumors. And what we're using there is more complex cell programming that actually renders the cells insensitive to the environmental drivers that we see in solid tumors, whether it be checkpoint, TGF beta or also the fundamental challenges around the persistence that we do see you have to be a real challenge for solid tumor CAR-T programs. I'll stop there and I'm looking forward to the conversation with the panel.

Thanks, christian. Next, we have Teri Foy, the SVP of Research and Early Development for I-O and Cell Therapy at Bristol-Myers Squibb.

Thanks, Matt, and thanks for the invitation. I don't have any slides to share, but I'll just say a few opening remarks. At Bristol-Myers Squibb, we're really excited to be at the forefront with in the cell therapy field. It's really an exciting time to be part of it. We're excited to have the first -- to be the first company with 2 CAR T cells approved for 2 different targets: our product for CD19 and Abeta for BCMA. And behind that, we're building a deep and diversified pipeline of next-generation programs, both in heme malignancies as well as solid tumors. Of course, we have a long history of immunotherapies in solid tumors with our Bristol-Myers Squib history of checkpoint inhibitors. But we also have a strong history of expertise in he malignancies. And so we're building on that foundation and advancing our cell therapy platforms. We're excited to be part of this field. We know that it's going to be challenging. We have one of the largest data sets of CAR T cell translational, clinical and manufacturing data on which to learn from to build our next-generation pipeline. We're making significant investments, both with partners to extend our innovation platforms is -- as well as internally in building our pipeline, both to next-generation target from next-generation engineering approaches as well as off-the-shelf approaches. And so we're really excited to be part of this field and be part of the next wave of innovation here. Thanks for the opportunity to participate.

Thanks, Teri. David Gilham, the CSO at Celyad.

Yes. Thanks, Matt. Thanks for the invitation, and good morning, everybody. So no slides for me as well, the following 2 reasons, please. Maybe because I'd like to just give a bit of a background to Celyad really is a story and the fact that the company itself moved into an immuno-oncology focus and specifically CAR-T clinically around 2017. And that's based on technology that was licensed from Darker College and primarily focusing around a receptor called NKG2D, which is a natural killer cell receptor, but expressed on a CAR-T. The real attraction of NKG2D is multiple legal targeting. And I'm sure we'll get to talk about some of those benefits in due course, but it's an interesting receptor. And we see really quite some time, taking us through first-in-human studies, demonstrating safety of this approach. So the safety of using a natural receptor and also with some evidence of initial clinical activity. Now in the background, we were working on some differentiated allogeneic technologies and those that are now coming to the fore. We've combined the first allogeneic technology, which is a peptide-based approach with the NKG2D receptor. This is, I believe, my co-panelist may be able to inform me otherwise. But I think this is probably the most advanced allogeneic CAR-T in the solid tumor space. We've treated around 15 patients or so with colorectal cancer. We'll be reporting an extension to that series in the very near future. But the direction of flow for that particular trial has given us to the stage where we're very excited to -- we're looking to initiate a trial in combination with KEYTRUDA with our colleagues at Merck later this year to really ask a question of whether the immune responses that we're seeing being generated by our AlloCAR T could be boosted by the specific checkpoints. Now moving -- continue to move on the allogeneic focus. We've also developed our own proprietary interfering RNA approach to generate further allogeneic . And we've recently just shown our first clinical data with a BCMA CAR, showing engraftment a lack of GVHC, but also some evidence of clinical activity, but early stage kind of activity with that particular CAR, which we feel substantiates the platform. And we're very excited to be looking to go forward to not just use that shRNA approach that may generate further allogeneic candidates, but be able to take that further in terms of dealing with some of the issues that Rafael mentioned earlier. How we can use different shRNAs and a multiplex approach to try and generate T cells with bespoke and refined functionality. So overall, Celyad has been an interesting story. It's a company that started in the autologous space. and has really moved into the Allo. And I think we've learned an awful lot during that time, particularly around manufacturing and delivery of these cell products. Thanks, man.

Thanks, David. And lastly, Alfred Slanetz, CEO of Geneius Biotech.

Thanks, Matt, for having us at this conference. Next slide, I guess. So what we're taking is a different approach to bringing immunotherapy to all patients. And really, what we're focused on is perfecting the autologous cell manufacturing such that we can outgrow autologous cells from 90% of patients' blood draws and do it plus or minus 15% in terms of our GMP release specs. And it's taken a lot of very careful work as I know all of you appreciate to be able to accomplish that. But really, our product then doesn't need any of the chemotherapy conditioning or IL-2 or have the need for the lymphodepletion types of agents. And therefore, is really designed for a very different use, and that is outpatient first line therapy, which we hope to be able to bring this to even the community oncologists in the longer term. We have absolutely no T rate, no exhaustion, 40% central memory and up to 20 different tumor-specific targets that we can target all of the same product. all identified generally by liquid biopsy sequencing to identify the specific targets. We have a management team has been involved in the CAR T world, I was at bluebird, Wayne Froland actually was involved in KEYTRUDA manufacturing in the past. And our manufacturing, we focused on very low cost of goods. And pipeline really is in 2 areas: lymphoma to really to draw and sort of compare and contrast with the CAR T AlloCAR-T work and then CTIL's melanoma lung cancer. And we have a product with potency really great tumor specificity multiple targets, which reduces escape, but also actually allows us to address the heterogeneous tumor and solid tumor area and also opened up the microenvironment. We've engineered certain things into the cells to do that. Next. And really, these are designed, I think, as we discussed to have a very low cost of goods. Our goal is to try to push envelope and get into a situation where because of the outpatient and low cost of our manufacturing we can produce a product at $100,000 per patient for the types of therapy and no need for conditioning. As I mentioned, we have met very stringent quality attributes. And really, what we're doing is focusing in parallel the full genetic diversity of the individuals, MHC and also T cell receptor repertoire on the tumor-specific mutations and rearrangements. Next. And this just shows you that in multiple production runs, we are fairly reproducible in terms of being able to target at least 18 out of our neoantigen targets in the same product. Next. And the killing is exquisitely sensitive. We've never seen killing of wild-type cells, only normal cells -- or only cancer cells rather. As you can see, we have about 80% or 50% to 80% killing against specific mutations, which are present on the person's tumor in this case for G12D and EGFR. And this really leads to a very good, we believe, specificity will lead to a very good toxicity profile for these types of approaches. Next. Our pipeline then starts with lymphoma, as I mentioned, and in approximately 3 years to the filing. We have other viral types of targets. These are actually targeting EBV latent antigens and also viral targets. But really excitingly, as I mentioned, we're focused really primarily on tumor-specific antigens. And there, we would have data in the proximately 2 or 3 years as well. And the goal is to start in melanoma lung cancer and go into a variety of solid tenders. In terms of really where we are, we think we have a really good potent complete specificity, multi-targeted product, which opens up tumor microenvironment. We've been reproducible. We do this from the blood and we have multiple targets and single production run. We've really balanced sort of the need -- the best product being completely matched to that patient's patient, but also be cost effective and time sensitive to address it. We can see, of course, internal antigens as well as external antigens, the internal antigens being a lot of the drivers of what causes proliferation metastasis. We're using, as I said, really Nature's computer, the millions of years of evolution that have resulted in the genetic diversity of the NMHC and the TCR to focus on the actual best targets for that patient. And then really, we're focused on trying to change the standard of care with modest pricing and low toxicity. -- so that patients can be treated in a local community cancer clinic and also have a potential for first-line therapy. So thank you.

