Hello everyone, thank you very much for being here, either in person, via webcast or in the replay. It's a great pleasure to have you here with you today, with our management team and our leading clinicians from around the world who share their enthusiasm on our development. Today, we have the honor to be hosting Jean-Pierre Delord, M.D. Ph.D., General Manager of IUCT Oncopole of Toulouse. Adel Samson, M.D. Ph.D., Clinical Associate Professor, CRUK Clinician Scientist, and Honorary Medical Oncologist for the University of Leeds. Pedro Romero, M.D., Editor-in-Chief of the JITC and Deputy Scientific Managing Director of the Lausanne branch of the Ludwig Institute for Cancer Research, who are here with us in person. Also, Christian Ottensmeier, M.D. Ph.D. F.R.C.P., Professor of Immuno-Oncology at the University of Liverpool, The Clatterbridge Cancer Center, NHS Foundation Trust. Matthew S. Block, MD PhD, Medical Oncologist at Mayo Clinic, who will participate remotely. Having their support today gives us a lot of confidence on the products that we are developing. I would like also to thank a lot of the organization team and our support companies who helped us make this a reality, and special thanks to Lucy. Great job like always. Today, we are going to discuss the latest progress that we have made on our clinical trials. You will see that the data that we are generating in patients can really be game-changing in oncology. Maud and our clinician colleagues will walk you through the very promising data that we have generated. Also, science is at the core of who we are and what we do at Transgene. We are 160 employees, all committed to pushing the boundaries of innovation to better treat cancer patients. What you will see today is also the result of the hard work and daily commitment. Thanks to all of them. Today, we will take the time to deep dive into the way we design our candidates and showcase their highly innovative content. Éric will discuss our highly innovative platforms, myvac® and Invir.IO®, that could be created thanks to our years of expertise in viral vectors. Our uniqueness also comes from our capability to partner with best-in-class innovators. Our collaboration on artificial intelligence with NEC precisely falls into this category. When it comes to developing a product such as individualized vaccines, we biologists have ventured into the wild world of artificial intelligence. I believe it takes humility to go and partner with specialists, but by doing this, I'm certain that we have integrated one of the best, if not the best, artificial intelligence tool for antigen selection on Earth. Combined with our vision, with a powerful vector and with an adequate clinical positioning, and with the commitment of our employees and clinicians present today, this could be very well translate a landscape changing opportunity for patients. Our uniqueness also relies on our conviction that the intravenous route could be the next big things for oncolytic viruses. With TG 6002, we have demonstrated in the clinic that this route of administration is feasible. This is important for TG 6002, this is important for the Invir.IO platform, and this is important for the partners such as AstraZeneca. That will help us go beyond what we thought could be the limitations of our technologies. There are also a few topics that are of great importance that we will not have time to cover today. For instance, manufacturing. This is key in our industry for both our platform, oncolytic viruses and therapeutic vaccine, and even more for individualized treatment. As you know, we have two technology platforms, therapeutic vaccines and oncolytic viruses, four products in the clinics and lots of ideas and new ideas that will be communicated today. I now would like to welcome Maud, who will be hosting the session on therapeutic vaccines. Thank you, Maud. Thank you, Hedi. I am very glad to introduce and to share this session on cancer vaccine in the treatment of patients with a solid tumor. As you all know, immunotherapy has become the fourth pillar of cancer treatment besides surgery, radiotherapy and chemotherapy. This has been possible thanks to years of intensive work and strong confidence in the potency of immune system to destroy cancer cell. This has led to the advent of checkpoint blockers, and we know that checkpoint blocker have demonstrated their efficacy in many solid tumor types. Having said that, the response rate to checkpoint blocker remain in a range of 20%-30%. It means that for a significant proportion of patients, they are not responding, they have a primary resistance to immunotherapy. In addition, among patients who respond, unfortunately responders relapse and develop secondary resistance. Obviously there is a need for a new therapeutic option to overcome primary and secondary resistance. There is many developments in that field, but we believe that vaccine, virus-based cancer vaccine can be a potential game changer. The concept of cancer vaccine is very attractive. You inject vaccine at a distant site of the tumor where the immune system is functional, and this vaccine will promote de novo in specific immune T cell response and also amplify existing T cell response. Very attractive concept, but you very well that clinical data coming from the past clinical trial has been disappointing so far. Obviously we need new technology to make cancer vaccine a success. The first key element for success of vaccine is choice of the antigens. In the past years, the focus has been on onco-fetal antigen or antigen which are expressed by tumor cells. Those antigens are poorly specific because they are also expressed by normal cells and they are poorly immunogenic. That's the reason why Transgene focused its R&D development on two categories of antigen. The first one is oncovirus antigen, and more specifically TG4001, which is targeting HPV. You know that HPV sixteen specifically, is a strong oncogenic driver and is associated with a number of tumor like head and neck cancer and anogenital cancer. The other category of antigen where our focus has been made is neoantigen. Neoantigen are specific to each patient's tumor, so it's a personalized vaccine approach. Obviously, neoantigen produce very strong and potent specific immune response. The other key element for the success of a cancer vaccine is the choice of the platform. Antigens by themselves are insufficient to promote a strong stimulation of the immune system. They need to be adjuvanted. There are several platforms which are used, and some of them are illustrated here, like dendritic pulse vaccine, DNA, RNA, peptides and viral vectors. Obviously what we have seen is that viral vector induce a strong and durable stimulation of the innate and adaptive immunity. Having said that, it is important to show you and to convince you of what I am saying, and I will leave the floor to Eric, who will provide us with a demonstration of the potency of viral vector and more specifically MVA. The floor is yours, Eric. Thank you, Maud, and hello to everyone. Next slide. I'm Éric Quéméneur. I'm the Chief Scientific Officer of the company. I will, as Maud said, give you some details on the value of viral vectors compared to other available technologies for the design of cancer vaccines and share with you what we think are strong points in favor of MVA compared to other possible viral vectors. Rightly, Maud also recalled that designing a successful vaccines is an art of combining the choice of right antigens from neoantigens to oncoviral antigens. It's also a question of selecting the most suitable target populations. You'll see in our presentations that we take a lot of care in analyzing and phenotyping our patients during and after treatment. The third pillar in this successful design is the selection of the good vector. We have chosen viral vectors for the reason I will detail here. The first one is of course to rely on their strong biological properties. Viral vectors are intrinsic immunological devices or nano-machines built to interact with the human immune system. We know that most of the vectors would be able to stimulate B cell response as well as T cell response, and pox viruses are well documented to induce a long-lasting T cell memory response. Second aspect of that is the ability to have a potent antigen display. We also know that, among the viral vectors, pox viruses are strong at infecting antigen-presenting cells and DCs, in the periphery but also in the tumor surrounding. The second criteria of choice is of course the easiness for clinical development. Viral vectors, as opposed to viruses, have been used for decades in the framework for smallpox vaccination, so there is a well-documented track record of safety. Furthermore, the flexibility and ability to design a large diversity of viral vectors based on MVA has been also documented over the years, and we take that as an asset for our technological development. Third, but not least, the industrial development point of view is important. This experience we have built over the years with MVA is also very instrumental in our development, in particular for the personalized neoantigen vaccine that depends on our ability to be fast and efficient in designing a product for every single patient. Among the various technologies, of course, there have been a lot of competition, and most of the community has been informed on the merit of mRNA-based technologies. That is nowadays a direct competitor to us. You might remember that before COVID, we also had to face against plasmid approach or DC-loaded systems. We still believe that even after the COVID period, viral vectors have some differentiation point. What we learned from the mRNA development is that they are very good at raising B response. Still discuss whether they are good at raising long-lasting memory response. Despite major investment made by our competitors, we know that there are still bottlenecks in being able to design a product for a single patient. Viral vectors, of course, present a lot of advantages in terms of mode of action and manufacturability. Of course, there are always drawbacks, but our experience with MVA make those limitations something that we don't consider that much considering our experience with the vaccine. If I move to the kinetics of the immune response, what is specific and strong with viral vectors is that they act very early on by stimulating innate response, and that's very important. Most of the technology around peptides or mRNA need some adjuvant component in vaccine for boosting the innate response. We know that we have strong factors in the MVA to early on in the game stimulate NK cells, innate immunity at large, that will be priming the secondary engagement of helper T cells and effector T cells. This kind of long-lasting activity is also forming a key component of the value of the viral vectors compared to other technologies. Many lessons have been acquired during the development for almost 20, 30 years with MVA over the years. The first one is that the vector has proven to be usable in several routes of administration, from intravenous to intramuscular, intra-tumoral, even intraperitoneal. Many experiences and the way we could better stimulate all components needed for strong anti-tumor response. The second, of course, is obviously the ability for large reengineering of the vector. What is noticeable is the large cargo capacity, with the capacity documented to be up to 25 kB. That allows us to insert either very large antigen sequences plus stimulating components like cytokines or antibodies that can be built in the vector. The one important point is also the ability to really nicely interact with the several steps in the process of mounting a long-lasting immune response, especially the ability to engage intrinsic adjuvant properties of the vector, especially manipulating the TLR3 pathways, plus the ability to also favor the Th1 skewing in the T cell response. Here are two examples of recent products we have developed. The first one is TG4001, for which Jean-Pierre Delord presented the recent data. Please notice in this design that we have made the choice of encoding the entire E6/E7 protein sequence. To present to the immune system the largest set of T and B cell epitopes that one can imagine. They are all available for presentation, and we know also by antigen processing that we are not restricted in terms of HLA typing. Any kind of patient type could manage a presentation of the vaccine. We are not restricted in that sense. For TG4050, which is the product expressing private neoantigens, the... Our expertise in protein engineering allowed us to make this design presented here with three different promoters encoding three polytopic cassettes, each expressing ten neoantigens linked all together by a small link of peptide that was optimized for optimal antigen display. These are two examples of what could be achieved and of course, much more is to come in terms of molecular engineering. If I come to the clinical experience, these are again three examples of recent developments. Some of you might consider that they were not success, but what is clear is that all trials demonstrated the superiority of the MVA arm compared to the reference group. We had very significant response rates and benefit of survival in those advanced patients where most of the immunotherapy approaches failed. As Hedi said, research is very active in our field, and internal research has been focused on optimizing the way we design our product. That's key for the TG4050 program, because you might know, and that will be recalled by Christian, that a fast responsiveness to the vector design is needed. In a matter of months, you have to design from the tumor sequencing a customized product in less than three months if possible. We have developed the very specific techniques to be fast in the cloning and characterization of the vaccines. We also optimize the way we design our polytopic assets. We have an internal software that is called Vac Designer that help us make the better prediction of the way we should organize immunopeptides in the vaccine design to be sure that we have a peptide for every single one. The curves would highlight the progress made over the last two years in the rate of success for antigen design. Today we are close to 90% of the patients that could be successfully treated and proposed the vaccine when the trial is right. The box on the right is an example of brand new discovered pox virus that is not yet in the clinic, but for which we are working at positioning. It's a PCPV, another pox virus, different from MVA and other related vaccines in the sense it can induce very large amount of interferon alpha at the site of administrations, so boosting dramatically the innate engagement. These are examples of the progress that has to be done in parallel to the clinical designs. In summary, I hope that I convinced you that viral vectors are very differentiating carrier system for vaccine presentation. MVA, among the viral vectors, is well suited for T-cell response and very different from other type of immune response generated by other small viruses, enabling the presentation of very large antigen sequences. What is also important is that the experience gained over the years make us very reactive and able to design efficient clinical trials. Next, presentation will be Dr. Jean-Pierre Delord, and maybe a few words from Maud. Thank you so much, Éric, for this convincing demonstration on the capacity of MVAs to stimulate the immune system. Now we will have the presentation from Jean-Pierre Delord, who is the General Manager of IUCT Oncopole of Toulouse, on the updated clinical data we have on TG4001, our HPV16 therapeutic vaccine. Jean-Pierre. Thank you, Maud. Thank you to all my colleagues from Transgene. We work with Transgene for more than 10 years, I guess, because we are at the beginning of the story of the MVA proof of concept studies. I can't remember exactly when it was. I guess more than 10 years ago. It's always a pleasure to meet my colleagues and to work with them every day. I'm a medical oncologist by background, and I'm happy to present you some of the first results we have with proof of concept study, the first in human and the beginning of phase two trials with TG4001, which is a therapeutic vaccine currently in phase two. I can Present. Merci. I'm the current general manager of a new big cancer comprehensive center in Toulouse in France. It's a pleasure to work with all the teams, especially those who are implicated in clinical research. We have a terrific phase 1 team and also at the translational level with Professor Ayyoub, who is one of the key opinion leader in the field of immunology and especially the T lymphocyte specific of antigens and especially tumor antigens. Next slide. This is a team involved in the head and neck program. You can see all my colleagues involved in the recruitment and the care of all the patients we have recruited in that trial. This is one of the key messages. The slide has been, I guess, asked by my director of communication. I'm just happy to say that last year we published close to 700 papers, including the clinical and the fundamental research. It means