Great. Thanks so much. I hopefully we could sharing. We can get into the kind of more Q&A. Thank everybody for those introductions. So Teri, I'd like to kick off the discussion with you, mainly because of the recent announcement of the positive transform results. And I realize we don't have the data itself, and we'll say that for a medical meeting. But just from a high level, how important do you think that is for the field, the first randomized trial, really moving up the line and going up against a standard of care that's been used for a long time. So maybe we can just kick that off.

Yes, Matt, thanks for the question. We're really excited about this data. And just to reiterate what you said, we announced a month or so ago that [indiscernible] the CD19 CAR T cell showed benefit over standard of care in relapsed/refractory large B-cell lymphoma, that standard of care being a high-dose chemotherapy and stem cell transplant in a randomized trial. And as you point out, that is the first time a CAR T cell has been in a randomized study versus standard of care and showed a better outcome. And so it's a really exciting data set. It's the first time for a CAR T cell demonstrate that better activity than standard of care. It's really important for the field of cell therapy. It's also even more important that we're able to think about bringing this alternative therapy to benefit patients in earlier lines of treatment. And that's really what we're all trying to do is benefit the patients and to now be able to say head-to-head comparing to sort of the gold standard, that we can show that a CAR T cell and bring a better benefit to patients. It's a really exciting time. So I think we're proud of that data, and we'll look forward to sharing it at a metal meeting later this year.

So if we look forward to the results. Gilead did give a pretty impressive EFS hazard ratio in their press release. I guess more broadly, there's some talk in other indications such as adult leukemia, Christian or maybe AML, Dave, I know you guys are working on there a little bit. As far as do you put a patient into transplant after a CAR-T therapy? I guess is there any read-through from some of these transformed results? I know it's in lymphoma so much different indication, but maybe Teri or Christian, do you think this will read through to how patients should be treated in some of those other hematologic malignancies?

I think from my perspective, and I'll focus my answer on acute leukemia and for this part, I think it's very clear that if you -- that you would like to use actually the CAR T therapy as it relates possible. And I think the best surrogate for that, that we have been -- be able to see in the field is certainly the early use of blinatumumab in frontline consolidation, which showed pretty much a doubling of the -- both the CRA as well as a massive improvement in terms of durability of effect that was basically demonstrated. We're looking at overall survival being above 50% of 3 years out of that trial intraplant consolidation. Now what we do know is a product like [ ObsiLis ] at least twice as active. And when we look at the durability that we can see, it seems to be in a different ballpark in terms of long-term effect we can induce. It makes every sense in the world to move these products up. There's also another aspect of that is, and that has a lot to do with the fact that both the disease as well as the current therapies have a huge toll on the patients. There's a reason why patients ultimately merge them to come to some form of sepsis. And it is all down to the fact that the new system has completely shut off on 1 hand because the regeneration of the immune system is blocked by the disease itself. And then we're using a horrendous amounts of chemotherapy to sort of manage these patients through, which have detrimental effects on the immune system overall and the performance to pay these patients experience. So from where we're sitting to go early with a T cell-mediated therapy, I think, makes every sense in the world, and I think it's very nicely prototyped through the work that I've done with my old team on the early stages in consolidating with plan.

Is there anything to add on maybe a broader read-through from Transform? David, do you have anything to add on maybe...

I would just agree. And I would add, what we're looking to do is replace transplant, not bridge to transplant. I think that we've shown that these therapies can have transformative outcomes for patients and really looking to bring those therapies to earlier lines and replace these harsher and not necessarily curative therapies.

I mean if I could compare and contrast really. So looking in on the outside from the CD19 space doesn't have a CD19 asset. But we are working with the NK2GD and AML. It seemed pretty clear to into the various clinicians that there was certainly with CD19 CAR and the fact that they were tending to move away from autograft in particular and using CD19, which sort of demonstrates really the fact that the clinicians have confidence. And so the pharma colleagues here, this is really important information that BMS is generating to show, without a doubt, that there's a benefit of this on the therapy. In the AML space, the comparators and other acute leukemia. The CAR Ts are basically not able at the moment to generate the sufficient level of clinical response. And so by default, the clinicians, if there is a patient who gets to CR or approaching a CR, will tend to transplant those patients because it is very clear at the transportation is going to give the first clinical result. But I think this is only a very early stage, just understanding how different leukemia works and how to generate clinical responses that can challenge transplantation is going to be very important. But it's an important point that perhaps outside world doesn't quite grasp is the fact that there's not 1 leukemia. There's very many different types of leukemia and we're trying to challenge some of those in a very different way. And the CD19 is a real success story that backs Blueprint for other leukemias and implements in the [ hemo social ] space.

Those [indiscernible] David that you were making with regards to the confidence of the physicians to not resort to the standard of care and consolidate with the transplant. And I think the only thing that actually keeps a position from doing that is you have can actually provide continued measures of the molecular disease status of the patients, you want to continue to show that the patient is in molecular CR. And the second thing you need to be able to show is that indeed, the CAR-T actually persists over time. There's 2 pieces of information, you can have confidence of not actually resorting to the transplant. Let's remember the transplant in -- on the acute leukemia side actually comes with a very significant level of mortality associated with it as well. So it's not a freebie, it's actually a key approach and particularly if you're looking at a second transplant, which is certainly a major challenge that we haven't been ale. But I think having information, very deep information with regards molecular status of the response and persistence, I think it's the 2 pieces that you need to have in place for the physicians to actually have confidence and actually just really give it a go.

I think -- so just to add to that, Christian, I couldn't agree more. And just really to highlight the difference in the AML space, where there isn't a molecular signature because the number of fusions are so vast, there is an obvious report. So I think there has to be developments in many ways to really support the whole therapy.

That's a good transition to the next kind of topic. Cell kinetics is something that has talked about a lot, both comparing, contrasting auto and allo and different indications. And firstly, I think we're starting to see that in something like leukemia's cell persistence matters, in lymphocytic leukemias, whereas in lymphomas, maybe it is more of an initially you see a kind of Cmax-driven long-term benefits. So Rafael, given kind of the allo approach you guys are doing, maybe you can start off there with how you're seeing cell kinetics matter in different indications. Maybe how that's influencing how you want to design constructs maybe for something like solid tumors?

Yes. Matt, that's a really good question. I just wanted to say that pertaining to the previous question that this was the choice of endpoint of event-free survival, I think it was brilliant on a great partnership with the agency, and we demonstrate the poor prognosis of these patients will can tail either at the time of salvage chemotherapy or at the time of transplant. And so how CAR really proved to be superior when one uses this endpoint. So with regards to persistence, I think in clinical trials, second-generation CARs incorporating for appear to favor delayed a longer CAR-T-cell persistence. There are other factors that really go with regards to whether persistence and expansion are long-lived and what their impact is in long-term clinical outcomes and that has led people to speculate how -- what the importance of it is. Factors like what is the makeup of the [ infusate ] of the cells are given the sense phenotype, the central memory composition, the so potency and this density and presence of the target. I think there are other factors that are completed as well and are difficult to sort of piece apart from persistence itself. But it is true in leukemia that in vivo persistence and barenalotemial activity are associated with increased risk of relapse. For us, in the allogeneic setting, we have to deal with rejection. And of course, persistence is limited by the ability of the body to actually tolerate these cells. So we maximize persistence by suppressing the immune system and by trying to evade immune rejection of the graph by eliminating endogenous T cells. And we've been able to have persistence of -- with chimerism of about 6 months and counting. And we do this by suppressing reactive T cells with anti-CD52 antibody and not been out the CD52 receptor during manufacturing in the graph. And so this allows us to redose when the T cells start to research. And we have a pretty good idea of when that happens and what the dose of anti-CD52 antibody is required for the T cells to cross the threshold that we believe is important with regard to innovation. But I'd just like to finish these comments by saying that there are -- there's a lot of work, particularly in the allogeneic setting where gene editing can be more prolific to increase persistence such as all immune defense receptors where the cells actually defend themselves against rejection or manipulating CD70, CD27 interaction or B2M, which we see other companies doing or other factors in the MHC complex. So this is something that is important to watch, and it's going to be really important, I think, in the setting of solid tumors. But -- and I'll just finish by saying that we've seen tremendous responses that are incredibly long-lived in patients are having had the best expansion and the best persistence. So I think this is still an unanswered question, and it's even more so in the solid tumor where the first inroads, if you will, into this topic, are being made now. So it will be very important to see how persistence and expansion plays with the panoply of other factors that are important for T cells to actually exert an antitumor response in solid tumors.

Teri, I guess, on that similar topic. [ Beyond what's ] unique in the defined kind of ratio of CD4 and CD8, do you think that -- at this point, how do you think that really affects the cells? And is there something you're going to look to take forward into maybe other programs?

Well, we certainly know from our data sets that our safety profile is good compared to the other CD19 CAR T cell products. And we do think that is a result of the defined composition. In terms of persistence, our persistence of the CAR T cells are, I would say, on par with the other products. So I'm not sure that it impacts per system per se, but it certainly does appear to benefit from a safety perspective. I would say, to address the question in general though, I would agree with what Rafael said that there's many factors that influence persistence and function of the T cells. It's not just whether they're there, it's whether they're functional as well and that's going to differ between diseases. And even within the disease, we see patients that have long-term responses. We see patients that relapse early or relapse late depending on either lymphoma or myeloma. And certainly, as we get into solid tumors, I think we'll see the same thing. We'll see that sort of differentiation even within a disease. And the many factors of tumor intrinsic factors as well as the tumor microenvironment influence on the CAR T cell may influence persistent and functionality of those cells. I think it's -- it goes beyond just the kinetics of the cells that you put in. It's what happens to them kind of once they're there. So I think there's still a lot to learn, but I think we can build off of what we're learning in the hemalignancies, for sure.

Sure. And Alfred, as a nongene modified approach, how do you think about persistence of those cells?