that we publish two papers a day except on Sunday afternoon. The team can have some rest. There's no big issue with people. As you probably perfectly know, HPV is worldwide health concern. If you read the paper published by the World Health Organization. Each year, more than 700,000 people could be facing a HPV-induced cancer. In the western countries, we consider that more than 25,000 patients a year have diagnosis of cancer induced by HPV, mostly oropharyngeal, perineal, and of course, as you perfectly know, cervical cancer. In France, the number of new patient a year is between 5,000 and 6,000 a year. Next slide. This is a slide describing what are today the standards of care. I must say that treating patient with a HPV cancer is unfortunately still a prehistoric story because we're treating this patient with only surgery sometimes, but mostly radiotherapy. Even if the radiotherapy has made in the recent years a lot of technical progress, it's still a treatment which provide to patient a lot of poor condition after cancer. We usually use with that radiotherapy very old chemotherapy drug called cisplatin. Cisplatin is also one of the most dirty chemotherapy drug we have in the daily practice. Very efficient to radiosensitize the cancer cells, but still very tricky to use in the daily practice. That's why once a patient relapse with an HPV-induced cancer the options to treat that patient are very limited, and the prognosis of metastatic HPV-induced cancer is very poor. The other targeted drugs or chemotherapy compounds doesn't provide any good option for treatment and to modify the overall survival of the disease. That's why immunotherapy, first generation immunotherapy like PD-1 antibodies has been tested in the clinic. The results are not so good, but it is also the sort of proof of concept, considering that some patients will benefit from the treatment. If you pool all the data, trying to see what's going on with HPV-induced cancer patient when the disease become a metastatic disease, this kind of treatment works between 10%-20% of cases. It's not enough to modify the overall natural history of that cancer. We have some cases in the daily practice of patients who benefit from the drug with a partial or complete response. It means that reactivating their immune system can be an option for patients with an HPV cancer when it becomes a metastatic disease. That's why we have designed with the support of Transgene a clinical trial with the TG4001. As it was mentioned by Eric a few minutes ago. This is a DNA construction with two major epitopes of HPV, E6 and E7. Also the integration of a boost by IL-2, so interleukin-2, which is one of the major cytokines in inducing T-cell response against many antigens, including viral antigens. All the construction is done on a modified virus and injected subcutaneously to patient in the trial. Again, I guess that what you see on the left panel has already been presented, but I would just like to focus on the result concerning intraepithelial carcinoma of cervical cancers due to HPV. It was the first time in the literature where we could observe in the paper published by my colleague that complete disappearance in between one-third and 40% of patients with a tumor, what we call an intraepithelial tumor. It's just before the cancer becomes an invasive form of cancer. All the patient was treated after the vaccination by surgery. You can see that the response rates, which was a complete histologic response rate, in that subgroup of patients with an infection due to HPV 16 was observed. Again, it was really a strong signal showing that the therapeutic vaccination could do something directly against transformed cancer cells. This is the characteristic of the patient included in the phase I part of the first human trial with TG4001 plus PD-L1 antibody. We directly do the phase 1 trial with the PD-L1 antibody. As you can see, I'm not going to go deeply in the details, but as you can see, the patient were what we call IO-pretreated patient. It means that the number of organs involved by the metastatic disease was superior to two different sites, and the number of previous line of treatment was between two and three. As you can see, the patient, of course, was a patient of first human trials. They were still withstanding, but they had been previously heavily pretreated with other anticancer drugs. The schedule of injection was the following one. There is an injection phase when we vaccinate the patient once a week during the two months. And then we do a new injection every two weeks. And of course, the treatment with the PD-L1 antibody was used all the time, and the tumor assessment was done every two months. This is a very conventional classical schedule for assessing the safety in phase I trial, but also the efficacy like we do in again classical phase II trial. We have already observed that TG4001 plus avelumab has definitively an anti-tumor activity in the overall patient population. We have more than 40 patients with disease control. I'm going to go more in the detail concerning the level of disease control we did observe. Response rate, it means tumor shrinkage. Just to let you know what we call a partial response in oncology, is that we have the volume of metastasis that decrease of more than 30%. It means that we can see with a classical CT scan or MRI a clear tumor shrinkage that is reproducible at or at least two times after the first CT scan compared to baseline. What is also very enthusiastic for the people in the clinic was that we observed one complete response with the patient that are still doing well. This was a very special patient with an anal cancer and with metastasis on the peritoneal surface. This is something in oncology when that kind of cancer is reaching the peritoneal surface. Usually, the chemotherapy doesn't work at all. The prognosis of that kind of tumor invasion is very poor. Usually, patients are passing away after only a few weeks or few months, whatever you do using conventional chemotherapy. The question we ask was, it's the way we're doing first in human trials in oncology. We usually include first patients that have been already treated with a large number of drugs already registered in that indication. What we have observed is that the patients with very large number of metastases, and especially those with liver metastases, had a very poor prognosis. For many reasons, I don't wanna go back to scientific background concerning liver metastases, but we all know in the clinic that liver metastases have a special prognosis in oncology. We did observe that those patients had no benefits under treatment. Excluding the patient with that special clinical presentation of metastatic disease, we observed that the overall response rates were observed in one-third of patients. The overall survival was more than one year. Meaning, again, that the proof of concept was there. Adding a therapeutic vaccine and a monoclonal PD-L1 antibody was probably doing something that was very, it was at least a very good signal in that very special subgroup of patient. This is another way to show you the results. It's something we call a spider plot. It's a longitudinal report of the size of the tumor from one patient compared to the other. As you can see here, as an example. The size of the metastatic disease is increasing very quickly in that subgroup of patient. For a large part of the patient, the size of the metastasis is decreasing very largely with the size of metastasis of less than 50% of them. One patient with a complete response very quickly. All the patient with what we call a long-term effect of the vaccination and the stimulation of the immune system with a follow-up and we're waiting for the next follow-up in the next few months. I guess we will get it very quickly. There is some patients with a very long-term benefit with no sign of any progression of their cancer. It means again that without chemotherapy, without radiotherapy, you can control or probably do more than a simple control of the disease, but obtaining a clear remission of the disease in one third of patient again. Next slide. Now this is the first results of the monitoring of what we observed in each patient. On that panel, this is the level of response against E6 and E7 at baseline before receiving any injection in the trial. As you can see, there was no patient with any cytotoxic T cell specific cells in PBMC that has been assessed by ELISpot. After the vaccination period, there was seven patients with very clear responses observed at the peripheral level. It means that probably the very good signal we observed during this phase 1 trial was clearly due to the induction of T cell response specific to the antigens of the HPV virus. This is what has been observed at the tissue level within the metastasis of patients. We have made a sequential biopsy program in order to see what's going on. Most interesting result is probably on the median panel, where you can see that CD8+ cells, it means the cells that contains the T cell, the memory T cells, that are in cancer, the tumor cell memory. But also those who are the effector of the cytotoxic effect were increasing in numbers within the tumor. It means that all the T cells were probably acting close to the cancer cells, and the tumor shrinkage were linked to the cytotoxic effect of the combination of the vaccine and the PD-1 PD-L1 antibody. There is a current comparative randomized phase two trial. We have decided, in actual discussion with all the investigators and our colleagues from Transgene to exclude, from the phase 2 part of the current trial, the patient with liver metastases. As I mentioned before, we do believe that these patients do not have a clear benefit from the combo. We are currently including patients with a recurrent metastatic and of course have not been previously exposed to cancer immunotherapy, and whatever the level of expression of the target of PD-L1 antibody. It's a randomized trial where we compare directly to the standard of care, which is PD-1 or PD-L1 treatment, the combo of TG4001 and PD-L1 antibody. The primary endpoint is a mix between tumor shrinkage and long-term benefit we call a progression-free survival. We are expecting the results very soon because the recruitment of patients in that trial is good and a little bit superior to the recruitment that was expected at the beginning of the trial. I guess between I don't know, but between the end of this year and early next year, we will get all the clinical results concerning that comparative phase I trial. Thank you for your attention. Oh, yeah. No. Okay. Sounds good. No. The most important information is that the interim analysis are expected in the next three months. Yeah. Before the end of the year. For Christmas. Thank you so much, Jean-Pierre for this excellent presentation. Truly, Jean-Pierre has shown the evidence that the combination of TG4001 and avelumab is active in advanced HPV-associated cancer. This is a randomized phase II trial, which is actually the first phase II randomized trial in anogenital cancer. As previously mentioned, we are expecting very soon the result of the interim analysis before the end of the year. We are going now to move to the other categories of antigen I mentioned in the introduction, meaning the neoantigen. Here we are speaking about a personalized approach, meaning that a batch of vaccine is prepared for each individual patient. Here, the clinical positioning is different. We are testing this personalized vaccine in early stage of the disease as a single agent, and again, it is directed towards the specific tumor-specific neoantigen of the patient. What do we mean by early stage of the disease? You know that when a patient is diagnosed with cancer, usually the tumor is surgically removed. Unfortunately for some patients and for some tumor types, the diagnosis is made at an advanced stage of the disease, which means that the risk of relapse is quite high. Even though the patient is in apparent clinical remission, based on the standard imaging procedures, there remain some micrometastases which may lead to relapse. That we call minimal residual disease. This MRD offers a very good opportunity to test a personalized vaccine as a single agent. The goal, of course, being to delay as much as possible the relapse and even if possible, to prevent relapse. We have two ongoing phase I trials, and we have selected two indications to test our personal vaccine TG4050. One is ovarian cancer and the other one is head and neck cancer. The patients who are selected in our phase II trial are patients in clinical remission, so meaning in with minimal residual disease. We know that in the those two indications checkpoint blockers have limited or no efficacy. It has even been shown recently at the last ESMO Congress in head and neck with pembrolizumab, a negative phase III trial in that setting. The immune system of those patients is functional because they received a limited number of previous lines of therapy. Those two cancers have a low to medium tumor mutational burden, which means that it makes easier for us to identify the neoantigen, the immunogenic neoantigen that will be included in the vaccine. Maybe a few words on neoantigen. What do we mean by a neoantigen? Neoantigens are antigens specific of each patient's tumor. They are supposed to induce a very, very strong immune response. They are 100% tumor specific, so they have not been subject to central tolerance. They are more immunogenic, obviously, than the so-called classical shared antigens. We are able, in our vaccine, to introduce up to 30 neoantigen. For doing that, obviously we need good tools. One important tools for this identification is obviously the use of artificial intelligence. We will have now a presentation by Cédric Bani-Jama, who is the scientific leader of the neoantigen program, on the use of artificial intelligence in general, and more specifically in clinics. Thank you, Maud. Good afternoon, everyone. Just during this afternoon, a step aside to speak about this term of artificial intelligence. You will all be hearing it more and more. You've been hearing it over the last few years, and many of you are watching it with some wise skepticism. The first things I'd like to point out is that what we call artificial intelligence encompasses a large array of technologies. There are large number of approaches that are already widely implemented in our daily lives. There are many uses of so-called artificial intelligence system. Some very trivial in your phone, in many application of your phones, and some in more regulated environment. In banking, a lot of use for artificial intelligence have been emerging those last year or even in some critical application, highly regulated environment like in air transport. You find a lot of artificial intelligence in commercial airliners, particularly those made in Toulouse close to the E.U. city. Healthcare is no exception to that. Today, there is more than, or at least there is at least 100 different devices that are registered by the FDA that use some sort of artificial intelligence or some sort of machine learning to make decision about the care of patients. This is in various medical specialties. It's essentially in the U.S., it's in oncology, but also in critical care or in some other application. Some of those decisions are critical for the patients. For instance, you get system for the triage of patient in intensive care. You get eligibility for some critical therapy procedures. Those systems have reached a level of maturity to allow them to be robust enough to have a regulator review them and approve them for use, which is critical for the patients. This evolution is not something that happened from one day to the other. It's actually an evolution that has taken decades. It started very early. When you look at in the late sixties or early sixties, they already had what they called intelligent machine, and it started with a diagnostic. Those machines were used to automatically generate data for the patients. Mainly they were some very early markers for use in critical care. By quantum leaps, artificial intelligence technology has evolved from intelligent machines. We went to statistical learning or so-called machine learning, which are like traditional approaches of artificial intelligence to move toward deep learning and what we call now foundation models. Foundation models are the most advanced models. They're not very much used in healthcare now, but they have some other use, especially in language or translation, for instance. Today with TG4050, we're really making use of what we call deep learning, and I'll come to that later. In parallel to those evolutions and those progresses in machine learning technologies, oncology had like a parallel track of complexification and sophistication a long time. Of course, the care of cancer patients have improved a lot over time over the last decades, but this came with a trade-off of complexity because we started from one indication, one recommended treatment, very easy, very simple guideline to apply as long as you have the diagnostic for the patients. We moved then to a more characterized, more molecularly characterized disease. We targeted therapy. That's what I called here oncology 2.0. We went through a more and more data-intense process of characterization of the patient and therapeutic decision-making. With TG4050, we are at a convergence point, basically we're at an edge where the data you get for every single patient, which is a whole exome sequencing and a full transcriptomic of the tumor and the patients, where this amount of data becomes too high to comprehend for the human brain and the single physician. That's where artificial intelligence is really of help. This help comes only because the performance of this technology have increased a lot. I've spoken about deep neural network. I will explain how they work later. Deep neural network are the AI technology, the machine learning technologies that allowed those approaches to have a much higher performance over the last decades. Why? Because those approaches can comprehend and take into account a lot of data, a lot of input data, without necessarily deciding which part of the data is important. It just takes all the data during the learning process, have this data used to learn and to teach the system and come out with a conclusion or come out with information. Unlike other approaches, there is no human, I would say prejudice or human hypothesis on which part of the data is important. It's not this gene or this gene. All the data gets in, and the system itself will select what is relevant or not for the final outcome. This has driven a lot of performance. For instance, there is this famous example of computer vision in a given task, where machine learning had a 26% error rate in 2011. Today, we are less at 3%, which is basically turning the table face to the human contender, which is around 5%. Those approaches had allowed a tremendous increase in performance. I've spoken about neural network. I'm not going to get into details about the mathematical way neural network works. But what you have to keep in mind, it's basically a compression of information. You start from a complex information, and here the complex information is a picture. A picture is a complex thing. There is a background. There is a what is on the picture. It varies from one angle to the other. There's a lot of complexity, basically, in the input data. You go through this process, and you come up with the information you want, which is the animal that is on the picture. Is it a cat or is it a dog? This is a very simple information that goes into only three letters. You don't need all the information that is in the picture. Basically, a neural network is a way to start with a lot of information, many of it relevant, many of it irrelevant, and come down to the very single information you want. It's transforming a lot of data into relevant information. How it does that? It does that because it's organized as unit of calculus that are like neurons that are organized in layer. For instance, the first layer is going to take each one small part of the picture. The second layer is trying, I don't know, to look for whiskers, for instance. The third layer is to say, "Oh, I see many neurons of the second layer. I've picked up whiskers, and they're organized symmetrically around the mouth, then it's a cat." Then you come up with the information. It's a way to lose the complexity of the information, gain the relevance, and compress the information. When you think about it, I started with this cat and this dog, but then you apply it to cancer. Cancer is actually a data mess in terms of genomic, because you get variability across patients, of course, and this we all know, everyone is different. You get variability across the diseases. Many people with the same disease, but the disease will have a lot of different molecular feature. Within the same patient, you'll get variability across space, meaning in different position in the tumor, into different metastases, but also along time. Because when you treat these patients or the way you let the disease evolve, it has also changes into its complexity and into its genome. A lot of variability, highly heterogeneous, highly parametric, and the point with the immune system is that you get all this mutation in the tumor, but only a very small fraction of those mutation are going to be relevant to make a vaccine. Actually, if you look at all the mutation, one of the numbers of relevant mutation for the immune system would be between 1% and 5%. It varies, but it's only a minor fraction. When you wanna do a vaccine based on those mutation, you better have a system that help you to select what are the relevant mutations. That's what we do, and we use deep neural network-based technology or a flavor of this approach that was developed by a Japanese company called NEC that was mentioned before. NEC have been working on two aspects. The first aspect is the mathematical basis and the technology of machine learning. They've been one of the drivers in this field with some of the top machine learning scientists working in the research laboratory on one side. On the other side, they've been working on the learning of this system by generating data, of course, public data, but as for proprietary data, to feed the system, to have the system learn how to recognize the good antigens. This is something that happened throughout the last decade. They've been doing that for almost 10 years now, and that's the system that we have built into our TG4050 process to design the vaccine. This vaccine is now into its first phase of study. A lot of people wonder whether this is going to be a valid, so-called validated system, whether it's going to be efficient. It's too early to answer now, but we do have some data that are very encouraging. First, the colleagues at NEC, when they wanted to implement that in the clinic, they did some exercise to see whether this system had the capacity to detect what are relevant mutations. They've started with existing data on T-cell therapy that were established by the Rosenberg's lab. You know what he's doing. Steven Rosenberg, he's taking patients, looking at the mutations, and testing mutation by mutation, which one is actually leading to an immune response that is beneficial for the patients. Across 7 clinical trials, he did that with more than 700 mutations, and they identified 23 clinically relevant mutations. What we did here in this exercise is that we take the system we use into our trial, we feed the system the 700 mutations, and we ask, "Among those 700 plus mutations, can you tell me which ones are actually relevant for the patient?" We see whether this system is able to find back the 23 that Rosenberg has shown as being beneficial for the patient. It depends, of course, on the question of threshold and so on, and those familiar with the performance of a diagnostic will see that we get an AUC under the ROC curve, which is more than 0.99%, which is an excellent performance for such system. If we were to use classical approach based on expression and binding like nivolumab, we would have a much less performance. This is obviously a retrospective exercise, and it's not data from our studies, what I've just shown. Now when we look at our patients and our phase one that we are running, we have data. At the time, we had data for six patients. We did this little theoretical exercise. For those six patients, we had more than 2,000 mutations that were expressed and somehow binding in terms of MHC and in those six patients in total. The total for those six patients, we generated, of course, six different vaccine, and we used one hundred fifteen different targets suggested by the AI. Among these 115 targets, 36 turned out to be immunogenic in the patient. You look at 36 positive by ex vivo ELISpot out of 115. Is it a lot? Is it not enough? We don't know. But still, if we were to select them randomly, just looking at binding and looking at expression, we would have less than one chance out of 118 billion to achieve a similar performance. I can tell you today that the AI system will do a decent job, a sufficiently good job to treat the patient and deliver to the promise. What I can tell you is that it's doing much better than having like a very classical, "I look at what is expressed and what's binding, and I select it." One chance out of 118 billion is not a lot, and you have much more chance to win at the national lottery than achieve this type of performance. This is antigen design, vaccine design. We have other use of AI because basically we start from the same data. We generate a lot of data of each patient, and we try to create value out of this data beyond having the vaccine. One of the first things that comes into his mind is to use this data to characterize the immune contexture and characterize the patients to better understand which are the patients that are actually responding to the vaccine, which are benefiting from such approach, and which one that probably benefits less and that we probably should not treat. Using this type of system, we can characterize each one of our patients and compare it to the diseased population globally, so we can see how our patients compare to the global diseased population, whether they are rather hard to treat, whether they were more like good prognosis patients. That's something that we can do, and we can classify them from immune desert to immune enriched and so on, and those classical immunological characterization. As well as patient by patient using AI. For each individual patient, we can create a profile and see whether this patient would normally benefit from other existing therapies, whether we are actually having a patient benefiting from the vaccine, while this patient would normally, according to the existing literature, have less chance to benefiting from existing therapy. So basically, it's a way to better understand our patient and better target our vaccine to the right populations. That's something that we do for exploratory purpose, but that we do on a routine basis on our patients. In a nutshell, massive amount of data, systems that are getting mature enough to achieve regulatory review. It's adaptive, it's learning from each time it works. It basically fed back with the data we're generating, so it's improving over time. It's only at the start, 'cause today it helps designing the vaccine. Tomorrow it will also help target the patients and provide further information to the clinician that is probably of value, beyond just providing the vaccine. Now, I'll let the floor back to Maud, and we'll get into more information about our clinical trial on the neoantigen vaccine. Thank you so much, Cédric, for this very well-documented presentation. We have now in our hands and in clinic a very innovative vaccine based on this personalized approach. We use the MVA, the MVA vector, and we have seen with Éric the capacity of these viral vectors to promote a very strong and durable stimulation of the immune system. We have also in Transgene the GMP manufacturing capacity. We are able to produce for each patient the batch needed to treat him. We have the technology which are needed to identify the immunogenic neoantigen based on the artificial intelligence with the support of NEC, which is also supporting 50% of the cost of our development. We have also internally designed this Vac Designer, which is able to optimize the selection of the cassette and the manufacturing capacities for the vaccine. Last but not least, we have also skilled and very enthusiastic investigators. Thanks to them, we are able now to present you clinical data. I will leave the floor to Christian Ottensmeier, who is a professor of immuno-oncology at the University of Liverpool. Christian, floor is yours. Thank you so much, Maud, for the very kind introduction. I can hear myself echo, so I'll try and do my best to only speak once to you. I think the really extraordinary perspective here for us is that this trial suggests that this approach is safe. It's immunogenic, and it looks as though it does what it's supposed to do. In the early suggestions, it seems to reduce the risk of recurrence for patients with head and neck cancer. My presentation will look at some of the data that led us to this trial, review with you some of the early data that we have, and then I will offer you my personal prejudice on what we are learning here. Cancer immunotherapy in a nutshell is an intriguing challenge. If cancer develops, the immune system must have failed. For the clinicians and for the scientists among us, we need to try and work out, can we reestablish immune control? What does that immune control look like? At the patient level, that's quite simple, the cancer disappears. At a cellular level, what does this actually mean? If we can understand what it means, how can we get there? If we think that the immune system is able to attack cancer, then we need to have a look at the first examples of successful immunotherapy. Clearly, that is very well recognized in allogeneic transplantation, where giving a healthy immune system into someone who has cancer and who has failed effective treatment can restore the immune control of the cancer but at significant toxicity. Since 2011, the world has changed in solid tumor oncology also, and immunotherapy has become a standard of care treatment with really stunning clinical benefit in individual responders. Nonetheless, the outcome is that only about between 10%-25% of patients currently benefit across the board in many solid tumors from immunotherapies with anti-PD-1 or anti-PD-L1 antibodies, and I've listed the tumors on the right-hand side where such data are commonly recognized. When we look at the immunological underpinning features under the microscope, then it's relatively simple to count T cells that are in the tumor. In the slide on the left, in the left part of the slide, in the histopathology image, you can see a group of adenocarcinoma cells with a lumen in the middle, the nice pink rounded cells. In between the darker dots are the immune cells, in this case, CD8 T cells that are trying to attack and to remove the tumor. This simple quantifying the immune cells already gives us a really good sense for what might happen to the patient as a cohort, but not to the individual patient. On the right-hand side, I've picked out a reference dataset from HPV, from head and neck cancers. In blue, those tumors that are driven by human papillomavirus or HPV. In black, the tumors of the type that we are vaccinating the patients against, the HPV negative tumors. You can see that if you have lots of T cells, and these are directed against the virus, many patients survive, the top blue line in the survival curve. Whereas if you have tumors that are not driven by human papillomavirus, that are the standard smoking and alcohol-induced tumors, and if you also have tumors that are low in these immune cells, the bottom line in the black lines, that more than 60% of the patients eventually die. That's the group of patients that we're targeting here in our trial. The sheer enumeration gets us so far, but what we need to understand is what do these T cells do and what do they recognize. I've just put in the red arrow here to identify that that's the group we're targeting in our trial. Over the last decade or so, we've learned much about the immune cells, and, specifically, we found out that particular groups of immune cells that live in the tissue and do not recirculate are really important. We found out that it's not just the number of immune cells, but also the quality of the immune cells. This is a dataset of just short of 600 patients, in this case with lung cancer, where the stacked bar chart on the left identify the type of immune cells in three categories. On the very left, few immune cells, on the right, many immune cells, and in the middle, an intermediate amount. The colors in the stacked bars identify the cells that are, of this tissue-resident feature, and they are the pink ones. That on the right-hand side in the survival graph identified what happens to patients whose tumors have lots of immune cells of the right kind. These tissue-resident memory cells identified by a protein called CD103, which is essentially a molecule that makes the T cells stick in the tissue. It's the functional outcome of this tissue residency program. You could see that if you have those, even if you have lung cancer, then you're likely to do quite well. If you don't have these cells, then you do really badly. We can count these quite easily by immunohistochemistry, and the bottom-hand panel identifies on the top row a TIL-high tumor with PD-1, CD8, and CD103 in the three panels, and at the bottom of the same slide, a tumor that is low in these immune cells. What we've learned, and many groups have now confirmed that these tissue-resident memory cells are highly effective killers. They expand in the cancer tissue. They express unique and actionable targets. As I've shown you, they're easily quantified. Very importantly, they underpin tumor control in mouse models very easily to model. In humans and in mice, these cells are activated by anti-PD-1 antibodies, but you need to have enough of them for that to be the case, and then can be trained, expanded, and activated by vaccination. Of