Well, we start with a very defined product of roughly 60% CD8s and 40% CD4s. And then within that, 40% of the cells are central memory. So what we have is, of course, an autologous product, right? So we don't have to deal with any of the risk of rejection, and the cells' proliferation is driven by antigen, which is there, and we're targeting up to 20 different antigens, which are present in the patient's tumor. So that's driving the near-term expansion. But then we have the ultimate persistence, which is memory, right, that goes on and even continues the immune response while the tumor is there but also kind of has that long-term persistence that should allow for better immunosurveillance if we're successful in a long-term response. So we're very much believers that persistence is very, very valuable in treating cancer and targeting multiple antigens with that persistence is really valuable because of tumor heterogeneity in that.

David, especially in the solid tumor setting, how are you guys thinking of it as far as giving multiple doses and everything like that?

Yes. I mean I guess there's a general first point, we'd be debating the relevance of systemic persistence for about 30 years or so. And to be honest, we still don't have the answer. It does feel a little bit like drop in your car keys at night on the street, they're looking under a street light just trying to find them because we access the peripheral blood and assume that what's happening in the price values occurring elsewhere. So it's quite hard, Matt. The issue for solid cement is that whatever we measure in the peripheral blood, are we simply measuring the cells that have not got to the right place? Are we measuring the right things are the wrong things? And we just don't really understand that. Moreover, as Rafael mentioned, looking to generate cells that have an earlier phenotype, and Alfred so discussing about earlier memory cells, cells that have a more naive phenotype. By default, these cells go to the lymph nodes. They're not going to sit in the peripheral blood. So is the reduction of persistence and seen as a failure because we're using a more naive cell. I think it comes down to really understanding the biology. And the specific issue around solid tumors, in particular is that, of course, we're limited on how much information we can get from patients. We can take biopsies, but it's really a shot in the dark to try and get some samples and they try to tease out some information. But we don't really have imaging technology, non invasive imaging technologies that can really help and have the sensitivity to detect the number of cells you might expect in that particular place. So that's a long answer to say I suspect the systems on metrical blood hopefully, is not so important when it comes to solid tumors. But the question is if whether it's a correlative readout, it's just quite hard to understand at the moment. I think until we really understand more around the biology of what's going on in terms of the clinical response.

Christian, I know you guys, you had a slide on persistence. You talked about it a lot. Maybe give your perspective.

Yes. I think it is, as I think was mentioned, the disease setting matters a lot in terms of how we think about persistence and the importance of persistence and that we need to under what's ultimately driving the particular disease. As I mentioned, in leukemia, acute leukemia, your driver cell is very, very close to stem cell. It's hardly differentiate it. And so at its enormous proliferate potential. And the only way you can do that is making sure you get rid of all of those cells, otherwise, disease will actually be able to come back and repopulate. So that's on one end of the spectrum. But then we obviously have different situations in, obviously as discussed before in lymphoma as, where the disease is actually not in ], it's localized. And in fact, if you sort of can actually hit it hard enough, you seem to be getting a long-term outcome. And that's if you look at the start of data that you look at some of the other data will clearly indicate that, that is what we can get to. I think as soon as we move into the solid tumor setting, it actually gets more complicated. I agree with David's comment that what we're looking at in the bloodstream is basically -- actually, it's the T cells that are idle. It's the guys to basically go out of the work, kind of cruise around in your bloodstream, and eventually are looking for a new job to do. But the actual expansion of the cells actually happens at the side of the tumor. That's where the action is. And to David's point is we have no way of actually seeing that at this point. We lack the techniques to interrogate it. But fundamentally, I think it is quite intuitive to understand that when we're actually infusing our T cells into the patient, whichever way we generated them, those cells need to find that particular location where the target is, where the tumor is. And they will do that with very T cell does still. They will go into the marrow, into the lung eventually nodes, and then they need to rates and actually score and screen the tissue to find those lesions. Just from a probability perspective, you have in that setting, just very, very few cells that will actually arrive at a given lesion. And then we actually demanded those cells that they do a lot of heavy lifting. So what that means is that actually those cells have to actually have a very significant amount of resilience. Persistence is one of them. But the other aspect's actually has to do have to do the fact they have to actually counter the negative influences that they're going to encounter in those lesions. And I think one of the other aspects, I think, it's important to keep in mind is that solid tumors tend to grow relatively slowly. It's a long dance between the tumor and the immune system, but eventually the tumor wins by tipping the balance and employing enough defense mechanisms to overcome the of the immune system. So in other words, we're going after a set of structures that are a lot more sophisticated in that regard than in acute leukemia is or deals actually looked like. Because they cannot just win by the speed of proliferation, but they actually need to win over time. And I think that's the -- I think the aspects when we think about the problem is that we need to actually be able to provide cells that we're driving towards solid tumors, a true resilience, which is a very significant level of persistence, but also an ability to basically not being susceptible to the negative signals in the environment. That's pretty much what we do actually with our most advanced program, where we're kind of running through are running exactly through those steps. But it's -- indeed, it's a much more -- it's a much higher level of complexity to be able.

So a follow on to -- since this talked a lot about solid tumors here. I guess almost a multiple choice answer and don't say all of the above, even though it is probably all of the above. But if you had to design 1 element that you think would help the most with solid tumor settings, is it trying to increase homing to the tumor itself somehow? Is it that multiple antigen targeting to overcome heterogeneity? Is it the tumor microenvironment, all the suppressive factors or maybe one other? But again, I realize the answer is probably all of it or it's different by tumor type. But if you could try to give what you think that is the most important. Teri will start because you're top left for me. So you get to go first.