course, that takes us to the area that we want to talk about today. What is much less well understood is what these cells actually recognize. We know that they are tumor-reactive and that there are multiple possible antigens that can be recognized. The slide, the right-hand side of the slide we've already seen before, because it categorizes molecularly defined antigens according to how similar they are to what is present in our healthy cells. At the bottom left, the self epitopes, which are not very good at activating immune responses. Shared antigens that many groups have targeted with some effect, including Transgene in the MUC1 study. The oncoviral antigens that we've heard about from Dr. Delord, and then the neoepitopes, which are really quite appealing because there is no central immunological tolerance, an attempt of the human immune system to limit these cells built in because these are very different, they're generated in the tumor, and immunologically, they therefore would appear to set a relatively low bar. The key bit is that the immune cells that are recognizing epitopes generically and specifically neo-epitopes are being presented with these epitopes on antigen-presenting cells. On the left-hand side at the top A are the antigen-presenting cells, in our case, the tumors. Then these express molecules called MHC class I and class II molecules. In my head, I imagine these a little bit like a silver platter on which the cell presents its identity to the outside world. Along come the T-cell, sniffs the tumor microenvironment, and if it recognizes something, it latches on, and that's what the bottom part says. This is the T-cell receptor that does the latching on, and then the costimulatory molecules turn that connection, that latching on, into biological function. The point I'm trying to make here is that the T-cell receptor is unique for each T-cell type. Therefore, you can imagine it a little bit like a molecular fingerprint for the T-cell, and wherever that T-cell is, whether in the blood or in the skin or in the tumor, it will have that same T-cell receptor. We can consider and then exploit the T-cell receptor as a molecular barcode for a particular T-cell type with a known or unknown reactivity. In an ongoing program of systematic probing, which is in parallel to the work that the Transgene's team is doing, we're collecting blood from the patient. We're stimulating the cells with antigens, in this case, tumor antigens, to find reactive cells. We identify the T-cell receptor, i.e., the molecular barcode, and we can now say, "Well, a T-cell that is reactive to a particular neoepitope has this particular molecular barcode," and therefore allows us to have a unique identifier for this type of cell. If you then go and do the same in the cancer tissue and analyze all T-cells, you can then work out which ones of the cells that you've identified in the blood are also present in the tumor, and therefore allocate reactivity to the T-cells in the tumor without having to do very sophisticated analysis and single-cell rescue in the tumor microenvironment. This kind of tool is now becoming readily available. We can apply that to any antigen of interest, for example, viral antigens, shared antigens, of course, neoepitope antigens. The really neat bit is that we can do that now in any tissue, starting in the blood, going into the cancer, and also looking at sites of toxicity. This kind of platform technology begins to enable us to probe and to track antigen-reactive T-cells with known specificity throughout the patient. The initial discussion now some years ago with the Transgene team, we wondered whether such tumor antigen-reactive T-cells existed not just in tumors that have lots of T-cells and not just in tumors that are responsive to anti-PD-1 therapies, but whether these could be present in tumors that are immune cold. The prompt for this investigation was a patient who was in trouble, a gentleman with non-small cell lung cancer, and of course, he in this case represented a huge clinical need. This was a tumor with a relatively low tumor mutation burden, did not express PD-L1, and his cancer had progressed on chemotherapy and anti-PD-1 treatment and radiation treatment. We were lucky enough to be allowed to collect a large quantity of blood from him, an apheresis product, and were also able to access both the primary tumor as well as a metastatic deposit. Here's the morphology. The at the top are the T-cell densities, which really shows not very much. At the bottom panel, you can see that MHC class I is expressed, bottom panel on the right, and that MHC class II is also expressed. In theory, this tumor should be visible to T-cells. What we did then is to compare the number of T-cells, in this case using transcriptomics, and quantifying the markers for CD3, so a pan T-cell marker, CD4 and CD8, as well as markers of tissue residency, CD103, bottom left, PD-1 and PD-L1. You can see that in each of these instances, our tumor sits at the rock bottom of the cohort. Each round circle is one patient. Each filled circle is a head and neck cancer patient. Each empty circle is a lung cancer patient. We've contextualized the biology of the primary tumor in pink and the metastasis in turquoise in the context of a larger set. You can see it's perhaps not surprising that immunotherapy with an anti-PD-1 antibody wouldn't work because there aren't very many T-cells that correspond to being released. We wondered whether we could identify the neoepitopes that might drive this, and we sequenced both the primary and the metastatic disease. We found about 1,000 or so single point mutations. Out of these 1,000 mutations, 18 genes allowed us to predict neoepitopes. In other words, the transcriptome allowed us to suggest that these might make it into protein. We made a construct with or we identified with the peptide sequence and then in collaboration with the Transgene team, and the Transgene team actually made the construct and assembled a personalized cancer vaccine for us. We were not very quick in this process. By the time the vaccine was ready, the patient had progressed and then was no longer able to receive the vaccine. We were able to, however, test whether we could recover the reactivities in a surrogate model. The patient had an HLA-A2 restricted genotype in this MHC molecule, and just by circumstance, there is a well recognized mouse model which also makes the same MHC class I molecule. We tested in the blood of the patient whether the neoepitopes that we had predicted were able to stimulate T-cells, and indeed were able to find T-cells that made interferon gamma in response to the peptides, not the non-mutated germline counterparts. We were then able to look at the T-cell receptors of these reacting cells, and to our surprise, were able to track such T-cell clones both into the primary tumor as well as into the metastatic disease. Identifying that even in a tumor with a low tumor mutational burden, T-cells exist that have a counterpart in the blood that have a T-cell counterpart in the tumor, as well as in this one case, metastatic disease. We weren't sure, though, whether the vaccine would be able to activate immune cells, so we took this vaccine into a mouse model and were able to reproduce many of the T-cell responses that we had recognized in the patient, also in the HLA restricted mouse model, and CD8 responses to this, these epitopes could be recovered. Intriguingly, we're not able to find the CD4 responses, but that's unsurprising because the human CD4 genotype was completely different to the gene restriction element in the mouse. The science really hung together very nicely. The question then is, and that is what takes us to where we stand today, if we have enough of the right kind of T-cells and the PD-1 treatment is really good, solves the cancer problem, but that's only in a small minority of patients, and we need to exceed the state of the art by training more of the right T-cells. Now we are right back to where we started with our vaccination program. The aim of the TG4050 program is to translate this exemplar data into proper medicine and to establish the relevance of cancer vaccines and demonstrate the power of neoepitope vaccines. We spent a lot of time and many, many discussions with Maud and Kaira and the Transgene team about what the right setting might be for the initial study. We decided that the best place would be a setting where the patient had completed their standards of care, had the smallest possible tumor load, but had a really high risk of recurrence. We argued that this would be a setting in which there would be minimal immunosuppression because the treatment had finished, and the patient had small volume of cancer, but nonetheless a very high unmet clinical need. That's how we designed the trial, to assess whether we could prevent relapse in high-risk patients. The real puzzle was how would we go about measuring effect if there was no cancer to assess? I think it is really a testament to the collaborative group that seems to have been really successful. We were going to compare immunogenicity to the proof-of-concept data, test safety, feasibility, and then clinical benefit, and that's what's currently ongoing. The trial design is relatively straightforward. At the time of definitive surgery, the cancer is removed. We take a little bit of the tissue and send it off to the team in Transgene. The patient then completes their adjuvant treatment, and in parallel, in this time, the Transgene team generates the mutations. Or rather reads out the mutations and then assembles a vaccine. In parallel, there's of course the opportunity to look at the T-cell receptors of the immune cells that are already in the cancer. When the patient has undergone their resection and adjuvant chemotherapy and remains disease-free clinically, we undertake a leukapheresis and then randomize the patients either to vaccination up front or to vaccination at a later time point. We have a short six-week period of weekly injections, the induction period, and then a longer period of longer interval three weekly injections in which we are vaccinating the patient. Again, at those six weeks four, take a second large blood sample to assess what has happened to the patient. It's really quite astounding. The data here at the bottom left-hand, this tech bar chart, identify evidence from, in this case, two patients, in the ovarian study, and Dr. Block will talk to us about that in a minute, and on the right-hand side from two head and neck cancer patients. It's really quite intriguing that we not only expanded immune reactivities that were present before, but that we also induced significant numbers of de novo responses. In other words, T-cell reactivities that were not present at baseline. The immediate response is 10 reactivities out of 30 put into the vaccine. That's a really quite remarkable outcome because I think we also respect and recognize that the breadth of the reactivity will be important. For these dots here, each color identifies one patient. The top row identifies the baseline, the bottom row identifies the basic default. You can see that by and large, there is a really nice elevation of the dots in the bottom row compared to the ones at the top. Identifying the expansion of reactivities in a very consistent fashion. What we can also see is that as in the previous data sets that Tanja and I published with the MVA backbone, NK cells are activated. In the very left-hand panel, you can see the induction and activation of natural killer cells. This suggests that the virus is highly effective at engaging the innate immune system. Most excitingly though, what we're seeing is also that we are activating effector cells, both in the CD4 and in the CD8 population, the two bits on the right-hand side, and in conjunction with this, a decrease in the overall naïve populations. The T cells that are activated also have evidence of induction of effector markers and memory markers. What is really intriguing is that we had aimed to also monitor the quantity of the circulating tumor DNA because we argued in these patients there wouldn't be very much to be getting on with in terms of measuring radiological response, because by definition, the patient's in complete response at the time of entry to the study, so we can't really get any better than that on imaging. The OncoLearn program suggests that if you look in a more sophisticated way and sequence the quantity of individual circulating molecules that represent the mutations, then you might still be able to pick that up. The data is undergoing analysis. This is data from the ovarian trial that just identifies that ctDNA in this assessed in this method goes really nicely with a more classical marker of CA-125 protein expression, a commonly used marker for the quantity of ovarian cancer cells. The trial here is in HPV-negative tumors enriched for TIL-low. We know that 85% of these patients don't have very many immune cells, and so if you've got an HPV-negative tumor, you're almost self-declared to have a TIL-low tumor. We picked patients with high stage, so large tumor size, nodal involvement, and/or extracapsular spread. Really clinically, this is the bad end of the immune-low cell tumors with a median progression-free survival of just about a year. Just to reiterate, these are the patients that we've already identified in the previous slides as the group that we would be targeting. The study is, as I said, an adjuvant study after resection chemoradiotherapy. The vaccine is given single agent, and we continue until either disease progression or two years are up, and we compare patients who have been vaccinated up front in a randomized way to patients who are followed only, but of course, for whom there is also a vaccine. The really neat piece of this is that in the parallel control group, this will give us the opportunity to vaccinate patients with the same vaccine concept, but now with recurrent and measurable disease. Of course, that also allows us to test whether the unmanipulated interval between surgery and recurrence has been accompanied by any genomic or immunological changes in the tumor. I think in the absence of immunological pressure, that is unlikely to be the case, but we'll have to see. The trial then gives us two opportunities. One is to assess the progression-free survival in the vaccinated versus the non-vaccinated arm, as well as a smaller number of patients in whom we can actually assess whether in combination with standard of care, and the hope that that would be mainly anti-PD-1, this vaccine can regain control of the patient's tumors that have recurred. The data looks astounding to the oncologist. At the top arm are the data from the first set of patients, and you can see the swimmer plots that have been aligned by the length of follow-up. So far in the vaccine arm, those are patients that have been vaccinated early, there have been no relapses. Whereas in the patients who have been followed only, there have been 3 relapses. That is 30% relapses in the patients that have been observed, as opposed to 0% in the patients that have been followed up. Randomized, multicenter, so we expect this to be a true reflection of the reality, but of course it is a small trial. We're looking for a total of 30 patients. The trial is fully enrolled. 