I want to say it's all of it. I mean I know you don't want that answer. But I mean, I really do think for solid tumors, there's a lot of hurdles to overcome and you touched on many of them. And we could talk about how to overcome those, whether it's both combinations or multiple engineering. But as you mentioned, there's antigen selection on multiantigen targeting, there's tumor microenvironment influences, there's antigen heterogeneity. There's many things that need to be addressed. And I do think for solid tumors, that we will need to do a number of those engineering strategies and combination approaches to be able to overcome the challenges there. Trafficking, certainly, it is a challenge, too. And certainly, some tumors may be more amenable than others to allowing the T cells to penetrate. But I would say, antigen specificity and selectivity, I think we can overcome, but I think it will probably take multiple antigens. I do think the tumor microenvironment is probably one of the biggest challenges in that there are multiple factors influencing both the function of T cells and tumor intrinsic variables that allow the tumor to kind of take over and outgrow the immune system. So I think being able to attack it from multiple angles is going to be the key.

Okay. I want to go out on a little bit with a single most important thing.

Well, I don't know. I think they're all important if we focus on all of them. But I would say the 2 most important things would you be able to try definitely multiple antigens. Targeting antigens across the tumor and dealing with the heterogeneity is really important. And then having a cell phenotype, actually that can transform the microenvironment based upon that to a pro-inflammatory microenvironment. And that, at least in studies we've seen seem to be effective within a week or 2 even, if you get enough of those T cells expressing across the tumor to do that. And that's where the multi-antigen approach also is very, very helpful because it starts to transform that microenvironment. By delivering these cells that can create that inflammation, maybe get rid of the college and build it more infiltrate from other cells that just are recruited in as part of inflammation and then you change you -- turn the tide in that tumor.

Yes. To build on Alfred's point, and Christian's logic which I entirely agree with because the -- I think it's certainly a solid achievement, we're not going to -- the access is going to be an issue. So if we assume that there's a limiting number of T cells, they need to co-op whatever is there to try and help them. So this means beyond the engineering of the T-cell, it means how can we impact the local area. Producing cytokines as such are armored CARs, as called in the states, trucks in Europe, choose which one you prefer. But generating cytokines or such of other elements that can transform the local immune system, I think it was very important. And I would say just a selling point of NKG2D. And actually, one of the reasons why I joined Celyad is the fact that the ligands for NKG2D are also expressed on Tregs and myeloid-derived pressor cells. And certainly, the preclinical model suggested that the responses that were being driven in preclinical models were partly of the release in new system likely due to some deletion of Tregs. So the ideal scenario would be multiple targeting where we can also target the elements of the state, which are really immunosuppressive. But I guess, number one on my list would be to produce factors that can actually strongly co-opt that immune system in a local and more area to try and drive a longer response would be key.

Christian, anything to add?

Yes. I mean the way we approach the question, which obviously has a lot of complexity built in, is to actually take it in steps and ask first a question, "Can we actually have an appropriate level of therapeutic window", which for many of the antigens that we're choosing in solid tumors, particularly if it's classical cell surface proteins, often is actually more complex than we would like to talk about. So we started at that level, and we did actually a program in neuroblastoma using GG2 and quite a bit of engineering are receptor to be able to show that. The next stage, we felt that we needed to tackle once we've established that we have the therapeutic [ green ] go. And indeed, in that case, insufficient persistence to make us get actually to sustained responses or got the responses, but we couldn't sustain them, was to actually really work on the resilient sulfur cells. And we've decided that given what we're up against going into one by one step, we're probably not going to be young enough to see the end of it if we're doing that. So we decided we're actually going to build an approach where we take 3 additional steps onto that already snapped therapeutic window. One was basically making sure the cells are insensitive to the whole family of checkpoints, not just second, to make sure the cell is insensitive to TGF beta, which is sort of 1 of the decoy signal stat tumors like to use. And the third one, actually to give a low-level IL-7 signal into the cell to actually just keep the cell going. We call it kind of a happy gene. And when we do that, that's -- we believe we should actually create a product that has actually quite a significant advantage over and above a standard CAR T or a normal T cell in its ability to withstand a very difficult environment and that we're in the process of getting that level of product with that level of engineering back into the clinic. But that's the approach we took. So there's some level of simplification we have to make at some point. And then actually, we need to be bold, I think, on some of the next steps to sort of be able to crack the problem, frankly.

Yes. I would just mention that, I mean, this field is still in its infancy. I was impressed by statements like from [ Steve Rosenberg ] in the till field using ctDNA for BRAF, how he believes that most of the action happens really early on in the first 2 to 3 weeks in terms of eradication of the tumor, and people with CARs have shown that doing lymph node biopsies 2 weeks later, it's too late that you have to do it very early to actually be able to detect the CARs because there is tremendous tumor necrosis. When I was in the field of TCRs, I was really impressed to see the patients with tiny metastases in the lungs would have responses and how these T cells were able to circulate traffic, penetrate the tumor and get rid of metastasis that were 1 cm or less. I mean, all the factors that you mentioned are important, but I would like to emphasize the importance of the target, the specificity of the target so that one can give enough cells of the right phenotype to be able to exert an antitumor effect that has the right therapeutic window and heretofore that, that's been a little bit sketchy. And the other thing is about all these modifications that can be done to T cells to try to enhance their antitumor microenvironment effect, doing enhanced trafficking, et cetera. I think all of that is important, but I think we need a lot of translational data to really try to understand what is the deficit that's preventing these T cells from actually causing a really radical antitumor effect, and then engineer them around them. I think blind engineering may surprise us and may actually give us the wrong answer or side effects.

Yes. Okay. Some things I get touched on a little bit here in this last question were combinations. And there is some experience with CAR-T plus another drug given systemically. But the question was, "Okay, is it worth giving and do systemically or can you just engineer something else into the CAR T cell down the line to kind of overcome that?" So we've seen BTK inhibitors, we've seen chemotherapy, which is probably different than any engineering you can do, also PD-1 inhibitors. I guess, David, maybe you can start off with how much you think you need to have that chemo. A CRC is obviously a very tough disease. So I think it does make sense. But just other combinations when you're starting to think about moving earlier lines of therapy and into more indications.