20 patients have been randomized, so we've got 10 more to go, and then hopefully we'll be able to confirm that the data will remain the same. The state of the art then is that effective T cells are a condition without which cancer immunotherapy cannot work. We know that checkpoint inhibitors benefit a small percentage of patients. They awaken or release pre-existing T cells. That is great if there are enough, but it doesn't work if there are not enough T cells. Beyond anti-PD-1, the other checkpoint inhibitors have only yielded a relatively small incremental benefit. In contrast, cancer vaccines are, at the moment, the only tool we have to train cancer cells in the patient. My expectation is that they will form the backbone of cancer immunotherapy for the future, and that really will be starting off with vaccines in combination with something else, such as, for example, anti-PD-1 antibodies. That, of course, is for the next iteration of trials to test. TG4050 specifically combines all the features that are likely to deliver clinical efficacy. A highly immunogenic, a very well-tolerated and a really well-understood MVA backbone. We now know that the neoantigen targeting induces specific cells against the neoepitopes and astoundingly, that it is clinically feasible. The trial encompasses really several decades worth of clinical learning and of adaptive learning, identifying and optimizing clinical trial settings and feeding this into randomized multicenter trials. In summary, I think this is an astounding effort because it's safe and immunogenic, and the early data really suggests clinical benefit. If that is the case, then the sample size effect must be large if it is detectable in such a small cohort. I think the smart design will also give us a lot of evidence of the biomarker programs that were defining the mode of action, demonstrating the link to cell-free DNA, identifying markers of early recurrence, and then hopefully also demonstrating that in a therapeutic setting, so the place where most vaccines have not worked so far, this will be able to convince the community that this actually works. Of course, I'll hand over to Matthew Block in a minute, who will talk to us about the ovarian cancer trial. My belief is that this trial and this approach is set to deliver a landmark change in the field by demonstrating the clinical efficacy of cancer vaccination in using, in this case, the MVA-based platform. It will, I think, define a starting point for rational combinatorial immunotherapy because it will start from an optimal patient setting from which we'll be able to work forward and assess what we need to do differently in patients with less favorable setting. I think this will underpin immunological basis-based choices of combinatorial approaches, and of course, the extension to other solid tumors and clinical settings is obvious. I think no one will fail to notice that this is already in a tumor type that hasn't really been the hallmark indication for cancer immunotherapy. Clearly for me, and most importantly, I think this kind of approach offers hope to patients with really terrible and often fatal cancers. Really if we can achieve that, then this is really all worthwhile doing. Thank you so much. Thank you so much, Christian, for this presentation of this first very encouraging clinical data in head and neck. Now we will move to the presentation of Matthew Block, who is Medical Oncologist at Mayo Clinic, and we will speak more precisely on our trial in ovarian cancer. Matthew. Well, thank you. I am happy to be here virtually. I am from Minnesota, but currently traveling in Tennessee. By way of introduction, the focus of my research is to better understand the mechanisms by which cancers interact with the immune system and how they avoid immune-mediated detection and eradication. I study therapeutic vaccines as well as oncolytic viruses as ways to induce antitumor immune responses. I study immune checkpoint inhibitor combinations as means to potentiate antitumor immune responses. As a medical oncologist, I also take care of patients. I treat patients with gynecologic cancers, including ovarian cancer, and I treat patients with melanoma and other skin cancers. Ovarian cancer is known as the silent killer because it is most often diagnosed only after it is stage three or stage four. Patients with early-stage disease do not typically have symptoms from the cancer, and there are no effective screening tests. There's no ovarian cancer equivalent to the mammogram or to the PSA test. Patients who are fit for it receive aggressive treatment with a combination of cytoreductive surgery and multidrug chemotherapy, typically with paclitaxel and carboplatin. Remission after initial surgery and chemotherapy is quite common, but it is rare for patients to be cured of their disease. Relapse is also very common. You can see this Kaplan-Meier curve on the lower right from a relatively recent clinical trial, and you can see the progression-free survival of patients. Again, while remission is common, most patients have relapsed by two years, and over 90% of patients have relapsed by four years. Due to its high recurrence rate, ovarian cancer is the most lethal of the gynecologic cancers. Because recurrence is so common, many investigators have studied maintenance therapies as means to prolong remission after initial treatment. Bevacizumab, a monoclonal antibody targeting VEGF, is approved for use as a maintenance therapy. It's actually given both concurrently with chemotherapy and as maintenance therapy after chemotherapy is completed. It shows, as you can see in the upper left Kaplan-Meier curve, an improvement in progression-free survival. However, as you can see in the upper right, there is no demonstrable benefit in overall survival when bevacizumab is used with first-line treatment. Another approach that investigators have taken is to test PARP inhibitors, and these have shown a great deal of promise as first-line maintenance treatments. However, benefit from PARP inhibitors is not distributed evenly among all patients with ovarian cancer. Patients with a BRCA mutation, as shown in the bottom left Kaplan-Meier curve, have a dramatic benefit from PARP inhibitor therapy in the maintenance setting. However, I apologize that the size of the font is very small in the table, but the table is just to show that patients who have no BRCA mutation derive less benefit. We can classify patients as having homologous recombination deficiency. About half of ovarian cancer patients have homologous recombination deficiency, and these patients derive a moderate degree of benefit from PARP inhibition, whereas patients without homologous recombination deficiency derive marginal benefit, perhaps two-three months compared to placebo. While the maintenance treatments represent improvements, the improvements are incremental. They're primarily improvements in progression free, but not overall survival, with the possible exception of BRCA-mutated patients with PARP inhibitors. One of the things that we'd like to do in ovarian cancer is detect residual or recurrent disease before it becomes clinically evident. As I mentioned earlier, once ovarian cancer is clinically evident, the disease burden is very high. CA-125, or the secreted portion of MUC16, has long been used as a tumor marker for ovarian cancer. Once the CA-125 increases to above two times the upper limit of normal, we consider the disease to have recurred biochemically. However, this typically precedes clinical recurrence or symptomatic recurrence by about three-six months. A more modern way to detect residual or recurrent disease is through the use of circulating tumor DNA or ctDNA. This can be detected in plasma from patients by either next-generation sequencing or by digital droplet polymerase chain reaction. In some patients, this is more sensitive than CA-125 for detecting recurrent or residual ovarian cancer. However, ctDNA has not yet made it to prime time as a replacement for CA-125. What's been shown is that if we detect recurrent ovarian cancer earlier, we will treat patients earlier, but we don't necessarily see an improvement in survival. This is shown in a randomized clinical trial in which patients underwent serial CA-125 draws while they were in remission after first-line treatment. However, the results of those blood tests were not revealed to the patient or their doctors until they went to twice upper limit of normal. At that point, there was a randomization, and for half of the patients, the result was revealed to the patient and the provider, and they would start chemotherapy immediately when they had recurred biochemically. In the other half of patients, the elevated CA-125 was not revealed to either the patient or the doctor, and so they did not start chemotherapy until they developed symptoms or physical exam findings suggestive of recurrence. As you can see in the top Kaplan-Meier curve, the patients who had the CA-125 result revealed started chemotherapy an average of 4.8 months before the patients who did not have the CA-125 revealed. However, the bottom Kaplan-Meier curve shows that survival was nearly identical. We generally don't think that we need to start cytotoxic chemotherapy in an asymptomatic patient with an elevated CA-125. We typically wait until either the patient develops symptoms or we can see rapid radiographic progression. However, since it is standard practice to monitor the CA-125, we think that when patients present with an elevated CA-125, this may be a window of opportunity for disease treatment, including the use of vaccines and other novel agents. With this in mind, we designed the TG4050.01 study to look at the use of the TG4050 vaccine in patients with early recurrence of ovarian cancer. Patients would undergo their standard surgery and postoperative chemotherapy. This would result in remission for about 80% of patients, then we would recruit patients while they were in clinical remission. At that time, the patient would submit archival tumor and fresh blood for development of the TG4050 personalized vaccine. Then we would simply wait until the patient developed an asymptomatic relapse with either an elevated CA-125 or small lesions on a CT scan or both. If patients progressed rapidly and had clinical relapse, meaning symptoms, then they were not eligible for the vaccine and would go on to receive chemotherapy. Those patients who had asymptomatic relapse were treated with the vaccine. Here you can see the results in the first five patients treated on the study. We had one patient number four, who had both radiographic lesions and an elevated CA-125 at the time she enrolled, and she was stable for 11.4 months after the first injection before she gradually progressed. Another treated patient one was treated after her CA-125 became elevated, and she actually experienced a normalization of CA-125 and did not have clinical progression for nine months on study. At that time, she was still in remission, but unfortunately died of an unrelated chronic illness. We'll look a little more closely at that first patient. This patient was a 73-year-old who had been diagnosed with stage 3C high-grade serous carcinoma, who did not have a BRCA mutation but had an elevated homologous recombination deficiency score and a p53 mutation. Those are nearly ubiquitous for serous uterine-ovarian cancer. Her first-line treatment was paclitaxel and carboplatin for six cycles, which is the standard. She opted not to receive maintenance PARP inhibitor therapy nor bevacizumab. She was enrolled after her CA-125 had reached twice the upper limit of normal, and she had a confirmatory CA-125 that also was twice upper limit of normal. Her CT scan was mildly abnormal with some small lesions that were not meeting the criteria for measurable disease, and she was not symptomatic. As such, we treated her with weekly TG4050 for six weeks, and in that six weeks her CA-125 fell sharply, although not quite under the threshold of normal. Interestingly, her circulating tumor DNA rose during the vaccine induction period. After vaccine induction, she went on to receive every thre-week maintenance vaccine doses, and after the first of those, her CA-125 had normalized and her CT scan findings had also improved. ctDNA lagged behind but eventually went into the normal range. This wasn't the only threat to the patient though. She also had a history of hypertension and some episodic atrial fibrillation, aortic stenosis, and she ultimately died of flash pulmonary edema unrelated to the vaccine when she was in remission and had not had any side effects. To summarize, asymptomatic patients with evidence of recurrence by CA-125 or minor imaging findings do not benefit from immediate initiation of cytotoxic chemotherapy, but this represents a window of opportunity in which we can treat ovarian cancer patients with a personalized vaccine. The manufacture of the personalized vaccine is feasible and requires about four months from the time of study enrollment until we can treat the patient. While the number of patients that we've treated thus far is quite limited, the early results are promising and suggest that the TG4050 personalized vaccine as a monotherapy can delay or even reverse early findings of recurrent ovarian cancer. I'll stop and thank you for your attention. Thank you very much, Matthew, for your excellent presentation. Now before opening the session to question, I just wanted to just to conclude this those presentation on TG-4050, saying that, of course, every day we are accumulating new data and the new data will be the subject for communication in H1 2023 in large congress. They are very important for us because they help us to move forward and design what we are going to do in phase II for TG4050. Having said that, I would be very happy with, of course, our speakers to answer any question you may have, which may come from, of course, the audience, but from internet as well. I think Lucie will be the person who will provide us with the question coming from the web. Yep. Please use the mic so that people in the webcast can hear. Sorry. that have been rendered harmless, which allows us to make therapeutic vaccines and oncolytic viruses. I am wondering whether there's a running hypothesis on why this therapeutic modality actually works in anogenital cancers, but not in head and neck cancers, for example, that are HPV16 positive. Is there a running hypothesis? It doesn't have to do with the structure of the cancer or maybe the local immune environment? My second question is on TG4050 regarding the neoantigen vaccine or the vaccine design. You mentioned that you can include up to 30 antigens per vaccine and that you identified approximately 120 targets. I was wondering whether those targets included also shared antigens and whether you saw responses against these shared antigens between different patients. My final question is regarding TG4050 in ovarian cancer. Does an asymptomatic relapse always result in a clinical relapse, or are there also cases in which that does not occur? Thank you. Thank you for your question. Maybe I will answer the first question on TG4001, and Jean-Pierre can of course complete my answer. Actually, the reason why anogenital cancer were selected and no longer head and neck cancer is not related to an efficiency of the vaccine. It's very simple. Initially, we got head and neck cancer in the first part of the recruitment, and we got a response. Pembrolizumab got approval in the first-line treatment of head and neck cancer in combo with chemotherapy or less often used in monotherapy. For that reason, we had no longer any patients because one of selection criterion is not having received any checkpoint blockers. It's not related to any issue with the mechanism of action. Are you? Do you want to add something? No. Yes, I'm good with your answer. It's only a question of opportunity. In U.S. and Europe, PD-1 antibodies are registered in head and neck cancer, and there is for a reason I will not comment today, we still don't have any approval for patient with cervical and perineal cancers. Unfortunately, it provides patient for clinical trial. Can you address the next question? Yeah. I'll take the second question on the antigens that were identified. In fact, when you look at the total patients that were sequenced and for which we've done the exercise, we have identified quite a few antigens, and we have a pretty good picture of what's going on in those patients. Obviously, we do find that some genes are more mutated than others. For instance, I mean, a majority of our head and neck patients as well as ovarian patients are mutated, for instance, for TP53. When you look at the two or three genes, they basically mutated in a majority of patients. Nevertheless, these recurrences in a mutated gene doesn't mean that the mutation is the same. When we look at like more than 100 patients for which we've done the exercise, we did find the same mutations only in two patients or perhaps three patients that were like recurrent. I wouldn't say that we've been able to identify a shared pattern or the shared mutation that we could use from one patient to the other. This is only about the identification. It doesn't mean that those mutations are particularly efficient to a stimulation by the vaccine. Does that answer the question? Yes. The third part. The third question is related to the. I think there is a third part. Yes. I believe this question related to ovarian cancer. Maybe Matthew, can you answer this question? Did you hear the question? Yes, I'm not able to see or hear. Sorry. You did not hear the question. The question was related to the fact if there is an asymptomatic relapse, it is necessarily followed by a clinical full-blown relapse or not? That's the question. In the vast majority of cases, it is followed by a clinical or full-blown relapse, but the timing can vary widely. In some patients, there is a low amount of CA-125 produced, and the patient may relapse clinically even without an elevated CA-125. That's fairly uncommon. The most common is for there to be a gap of three-six months between the biochemical relapse and the symptomatic relapse. Sometimes it can take up to 12 months. What we saw in that first patient on the clinical trial where the CA-125 increased and then went back down and there were imaging findings to confirm is practically unheard of. We would not expect a patient to have a relapse and then eradicate it on her own, so to speak. Coming from Internet, we have two questions which are also related to ovarian cancer. The first one is why doesn't the chemotherapy in asymptomatic patients with CA-125 lead to clinical benefit? This is the result of the trial you showed, but maybe you can comment on that. Right. Why would the vaccine do better there? That, that's a great question. Chemotherapy for patients with biochemical recurrence can be very useful, but I think the importance of that trial is that the timing does not matter. In other words, if we delay chemotherapy until the patient is symptomatic, then the patient is exposed to less chemotherapy, has fewer cytopenias and other hematologic deficits from the chemotherapy, and has fewer other side effects. There is benefit to delay, but there is still benefit from chemotherapy unquestioned. I think the advantage of the vaccine is that it does not cause the same types of toxicities that chemotherapy does. I kind of tell patients that there is a limit to the amount of chemotherapy a patient can receive over a period of time. That limit is different in different patients, but we often reach a point in which we have to start reducing the doses of chemotherapy because of hematologic toxicity. If we give too much chemotherapy too soon, the chemo is still not curative, and the patient will ultimately need more, and sometimes she's unfit to receive additional chemotherapy when she needs it more. By contrast, the vaccine doesn't cause any cytopenias, and may, by working by a different mechanism, may eradicate some of the more slowly dividing cells that are not sensitive to chemotherapy. Thank you, Matthew. There is still another question for you. It's a different phrasing of what you said before, but how many months before symptomatic relapse would be considered as clinically meaningful for TG4050? You said that there is it could be three to even 12 months between the first sign of asymptomatic relapse and the full-blown clinical relapse. What would you consider as significant or clinically meaningful for a vaccine? Yeah. That's a tough question. What the FDA has considered clinically meaningful for, say, bevacizumab and PARP inhibitors is a benefit of about six months. Going from three to six to maybe nine to 12 months of a delay would, I think, be something that might lead to regulatory approval and so forth. Of course, what we want is what that first patient achieved, which was a remission, which who knows, that could have gone for longer had she not succumbed to other illness. Thank you, Matthew. Maybe a question for Jean-Pierre. What are key takeaway from ESMO 2022 that are relevant for the Transgene pipeline? I identify and I mentioned the adjuvant trial of pembrolizumab in head and neck, the KEYNOTE-412, which were negative and we were eager to get the results because it could have changed the therapeutic landscape of the adjuvant treatment of head and neck, but it was not the case. Maybe you have noticed other important news from ESMO which may have some impact. That's a difficult question. Yes. Probably I would say first as a whole oncologist because I've finished my training few years ago. I would say that if you don't reach treating metastatic disease, the level of 50% of response rate, we were discussing few. Mm-hmm ... few hours ago. You will never deeply modify the natural history of the disease. There is no example in the field of oncology, in my memory, where a drug, whatever it coming from, chemotherapy, targeted therapy or immunotherapy that deeply modified the natural history of a disease used at the adjuvant time. When you did not show on the metastatic disease that it's clearly modifying the destiny of patient, and of more than 50% of patients. I mean, it need to works very efficiently at the metastatic level to be one day efficient, when you use it at with a prophylactic strategy. It has been the case in the field of immunotherapy, when we use it for patients treated for melanoma or lung cancer or bladder cancer as an example. It has tried to be effective in the other subgroup of cancer patients. That's why, again, we do believe that what we do at the level of the beginning of a proof of concept, being able to modify what's going on in the metastatic disease in head and neck cancer patients, it is a very crucial step. If we succeed in showing that we do something good for 50% of patients, then I guess as a strategy, with HPV vaccine, as an example, we'll have to switch very quickly in an adjuvant setting. As far as the ovarian cancer is concerned, Matthew, I don't think you were attending ESMO, but there was an update on the PARP trial on PARP inhibitor in BRCA-mutated patients and the confirmation of the effect of PARP inhibitor, but in this specific subpopulation around 20% of the ovarian cancer population. Do you agree with that, Matthew? Yes. About 20%-25%, depending on the study, of patients with ovarian cancer have BRCA mutations. These patients have always been known to derive higher benefit from PARP inhibitors than patients without a BRCA mutation. There are some recent data suggesting an overall survival benefit in that 20% of patients. I don't think we know yet whether we are curing patients or simply delaying recurrence, but the delay is fairly profound. It's a benefit of years for those patients. We're encouraged by that. But obviously many patients, even with BRCA-mutated patients who are being treated with PARP inhibitors still develop recurrence. It's encouraging but doesn't obviate the need for new therapies. Thank you, Matthew. Yes. August Moran, Sabrina Garnier. I have a few, I would say one general question. First on design of TG4001. Just curious if you ever considered looking into IL-2v, use of IL-2v to avoid Treg stimulation as we've basically seen it, this approach being quite broadly used in the industry now. Actually it's a legacy design with an IL-2, which is not mutated and binding both to the dimeric and trimeric receptor, potentially. Nevertheless, it's given in the vaccine, which is non-replicative, so it's only expressed at the site of priming or close to the site of priming. It's very unlikely that it would result into a larger Treg stimulations and that there is a specific need to use a mutated IL-2. This is one aspect. The second aspect is that historically, both in TG4001 and the other past vaccine design that use the same IL-2, we have monitored Treg prevalence and both in the circulation and sometimes in the tissue, and we could not really see any form of increase after relatively intensive vaccination schedule. Yeah, it's a question. We don't have a definitive answer, but it doesn't look like it's being an issue here. Thank you. Question for myvac® and TG4050. Basically for myvac®, just curious if you could sort of highlight again how do you account for clonality of mutations within the AI algorithm? Ideally, the selection should be based on basically single-cell sequencing. This would be a gold standard where you know exactly what's the prevalence of the mutation you put into the vaccine. This is not something that we can do realistically, clinically. We can do it on a, like, experimental use, but it's not something that we can develop, at least. Today, for patient use. Still it's something we use to benchmark ourselves somehow experimentally. Now, in the current design and the envisioned design of the product for the routine practice, if I may say, what we do is that we reconstruct the phylogeny of the tumor by trying to understand which are the clonal mutations, meaning the mutations that we would find into basically all the clones, because it has been shown that those mutations are more efficient for vaccine design. Basically, what we do is that we give an advantage, or the system gives an advantage to those mutations when they are selected. We do that essentially based on the allelic frequency. It's a bit technical. On the clinical data, just curious if you basically, if you look, in the totality of the T cells and how many of those actually are, specific for neoantigens that encoded in the vaccine, and if you looked at, markers of T-cell exhaustion as well. Yeah, the answer is yes and yes, in short. We have looked at every single, basically mutation and how much T-cell response we got against those mutations. We have also made sure those mutations are only directed against the mutated isoform of the protein and not a wild type one or a closely related sequence. This is one thing, and the second thing is that we have characterized not specifically exhaustion markers, but many markers that are associated to T cells. What we could see is that those T cells have an effector phenotype. They do not expressing specific exhaustion markers, at least at the time where we have made the test. Obviously, we have very preliminary data. We'll come up with more. We'll come up with more characterization of the response and more information on the actual activation stages of those T cells. So far, so good, I'd say. Thank you. Thank you for this very interesting presentation. I have several questions. The first one is to Jean-Pierre. Basically, in the TG-4001 trial, you have a large number of patients that seem to show at least stabilization. Then some of them are relapsing. I was just wondering whether you had the chance to access tumor biopsies so that you would. Because this would be a perfect setting actually to demonstrate at least some immune response and possibly some tumor escape. Yes. That we have an academic second-tier biopsy programs. I'm not sure that the patient included in that trial and those who are relapsing after a long period of stable disease have accepted a new biopsy in the trial. I have to double-check that point, but I'm not sure. I know that we have also an academic program concerning PBMCs, and we're gonna also be able to see in the peripheral setting what's going on in terms of exhaustion of T lymphocyte and those who are specific for the antigen E6 and E7. I have no answer to that question, but it's clearly one of the question we're gonna address in the future because the collaboration between Transgene and the academic teams, like of Christian and the team of Maha Ayyoub is very efficient. The going back between the teams, that probably one of the reason why we are like collaborating with Transgene very much and for so long time. Since you analyzed the T cell responses and do you have in some way any correlation between the quality or the intensity of the T cell responses and the clinical responses? I guess I can start with the answer, but Cédric Bani-Jama will probably follow me. We're still expecting the results that are looking at, as an example, the complete phenotype of CD8, the tetramers, the complete sequence of the TCR of lymphocyte that Christian Ottensmeier will do in the future. This remaining question are a burning question, and we would love to share the results with you today. I guess this will be early next year, because at the biological level, that will be key answers in order to know what's going on. Before getting enough time to see what's going on in patients, as you- As it has been mentioned, there was three relapse in the control group and no relapse in the treated group. Is it a signal? I don't know today, but to be honest, it's not largest panel of patient to be today very confident and send you a signal. I would really love to share with you today what's going on at the level of CD8, which are specific of each antigens in all the trials, HPV on one arm and on the other arm, the 40/50. I have a quick question to Christian. May I? Or are we too late, or The last question. Okay, the last question. Basically, you mentioned that with the TG4050 trial, you had evidence that there was not only expansion of preexisting T cells, but also induction of new antigen-specific T cells. Did you have a chance because you mentioned that there was some follow-up with tetramers? Because then following, for instance, typically the quality of preexisting T cells before and after treatment following a tetramer sorting would give also some information about the functional status, the clonality of these cells and perhaps possibly see here the direct impact of the vaccine on the quality of the T cell response. Yeah. Yeah, go ahead, Christian. There'd be three pieces. One is what happens in the blood compared to those patients where we have flash-frozen tumor to do single-cell RNA sequencing in the primary tumor before resection, because that will give us really quite clear evidence of what was there before in the tumor. At the moment, the measurements are what is present in the blood before vaccination and then after vaccination. That already gives us two extra pieces of information. It will allow us to compare baseline reactive T cells transcriptomically to antigen-specific T cells that are present after vaccination also, as well as those T cells that are only present after vaccination. We'll be able to compare whether at a single cell level, the T cells reactive to neoepitopes after vaccination are different to the ones that were before vaccination. I think that will begin addressing the functional questions that have or the questions about the functional capacity and the exhaustion phenotype that have been asked already. I think what we would predict is that the cells, from what we know so far, will turn into a more activated immuno phenotype post-vaccination, and we hope to be able to confirm that at an epitope-specific level also. Yeah. Concerning the longitudinal monitoring with the tetramer staining, indeed for two-thirds of the response, if you take the total of response observed, two-thirds were absent prior to vaccination, and one-third is amplified after vaccination. Obviously, we do not know whether, I mean, it's a bit specific, but we do not know whether it's the same clone, and it's probably not the same clone. That's an answer that will come up with the TCR sequencing of positive tetramer stained cells somewhere next year, probably. Much more to come. That's good news. I would like to thank the speakers for their excellent presentation and their answers to the question from the audience. Very, very interesting question. It's time for a break. Five minutes? Ten minutes? No. A very quick break, because otherwise we'll really run late. A couple of minutes, and then we'll be back for OncoBx. Thank you. Thank you. Thank you, Christian. Thank you, Mathieu. Now in terms of freaking cool mode of action for oncolytic viruses. We'll then jump into examples of what was achieved together with Leeds University Hospitals in understanding the mode of action of the product in humans in several routes of administration. We'll get some questions for Adel at that time point before he will leave, and we will jump back on the normal program. I will be continuing the session, presenting you our recent development and the new product we plan to move into clinic in next year. We will have Steve Bloom, our Chief Business Officer, presenting opportunities for collaboration on this platform. We have heard about the mode of action, the development we can do with tumor vaccines. I just want to position oncolytic viruses. They are very different from the other one. They are also based on viral vectors, but we have selected replicative viruses that would specifically replicate in tumor cells, and that would be the site of action for this class of product. They will engage peritumoral lymph nodes, immune infiltration, and by targeting the tumor, be able to remodel the tumor phenotype. That would be basically the big difference between this technology and the previous one. If I just recall some basic, I'd say we usually consider that oncolytic viruses combine three mode of actions in the tumor environment. They would, by themselves, be able to infect, replicate, and induce a viral-dependent cell lysis. That's the reason for the historical name of oncolytic. We know now that most of their efficacy comes from the engagement of immune cells that will be able to reach the tumor replication site for the virus. Virus replication will act as a signal for engagement of both innate and adaptive immune responses. We also, as said before, use the potential of the vector to directly express in the tumor some payloads that would diffuse from the replication site and be able to either repress some immunosuppressive mechanisms or boost immunoreactive mechanisms. They would also be able to bring in the tumor some new modalities for activity, like what we have in TG6002, that is direct expression in tumor of some enzymes that will locally convert a prodrug into an active cytotoxic drug. We have this kind of targeted chemotherapy on top of the viral-induced cell lysis, immune recruitment, and potential new modality of treatment that would be targeted to the tumor specifically with minimal exposure of the serum and of tumor sites. What is important is also to confirm that we have at the site of viral replication all mechanisms that are necessary for triggering a strong immune induction. We published some time ago results demonstrating that, and confirming that, oncolytic viruses when replicating and inducing cell death, in fact, do match most of the major hallmarks of an immunogenic cell death. That's important. The release of all those signals and the combination of the signals is really instrumental in driving the immune response towards something productive, not only against the viral infection, but against the tumor antigens. What is also important for us is that using this oncolytic platform as a platform for the discovery and the development of new product to be able to demonstrate its versatility. We have a lot of examples internally and as part of our collaboration programs with other companies showing that a large diversity of molecular payloads could be encoded and expressed from the site of replication. This diversity