So I mean, so our example, I think, in colorectal is perhaps different to most. Obviously, the standard route, usually for [ tail ] CAR T is to use a silo preconditioning backbone for all the reasons we know about in [indiscernible] in ground. So we started in CRC, not thinking size flu first and foremost, but actually thinking can we build a CAR T on the back of standard of care. So the first trial that we've carried out with allo NKG2D is combined with FOLFOX. So these are patients who have seen FOLFOX multiple times, standard of care in metastatic colorectal cancer. They're relapsed or refractory to that condition. And we built our 101, the NKG2D allo to basically fit in with the cycles of FOLFOX. So there's generally 6 cycles. The way we give out 1 or 1 if we give 2 days after cycle 1, 2 and 3, the other generic CAR Ts, there's a transient period of liquor depletion is not to the same extent as any means as cy/flu. But some of the key benefits are that 5 FU induces the expression of the legs on COVID tumor cells actually known for a very long period of time. It also is likely to have some more direct effect than we would expect with cy-flu, which has basically 0 effect as far as we know on colorectal cancer. So this is really thinking about how can we use therapy that's bespoke perhaps for an individual indication and try to build the CAT on top. We see some partial responses. We're seeing some stable disease. Things are going in the right direction. It's early days at the moment. We're currently testing with FOLFIRI because we want to understand, can we think about using any standard of care that is used in late-line metastatic colorectal cancer. But in the next stage, because the responses we're seeing are encouraging, but they're not durable. And of course, in MSI-H patients, so microsatellite instable patients, checkpoints have a very major effect. We're not seeing those patients because they get checkpoints. We -- all of our patients are MSS, microsatellite stable. These patients will typically not respond to the checkpoint. But what we've seen is some of our transitional work, and this tips roughly out our work, I think translational work is very important. We have seen some increase in the patient's own T cell repertoire 4 months after the combination of therapy. So this is much later than the CAR-T and FOLFOX. In patients who responded, patients who failed to respond actually, there was no evidence of increase of the T cell rapporteur. So based on that, we're looking to work with Merck to give a PD-1 to see whether we can build the CAR-T and then bring in this combination at a later stage and see whether that can enhance the activity of the endogenous immune response. So this comes back to your original question, Matt, where we could, of course, engineer a PD-1 knockout in our CAR-T. But for this particular trial, we don't see the PD-1 checkpoint is working directly on our CAR-T. Our CAR Ts express PD-1 anyway, they're not exhausted -- And I doubt anybody around this table, had it been their cells will never be exhausted enough to be expressed PD-1 with they're manufactured. It's got to be a much longer time for the PD-1 to come on. So we're really asking the question about the patients on immune response. And can the checkpoint actually impact on that. So see the CAR T plus FOLFOX is really trying to drive an immune response that we're hoping that the checkpoint can then help to deliver for a further period. So in that situation, I think engineering the CAR-T wouldn't bring an advantage to that particular situation. And I'm just really using that as an example.

Teri, you probably have the biggest armor materium of things that you can pull from the Bristol-Myers shelves to do combination studies. I know there's some ongoing in multiple myeloma, but what else are you may be excited about or think would make rational sense?

Yes. So similar to David, we think about combinations kind of in 2 ways. One is combinations with existing standard of care therapies, where if we're looking to sort of complement direct activity on tumors with our CAR T cell activity. So a BTK inhibitor in CLL, for example, or an IMiD or a CELMoD in lymphoma or myeloma, a combination where the standard of care drug acts on the tumor directly and the CAR T cell acts to its immune-mediated T cell effects. But in addition, as you point out, we can combine with immune-enhancing agents, checkpoint inhibitors, cytokines, other immune pathway agents. And we found, for example, some agents like our CELMoDs actually have immune-enhancing activity. And so we're doing a number of combinations right now in both our studies with checkpoint inhibitors, but also in CELMoDs to look at the immune-enhancing effects both on the CAR T cells but also on the endogenous immune system. Cytokines as well, as we talked about earlier, one of the things that we believe is important is engaging the endogenous immune system in addition to the CAR T cell activity is an important feature in maintaining a durable response, we believe. So having these combinations that can potentially add to the overall immune response we think is really important. And to your point about whether combinations or engineering or which one is better, I think they both have advantages and perhaps disadvantages. Combinations, obviously, give you some flexibility in dosing and timing, which can be good. But engineering, obviously, provides opportunities to build in features into a single drug. And you may be able to build in multiple features into a single drug. But it does also potentially present some challenges if you need to think about the timing and toxicity of modulating those multiple pathways at the same time with your CAR T cell. So depending on where you're trying to target, you may want to choose one over the other. I think there's multiple approaches that can be taken.

Sure. Rafael, in RCC, it's obviously a tumor type that now has become dominated by PD-1 combination. So as you guys are thinking about your CD70 program, is it logical to -- you know PD-1 is going to not only maybe health CAR T cells, but help the kind of maybe an endogenous response as well. So how are you thinking about combinations as you move in solid tumors?

Well, this has been an obvious thing for us is to move it in combination with PD-1. And as I said before, we wanted to see whether there was evidence there to actually do such a thing because all these patients have failed checkpoint inhibitors as well as angiogenesis inhibitors. However, they still have a very T cell-rich tumor. And there is the possibility that one could engender antigenic spread, which would result in NTM presentation and potentially, PD-1 may have an effect as well. And in fact, we are obtaining still very preliminary, but we are obtaining biopsies from these patients, and PD-1 tends to increase as a result of interferon release. So it is a very kind of obvious thing to try. When one starts with a new molecular entity, generally, regulators wanted to do the classical 3 plus 3 study but we have plans to actually do extension studies, both with monotherapy as well as combination. And since these patients have already failed, which is one of the problems of drug track combinations is to now what is the agent that is actually costing activity. But since these patients have already failed checkpoint inhibitors, we should be able to see whether or not there is selectivity or synergy in the combination with respect to the CAR loan. So yes, very much in our mind, but still early days.