ranging from small immunofactor like cytokines or small effector ligands that could be encoded in viral genomes to very large payloads such as full length antibodies or large enzyme systems that were also expressed at high yields. We have this knowledge on the promoter design or ability to engineer recombinant cassettes to control the level and the timing of this expression in the tumor. I recalled in the very beginning of the lecture the fact that poxviruses are large cargo. It's well documented that they can encode for up to 25 KB of transgene cassette. Keep in mind, for example, a full length antibody would be 1.5 KB. Plenty of room in the cargo for targeted delivery in the tumor. I would give you some kind of flavor on the mechanism of action from this product, BT-001. As the name suggests, it's been developed together with another company called BioInvent in Sweden. Our good friends at discovering and putting in development very original antibody settings. In that case, it's a full length antibody optimized for the Fc gamma part of the receptor to engage the ADCC. The antibody would recognize the CTLA-4 target that is well known to be present in the surface of intratumoral Tregs. We also express the GM-CSF to act as a multifunctional cytokines in the tumor environment. It was demonstrated in previous programs like one with the T-VEC, now called IMLYGIC by Amgen or Pexa-Vec, our own product. It's also the case for LP 1 and LP 2, that GM-CSF would bring a lot in the possibility to activate and mature infiltrate monocytes into macrophages and antigen presenting cells directly in the tumor. It can also promote the skewing towards the immunocompetent M1 phenotype for macrophages. As you see here on the slide, we have optimized the design to reach the maximal functionality of the recombinant antibodies. What is important in this ambition is the good balance between the light and heavy chain. By using in that case two identical promoters, but in some cases might rely on two different promoters. We have been able to obtain the proper stoichiometry and functionality of the antibody directly in the tumor. For the demonstration I will just develop just after. We also have a surrogate product adapted for preclinical research in mice. That's pretty the same design, but the GM-CSF and the antibody would be customized for their mouse target. Just in a nutshell, relevant examples. I think you see here that we've been able to demonstrate that the product is active in syngeneic, so immunocompetent mouse models and a large diversity of these models. Here are displayed five of them that range from so-called hot tumor with very large tumor infiltrates in terms of T cells and even for some of them in terms of NK cells to cold tumors that are desert tumors with virtually no immune cells present ab initio at the time of the treatment. For those models we could demonstrate some activity. Of course, the hotter, the better. But you see here that even in very cold or deserted tumors we could achieve some survival, and it is associated with the tumor influx of T-cells and relevant immune cells. What is important also to demonstrate is that this activity is correlated with induction of strong and long-lasting CD8-positive T-cell response. When analyzing the specificity of the neoantigen-specific T-cells or infiltrating T-cells, it's also very important to demonstrate that we've been able to generate a majority of tumor-specific T-cells compared to anti-vector T-cells. It's about 1/10 of them. It's a question that we often have on, do we have a majority of T-cells raised against the vector or against the tumor antigens? This type of experiments clearly demonstrate that a very large part of the T-cell response is against tumor antigen. We know that this T-cells are mostly memory T-cells, and that could be demonstrated by re-challenge experiments in which for example, we treat mice grafted with one specific model and the surviving mouse would be re-challenged with another tumor that would be immunologically different from the initial one. We test whether there is a difference in terms of behavior upon vaccination between these two groups. What you see here is that the population that was treated with one specific tumor would have developed a tumor specific response that protect the survivor against re-challenge, whereas the mice treated with different tumor would almost all die within one month and a half. What is also important, and also demonstrative of a strong, systemic, T-cell response is, this type of experiments called abscopal response in syngeneic model where we, equip the mouse with two tumors. Only one of them will be, treated with the vaccines, and we look at the evolution of the size of the second one untreated tumor. You see here, upper panel, those tumor will grow fast in the absence of treatment. What is impressive is that, upon treatment on one side, you see that most of the mice, would also present control of the tumor growth on the non-injected flank. Only two of the mice did escape from this, treatment. That's demonstrative really of induction of immune response at the site of administrations, but a systemic response that could reach non-injected site of tumor. That's the rationale also for us to move into metastatic stage diseases, starting from either a local treatment or from intravenous administrations, hoping to reach one application site, but for which the answer could translate into action in non-treated tumors. When analyzing in depth the phenotype of the T-cell responses, that's the right-hand panel, you see that we have really been able in this injection at the priming site to induce a large expansion of the population of effector T cells. We have been able to decrease the proportion of exhausted T cells. That's true in the tumor but also in the circulation, and also to be able to demonstrate a strong decrease in the population of intra-tumoral Tregs. It's also partly true, but only partly true in non-injected lesions, but a significant amount of the Tregs are also depleted in this non-injected lesion. When further analyzing what are the pathways induced beside the induction or decrease of exhausted T cells, we can go in what is happening for other cell populations. By using differential gene expression analysis, we could confirm that a large component in this immune remodeling of the tumor microenvironment is associated with activation or expansion pathways of a large diversity of cells relevant for expansion of the T-cell clone repertoire, but also for mobilizing other type of innate responses. We could also demonstrate specifically that among the population that are induced, we have this cDC1 subpopulation of APCs that have been really boosted by the treatment. That's also very important in demonstrating the ability of the product to also favor antigen display and expand the T-cell repertoire. All those results plus, of course, many other results we've been able to collect in the last two years together with our colleagues at BioInvent, led us to the positioning of this product in the IT route as a start to demonstrate that, for example, superficial lesions treated with BT-001 can of course be controlled in their evolution and resolved spontaneously after induction of the virus. Also we will monitor activity on non-injected lesions and want to be able to demonstrate that we've been able to generate long-lasting circulating response that would act also on uninjected cells. The first part of the trial has been almost completed. We are enrolling the very last patient at the IV dosing plan in this dose escalation program. The next step, of course, that will be very relevant is combination with pembrolizumab. We already have the agreement from MSD for free supply of the Keytruda, and we should be starting this Part 1B of the trial at the beginning of next year. All those patients are, of course, carefully monitored for a long period of time. We know from our competitors that some responses might take some time to be detectable several months after injection. Today, of course, short-term monitoring has been performed, but we want to keep those patients under monitoring to be sure that we have a full control of what could happen in the months after our treatment. We could, from some of those patients, demonstrate that both the antibody and GM-CSF are expressed in the tumor. Of course, it's a safety study also. So far, no reports of adverse events. The product is well-tolerated. We have no spreading of the viral genome from the tumor into blood or biological fluids at the moment. Confirming, if needed, the high tumor specificity. We are, as I said, now entering into the IV dosing, where we expect a larger fraction of the patient to respond. For the low doses, we have been able to observe already one tumor control in one patient at the lowest dose level, which is an achievement, but, of course, very few numbers at that stage. Something important in our development of this in vivo platform is our collaboration with AstraZeneca. You might remember that we signed this collaboration deal three years ago. It is, it's a very large collaboration with that could lead up to five new products in the framework of this collaboration. We, of course, have had the challenge to develop five products, some being simple, although very complex. Really, this collaboration has pushed our technology to the limit. So far so good, I would say, in the sense that most of the preclinical programs have been performed in-house, and most of the products have been transferred to AZ for in vivo analysis. That's their duty to characterize the product in their challenging model in their premises. The good point is that for the first product delivered to AZ, we already have one option license exercised last year. We know we have worked together to make this product become a clinical development. We have got some good feedback from AZ confirming that they really want to move into clinics as early as beginning of next year. If I... Just a few words on what we are actually doing internally in terms of research. We'll have report from Adel. I've reported on BT-001 in the IT route. Really the key challenge for us is to be able to demonstrate that the product would be efficacious in the IV route to target advanced stage metastatic diseases. There are many limitations that have been reported to use oncolytic viruses in the IV route. Most of them are reported in this cartoon that was published some time ago by Marchini. There are many papers highlighting the limitations for using OVs in the IV route, and we are addressing that in the research. Here are the different avenues our teams are exploring. You see that, if I start from the below part, in terms of preventing neutralization by circulating antibodies, preventing capture by endothelial cells or whatever, non-specific interactions we could have in the circulation. One way would be to of course reformulate the product or change the surface properties of the virus. We have tried via collaboration with both companies or academia several technologies to be very direct. None of them have been so far demonstrating good efficacy. So most of our attempts to change the surface properties translated into loss of efficacy and loss of infectivity of the product. But we will look for new technologies if available. What is maybe more relevant is to be able to force the transfer from the blood to the tumor by local physical stimulation. You might have seen the paper we published some time ago, using ultrasound cavitation to transiently permeate the vasculature and favor the transfer into the tumor. The technologies developed by OxSonics is still prototypic, but we monitor their progress and hope that we will have something ready for clinical development in the next forthcoming months. Unfortunately, the available echographic machines are not suitable for this development at the moment. Another route is to boost by a combination with other agent, the replication or the ability from the very early infected sites to have a strong expansion of the viral infected cells. One maybe major contribution could come from the combination with nano formulated mRNA. We have this ongoing collaboration with small companies based in Boston called Combined Therapeutics. We are currently exploring several candidates for this optimization of the replication from the very early infected cells. Too early to report on the results, but we hope that in the very next months we'll be able to communicate at congresses what we have achieved so far. Maybe the more potential would come from the left-hand panel part of this figure. We have a lot of internal projects demonstrating that by changing and acting on the arming of the Copenhagen platform, we could change the immune phenotype, decrease some subset of populations, the level of immunosuppressive cells, boost the proportion of immune effector cells. Ongoing programs we might report also soon on those candidates. What is also another venue we are exploring is the use of targeted inhibitors for some immunosuppressive pathways like the interferon gamma intratumoral response. We might—you might have seen this collaboration we have with Medesis, who has provided a technology based on nanoemulsions for targeted delivery of inhibitory RNA to target precisely this pathway. Something also important is to demonstrate that the oncolytic viruses could combine with cutting-edge technologies used now in the treatment of solid tumors like CAR T-cells. We have this very efficacious collaboration with this Chinese companies called PersonGen, developing their own CAR T-cells for solid tumors, and who believes, as we do, that co-treatment with OVI equipped with a specific attractant for T-cells makes sense there. We are of course hoping to also very soon publish those results. All those development of course aims at selecting one product. You all know that research might be disappointing and a risky game, but these are the kind of two-pronged plan of action we have taken to be in a position to be the first having onco-viral approaches targeting the IV route. If I come back to the clinical portfolio, Transgene's. You see that we have our development around TG-6002, and I will leave the floor to Adel to comment on the very first results we have obtained with 602 in the IV and IT route. We also have this collaborative program with BioInvent on BT001. We also expect a lot, of course, in the very forthcoming months, around the development of a product together with AZ. Maybe to come later on, some development from our own research programs. I will at the end of the day report on our IL-12 expressing oncolytic virus. Of course, the best is yet to come with the other companies. Having said that, I will leave the floor to Adel, who is working at the Leeds University Hospital, acting as a medical oncologist and the lead of translational research team. Very strong and long-lasting collaboration with us and very happy to hear you, Adel, sharing your observation on this very important product for us in terms of lessons learned from the development. Thank you very much for the introduction, for the kind invitation to visit Transgene. The products that we have been working on are TG6002. Let me just see if I can figure out how to use this. Yes. Oh, no. Other way. The green one. The green one. No, the green arrow. The green arrow. The big one. Thank you. Thank you. I just wanted to introduce the team in Leeds. Running a translational early phase clinical trial requires a really very large team. The lead of my translational laboratory, Dr. Emma West, is sitting there in the audience. I have a large laboratory group, which requires the input of interventional radiologists and standard radiologists, a large early phase clinical trials team, as well as the surgeons. In Leeds, we are lucky that we can do this very efficiently, and we have a very efficient pipeline of oncolytic virus studies. In which types of GI cancers could oncolytic viruses make an impact? When I'm considering GI cancers that oncolytic viruses could make an impact, what comes into my mind is which ones are more likely to be immunosuppressive. For TG-6002, which encodes enzymes that convert inactive 5-FC chemotherapeutic into 5-FU, active 5-FU, and considering which ones are 5-FU sensitive. The cancers that come to mind are colorectal cancer, hepatocellular carcinoma, and cholangiocarcinoma, both in the neoadjuvant setting as well as in the second-line palliative setting. Then another consideration is the route of administration. We know that you can deliver an oncolytic virus into tumor via transdermal intratumoral injection, that you can definitely get the virus there. How far does the virus spread throughout the tumor, and how well-tolerated is this procedure? Intravenous therapy ensures that the virus, if it reaches the tumor, is well dispersed throughout the tumor, and the procedure can be repeated very easily in the setting of systemic therapies for patients with cancer. However, we don't know just how much virus gets into the tumor. A third way of delivering virus is through local regional delivery. These methods include intrahepatic artery infusion, isolated limb perfusion, or the peritoneal