Christian, you probably have the most experience on the panel with PD-1 combos and with DLBCL. I think still a little mixed on how much benefit that provided in that setting. But curious to get your thoughts on combo and then also moving forward, you are looking at some dominant negative shift to as a kind of engineered way to go about checkpoints.

Yes. I think for the first, I think question you have to ask is what is it really you try to influence. Do you try to influence the tumor? Do you try to influence the behavior of the T-cell? And I think that is important to understand. If what we're working with initially in the DLBCL side, is a program that were designed to actually have a slower buildup of activities. So we basically had a different kinetic in terms of the buildup using different [ caution with your ] domains. And what that does is that obviously, as the sales actually are engaged over a longer period of time, become more prone to actually upregulate PD-1, which happens typically after 2 weeks of action. And as such obviously, it could become susceptible to a PD-L1 [ triple ] negative impact. But to Amado's point or but to Rafael's point, clearly, the -- one of the things you do with a T-cell media therapy is you actually secrete the interferon gamma gamma and interferon gamma drive at the very side the upregulation of PD-1 looking. I think the data we had, if anything, would point to actually utility of PD-1 in that setting. But that doesn't mean that PD-1 actually or anti-PD-1 has a role in of itself in the DLBCL. I think that's been very nicely answered, the vivo data and the [ K2 ] data would suggest that the agent on its own doesn't have a significant impact of the tumor. So what we're looking to do is actually impacting the T cell not the tumor. And I think that's important to actually understand. If you then only think that what you do is primarily impacting the T cell you're bringing into the system, and that is where you want to have the action. That's actually where it makes sense to engineer properties into the T cell. However, if you -- what you try to do is actually have a broader impact on the disease setting or other immune cells, et cetera, that's where you have to be systemically present. And that's obviously where I have to -- if you use engineered and engineering approach, you have to secrete an agent, which is something we've been working on. Or alternatively, you use, obviously, a standard therapeutic that's systemically available. So it's really the question is what is the problem we're looking to solve? And what is the mechanism you're trying to combine or you're trying to solve for? And that is, frankly, the way we're approaching it. So if I want to actually have a sale that is a CAR T cell that is insensitive to checkpoints and better off actually engineering that into the cell because I minimize actually any form of toxicity on to other cells that it could have. If it is actually a situation where you try to actually get out of inhibitory effects of more than 1 particular agent. If you do that systemically, you start adding 2, 3, 4 therapeutics systemically available in that patient, which obviously raises a lot of toxicity risks, you're actually better off to actually engineer that into the cell and we end the call no longer susceptible to those signals, which actually it doesn't add toxicity overall to the system, but it addresses the particular problem you're looking to solve. So it really depends on kind of the balance and where -- what the point of intervention you really going to go after.

An interesting point, I'm talking about all this is I think one of the more surprising PD-1 combinations that have come out recently is kind of levatinib, the multi-gig, maybe VEGF focused TKI, but multi-TKI and not just in RCC or indications where maybe levatinib is used. So should we be thinking a little bit outside the box? Or is that a good way to go about kind of changing up the tumor microenvironment in addition to CAR-T? I don't know if everyone needs to comment on this, but if anyone has any thoughts on just these other things that aren't maybe immediately thought of as being a -- helping the CAR T cell. Anyone want to have a comment on that?

Well, certainly. I mean, other items that open up the microenvironment like an anti-VEGF could be quite helpful in certain tumors. And -- so I think that other combinations. So one thing on the PD-1, I actually think it's the ideal long-term combination for at least in our thinking, in terms of T-cells very busy to point out after a couple of weeks to keep the activity rolling, right, in the patient. And PD-1 really teaches us a lot about the antitumor response because it tends to work better when you have high mutational burdens, which means you have a higher precursor dose, if you will, of T cells to unleash at the tumor. And it doesn't work as well in the so-called lower mutational burden generally. And obviously, there's a debate about what is high and what is low, but sort of conceptually, one of the things that we feel for ourselves, given that we're targeting a lot of antigens that are present even in low mutational burden is by increasing the dose and combining with PD-1. You actually could maybe make PD-1 type of combination useful for virtually every patient, which could be amazing to bring the world of immunotherapy to people instead of today's chemotherapy, radiation types of early therapies. So that would really bring T cells and immunotherapy to really change the paradigm for these patients early on.

Right. So we have about 10 minutes left, and I think maybe we'll try to get to the last question, which we could talk about for a long time, is one of the biggest debates. But autologous versus allogeneic. I know we've all had this come up before. But my question is, should we really be thinking about how these 2 approaches can coexist? Are there going to be indications where one is better than the other, like we see with transplants? And I realize there's a lot of other factors that go when you're doing a total transplant versus a CAR T therapy. But curious, just everyone thinks it's one or the other, but maybe there is kind of an approach for both, at least in the next decade, let's say. So Rafael, I'll start with you as the key allogeneic player.

Yes. It's a great question. I think autologous therapies are the current standard and they're getting better and there are a lot of studies to read out and we're going to see more and more advances. I think this in different sort of buckets of progressive evolution, autologous therapies and allogeneic T-cell therapies and potentially iPSC therapies. And then down in the future in vivo reprogramming, which there are a lot of companies that are using either viruses or LNPs to try to reprogram T cells. And so this is going to be a continuum. Certainly, there are pros and cons to each one of these technologies. And I think eventually, allogeneic therapies, because of issues of cost, issues of access, the potential of doing it as an outpatient, if we can reduce the toxicity to that level. And more importantly to me, the potential to engineer around problems so that still in lymphoma, for instance, 60% of patients fall out of response. So engineering around these mechanisms of resistance can increase the benefit risk. I think all these therapies will have an opportunity to do that, but the allogeneic ones being off the shelf where the manufacturing happens much earlier have perhaps an advantage. But they still need to establish themselves. And obviously, we're working hard. I think we've created the most allogeneic patients so far in hemologic malignancies and we would like to be the first to actually have an allogenic product in the market. But initially, they will coexist and then I think the evolution will be dictated by the biology.

Teri, you're at a point where [indiscernible] will be second line transplant-eligible, we hope, obviously, putting approvals and everything. But where do you think you guys would explore an allo? And can you tell us anything about kind of how you're thinking about the 2 different approaches?

Yes. I mean I think, Rafael stated that well. I agree, I think that the whole field of cell therapies is there's an innovation cycle and it's a continuum. We're at the beginning of that with the first generation of autologous cell therapies. And I certainly think that today's approved cell therapies are the gold standards, but they will be superseded eventually by the next wave of innovation. And I do see that, that will take some time. It took some time for the first, autologous therapies to be approved. The regulatory hurdles are not trivial. There's a learning curve for scale-up and commercialization and all of those things. And the autologous therapies will continue to improve. As we've just discussed, there's lots of engineering features, manufacturing improvements, cost of goods and automation improvements that are ongoing right now both at BMS and as well as at other places. So I think that they will move in parallel. Innovation will move in both autologous and off-the-shelf settings. But I think the cycle will be rapid. And I can imagine as we move forward in the next decade, we'll see great progress, both with off-the-shelf approaches, whether it be allogeneic or PSC. I agree with Rafael that the advantage of off-the-shelf approaches, particularly an iPSC-derived approach in the ability to engineer multiple features into a cell that can be your kind of master cell bank, that you can continue to derive additional products from is a huge advantage of an iPSC-derived-based platform. And certainly, in vivo, if we ever get there, would be an ideal set of therapeutic. We're pursuing all of those different approaches at BMS. But it's all going to take time to prove out the biology. And I do think there's still a lot to learn on the biology of the allogeneic cell therapies.

Yes. David, you guys definitely seem to be pushing ahead with, I think, a focus on allo going forward, but maybe any other thoughts on kind of -- you have done both, so you have looked at both.

Yes. I guess from an internal perspective and certainly from the manufacturing perspective, allo has a lot less issues in terms of speed. It has a lot more issues in terms of batch-to-batch variability, [ QC ] release, et cetera. So it doesn't come without its problems. But in terms of stress on a just-in-time manufacturing, obviously, that's very different. I mean in terms of the original question, Matt, yes, so if we use -- a CAR T is basically an antibody stuck on a T cell. But after that, I think autologous and the similarity starts to fall apart. The source of the T cell is very different, the way that an auto CAR-T is used and can be considered to use is different to how we would think about allo, we're trying to engineer different elements into an allo as Rafael says. But I really think about them as a different entity. So I don't actually really see them as competition for each other because I think they're going to be used in a very different settings, which to my mind means that they will coexist, and I'm sure that will persist through a period of time. And probably the section will be, as Teri was saying, yes, it's going to be dependent on the patient need and how this fits in but also numbers. Yes. And so the reason why we were thinking about standard of care in colorectal cancer and thinking about an allo approach on the north of this, how do you deal with 100,000 patients a year to try and have enough product that could be moved backwards forward? So I think there's a various macro issues. And I would say that actually allo CAR-T probably the better comparative would be biologics, really, rather than autologous cars because I suspect that they're really be working in a different place. So I think there are different entities overall. They're having lived through the auto to the allo certainly from an internal stress perspective, the stress is certainly much lower from a manufacturing perspective in terms of generating the other, but it's not stress-free.

Christian, maybe any -- we're getting low on time, so not to cut you short, but a few short comments so Alfred can also get a chance to comment.

Sure. Well, I think the way I think about these questions is, first of all, I think we're remarkably about of predicting the future, I think collective. We don't like to admit it, but I think it's true. But I think what's always helpful to remember is put yourself in the shoes of the patient. The thing that matters in oncology is you've got to leave, and it's tagged to the outcome and it's tagged to the performance of the therapy. And frankly, as a patient, you don't care what the technology is. we can be enamored as much as we want with the kind of stuff we do for as long as it doesn't actually do the trick, it doesn't matter. And I think that's really what we need to sort of focus on, is really drive on really high transformative outcomes. And those will have a place and we'll have a need and meet the need. And frankly, probably be a bit less hung up on the technology. There's a lot of different ways to skin this cat and I think all of us are working on lots of different approaches here. But fundamentally, what we need to focus on is what really makes a difference. And those products will actually have a place and will be important.

Next question, Alfred.

Yes. I think that really, it depends upon the application. If you need something immediately, you may use the allogeneic approach. But the downside is, is that if you need a durable response, at least until CRISPR or other techniques come into play to reduce the immunogenicity, the duration of response may be an issue. So you may follow it up, for example with an autologous product in a second dose. Or I think when it comes to leukemias and lymphomas, that's more likely to be the sweet spot, particularly in more advanced patients who need something rapid. But when you get into the solid tumors or earlier lines of therapy, where you can have the autologous perfect match. I mean, that's really the ideal type of therapy for those patients. So I think that's where autologous will come into play to a far larger extent, particularly earlier in the treatment and the places for solid tumors and particularly.

All right. Well, I think that's all the time we have. Christian, don't give away the secret that we can't predict the future because then we might not -- we might be driven [indiscernible].

That's your department. That's not us.

No, I really appreciate everyone joining us today. I think we tried to touch on all the major topics. It's going to take data to answer these questions here over the next couple of years. But I think regardless of each individual project, the whole field is definitely moving in the right direction. So I wish you all luck in the next data update and look forward to it.

Thank you, Matt.

Thanks, Matt.

Thanks, Matt.