pleural routes. Today I'm going to talk to you about intravenous delivery and intrahepatic artery delivery. The first study that we undertook with Transgene was with Pexa-Vec. This is the preoperative Pexa-Vec study where we injected a standard dose of Pexa-Vec, Pexa-Vec being a modified Wyeth strain vaccinia that expresses GM-CSF. We injected a single standard dose ahead of planned surgical resection of colorectal cancer liver metastases. The primary objective being the presence of virus in the tumor. When you inject intravenously, how much virus gets into the tumor? Secondary objectives are safety and changes in the immune tumor microenvironment. Patients received a single intravenous dose ahead of planned surgical resection, approximately 14-21 days afterwards. What we found was that there were changes in the peripheral blood parameters, as you can see there in the bottom right-hand box, with activation of the immune system, increased PDL-1 expression as a result of virus delivery. We found that in two of the nine patients treated that there was tumor necrosis. In one patient, there was a complete pathological response following a single intravenous infusion of Pexa-Vec. In another patient, significant pathological response. The patients with complete pathological response presumably didn't need to have major hepatic surgery, but you don't know that until you take out the liver. What we also found was that there was an increase in neutralizing antibodies, as you would expect, and also that the virus was carried in the plasma component of the blood and not carried on the cells. We found the presence of virus in the majority of tumors that we treated, and this was at variable doses within the tumor. Our in vitro assays showed that the virus only replicates in the tumor and does not replicate in the normal liver. We also found the presence of large amounts of CD8 T-cells in the resected tumor specimens in all the patients that we treated. We showed that there was a long-lasting adaptive immune response against the cancer. We showed by way of an interferon gamma ELISpot that we get increased numbers of T-cells that are producing interferon gamma in response to a tumor-associated antigen, carcinoembryonic antigen, and that this continued at least three months following surgery. That single infusion of Pexa-Vec resulted in a marked T-cell response against cancer that continued well beyond the time of surgery. Just to summarize that, intravenous Pexa-Vec reaches metastatic tumors. There is an interferon response that induces innate and adaptive anticancer immunity. There's evidence for tumor necrosis, and there's the upregulation of immune checkpoints. However, there was a little hypotension in some patients, and we wanted to see if we could optimize the delivery of the virus to the liver tumors potentially via intrahepatic artery infusion. Moving on to TG-6002. This is the next development of the vaccinia virus. It's a Copenhagen strain vaccinia that encodes enzymes which convert 5-FC into 5-FU. It's being tested in two simultaneous clinical trials, one via intravenous delivery and one via intrahepatic artery delivery in patients with colorectal cancer. For the trial via intrahepatic artery delivery, it's patients with liver-dominant colorectal metastases. For the intravenous study, patients are randomized to one of two arms. In the first arm, patients receive three infusions on days one, eight and 15. In the second arm, they also receive three infusions but on days one, three, and five, and the doses are escalating up to 3 × 10^9 PFU. In the intravenous study, we found that TG-6002 replicates in tumor tissue without signs of further spread into normal tissues, that the virus does reach the tumors, and this is found by tumor biopsy, especially at the higher doses of TG-6002. In the patients administered 3 × 10^9 PFU, we found the presence of virus in the majority of patients. We also found the presence of 5-FU in the tumors, indicating that the virus is replicating, producing the enzymes that convert 5-FC to 5-FU, and the 5-FC is given orally. We also found that the 5-FU was sustained in those patients up to at least two weeks, and at times longer, following the virus administration. There was, again, an onset of neutralizing antibodies, as you would expect, but this did not seem to inhibit the activity of the virus. The peak serum 5-FU was maintained despite the levels of neutralizing antibodies. This was the findings in the intravenous study. The clinical results continue to be collated for that and will be released soon. Moving on to the intrahepatic artery study. The intrahepatic artery route is routinely used in patients with liver tumors and it is frequently used for patients with primary liver tumors but can also be used for patients with liver metastases. The idea here is that it ensures a greater dose of the drug is delivered directly to the tumor where it is required. In this study, we used the same virus. We used it in patients with liver-dominant colorectal metastases. We gave a single infusion on day 1, followed by 10 days of oral 5-FC, and if everything goes well, the patient has a CT scan after four weeks with the potential for a second cycle of the same treatment. This is the team here in Leeds administering the virus to one of our patients. It's done in the interventional radiology theaters, and you can see angiography of the hepatic artery there, showing perfusion throughout the liver. We treated 15 patients to date, starting dose of 1 × 10 to the 6 PFU, and escalated to 1 × 10 to the 9 PFU. We found that the treatment is again safe, that no maximum tolerated dose was reached. According to the protocol, that was as far as the study was planned to proceed. In terms of laboratory immunological activity, what we found was that in the highest dose cohort, so cohort 4, that we saw an increase in CD4 and CD8 T cell infiltration into the tumor, and this was much greater than we saw in the lower dose cohorts. We also saw an increase in PD-L1 expression in the tumor in cohort 4, but not so with the lower dose cohorts. Again, this shows 1 × 10^9 PFU is sufficient to draw immune cell infiltration into the tumor, and that there is immune activity leading to upregulation of immune checkpoints. These graphs are probably a little too small, but essentially it shows in the first three cohorts that we are getting immune activation. We are getting the expression of costimulatory immune checkpoints such as OX40, as well as co-inhibitory immune checkpoints such as PD-L1. Very nicely, we saw a decrease in CTLA-4 expression. TG6002 leads to a reduction in CTLA-4 expression amongst a cohort of immune cells, indicating either that those cells are less exhausted and more immune active, or that there are less regulatory T cells. Similarly with TIM-3, another co-inhibitory immune checkpoint. We also saw in the... This is data from just the first three cohorts. We are analyzing cohort four with the highest dose, that we do see increased T cell interferon gamma responses to the tumor-associated antigen CEA. We also see a little activity against the virus itself. T cell activity against the virus, which is no bad thing, because if the virus is inside a tumor cell, then elimination of tumor cells harboring virus leads to anticancer immune responses. This was an assay that we did very recently, looking at calreticulin. Calreticulin is a biomarker of immunogenic cell death, and we see little activity in the first three cohorts, but we see a peak of activity in cohort four, indicating 1 × 10^9 PFU induces immunogenic cell death in those patients. Looking for the presence of virus in the intrahepatic artery study, we saw very little virus at 1 × 10^6 PFU, a little bit at 1 × 10^7. The majority of patients being tumor virus positive at 1 × 10^8 PFU, with ongoing analysis at 1 × 10^9, but looking very positive in terms of the amount of virus reaching the tumor. That's the summary of our work with TG6002 in patients with colorectal metastases. Thank you very much for listening. I'll take any questions. Thank you, Adel, for this presentation. As said before, you will have to leave to fly back to the UK. If you have any kind of burning questions or one, two questions before Adel leaves, please do that now. Okay, good. Maybe I'll just. Yeah, the microphone goes to you. Thank you. Just a quick question. I'm just curious if you were able to see the virus spread within the tumor mass or also localized replication of the virus? Yeah, for pre-op Pexa-Vec, where we had the entire surgical sample, we could see specific foci of virus replication two-three weeks after injection. Presumably, when you inject intravenously, you get a little bit of virus that reaches the tumor, and that there are specific areas of replication within the tumor that you can detect up to three weeks after giving that dose. In the TG6002 studies, there was a biopsy that we're taking, and a biopsy will only tell you about the bit that you biopsied. But we were able to find virus at the higher doses in the majority of those biopsied, indicating to us that it's likely to be everywhere in the tumor. Maybe regarding TG4001 and TG6002. With TG6002, we see that there is in the plasma, we see systemic levels. Do we also see that for TG4001? To be very We are of course gathering the samples and we are now reaching the. IL dose where we expect those responses. As of now, no positive answers. Is the thinking then also that you need to have a certain threshold in the past to actually use the- You're right. Many discussions are ongoing. One of them being that we might have also been too high in terms of viral infusions, generating major, I would say fibrosis or whatever, retention mechanisms. We know also that we might have over-boosted the immune response against the vector, so this might be an explanation on one side. The other one is that it takes some time. Time points. I don't have in mind the time points, but also we have to carefully look at the optimal time points for this, sampling. Okay. Thank you so much. Yeah. Do you have an idea of the levels of 5-FU that you reach in the tumor? How do they compare with the standard chemotherapy levels? It's a question for Caius, who's been doing the assets. Do you get the question? Levels of 5-FU in tumor in comparison to systemic delivery of 5-FU. I think I don't have the number off the top of my head, but when we compared, we reach similar level to already administered prodrug of 5-FU. I see that the calreticulin maybe is too few samples, but appears in the higher dose. Does it correlate with the levels of 5-FU that you get? It seems there is a trend of correlation between the dose of 6,002 and the level of 5-FU, but it's hard to say because the number of samples are quite low. Okay. Thank you so much. Nice to see you all again. I'll see you soon, hopefully. Take care. Thank you. Bye. Thank you, Adel. Of course, if there are any kind of questions you'd like to ask, we'll be happy to transmit to Adel or even work internally on an answer on them. Next part of this presentation will be around our brand new product expressing IL-12. It's something we are the first now. It's the first time that we report on this product, so privileged communication. Here is the product design. You might recognize the way we construct our product, combining IL-12 cassette with a full length antibody. Compare with what we did with the BT-001, we have used a different set of promoters for the antibody part. And also, actively worked on optimizing the promoter selection for IL-12. The target product profile for this specific one is to target immunosuppressed solid tumors where immune remodeling of the TME makes sense. We know it's the case in many diseases that have proven to escape from primary lines of treatment, which might have generated insufficient tumor infiltrations or accumulation of T-cells that would be exhausted T-cells or imbalanced population of T-cells compared with the APCs in the tumor. Of course, in this development, we have in mind to fully exploit what we've learned from development of 602, and particularly to go in the IV route to target deep lesions and advanced stage diseases. The tumor-targeted IL-12 therapy has been documented for a long time. The challenge is that the therapeutic window for IL-12 is quite narrow. We need to reach a rather high amount of IL-12 in the tumor, but avoid reaching high level in the circulation. The maximum dosing reported to be 0.5 microgram per kilogram or kind of average C-max around 100 nanograms per ml. So we have those very precise constraints in terms of balance between the intra-tumoral concentration and the circulating concentration. We believe that the oncolytic targeting systems might achieve those very ambitious numbers. What is important also is that the product to be able to be positioned in the modern settings should be combinable with the approved checkpoint inhibitors or most recent chemotherapy. Of course, to be able to demonstrate that, used in the second line, they could synergize with those therapies. The question of having a product expressing both IL-12 and a possible second antibody, of course, is still being characterized in preclinical. We have another product that might also make sense with a single agent. But we believe that really having those two armings that would work synergistically in the tumor makes a lot of sense, especially if the antibody that I cannot disclose today would act on the balance between effector and Treg cells. IL-12 is a kind of a consensus. It's a must have, and of course, we'll discuss that on the next slide. The intensity of the competitions also reveals that it's a must have for in the therapy. IL-12 is a potent pro-inflammatory type 1 cytokine. It has been used in a large number of preclinical demonstrations, but also we have a lot of knowledge from early clinical trials demonstrating its ability to boost both innate and adaptive responses. Acting on multiple cell populations from NK to T cells at large, inducing differentiation of most relevant helper CD4 T cells, and skewing the mechanisms toward Th1 response. It would also indirectly act on tumor cell immunogenicity, boosting expression of MHC class I and class II molecules, but also working on macrophages and immune cells in the tumor. This strong activity is of course well demonstrated, but counterbalanced by the multiple reports on adverse event when using IL-12 directly in the circulation. We need some mechanisms to favor accumulation in tumor and avoid overexposure of other organs. This consensus on IL-12 is translated into a very high competition in the field. Multiple technologies have been explored in this perspective. The use of mRNA technology of recombinant proteins and fusion proteins to achieve proper targeting to the tumor and minimal exposure of non-tumor organs, and obviously also viral vectors. In the field of viral vectors, here are the key competitors that we have facing our development. You can remark easily that most of development at the moment are still in the IT route. Only RIVAL-01 developed by Turnstone and Takeda in North America has the ambition to go both in IV and IT, as we might have. You might have seen recently that there's been a big change in the leadership on these programs. Takeda stopped its development on this product, and we are waiting on what could be the next steps to move forward for this product. I would say large number of competitors with IL-12, but very few willing to address the challenge of going in the IV route. What can we say already in terms of activity of this product candidate? We have, as we did for BT-001, assessed the product efficacy in several models, from CT-26 as a reference model to very cold tumors. The colder one being LLC1, and also very challenging tumor being this breast cancer model, EMT6. We could detect responses in most of those models using monotherapy or combination with anti-PD-1. Again, proving that those two mechanisms are synergistic, but most of the activity in these combinations coming from the oncolytic virus. What was also important is to confirm, as we also did with BT-001, but now we have a larger set of immune characterization of the clonality and the diversity of the T cell responses. You see here that we've been able to characterize at the 10 and the 14s were quite early on in the process. This, the induction of both antitumor and antiviral responses. Mostly we have a recall from memory T cells against the tumor and also fast induction of T cells against viral antigens. We know that those responses are not very long-lasting, except for the tumor responses. Important to demonstrate that we have a strong memory cell mechanisms engaged in the antitumor response. You see here the results we could collect in the CT-26 model, demonstrating that even after a recent challenge, we have a very strong efficacy. Again, confirming that we have been able to raise a memory T cell response, and part of that being a resident memory T cells. What is also important is to demonstrate that a very efficacious response could be observed. If you just look at that after the session, compare this efficacy with the one we could obtain in the IT route. With BT-001, we are much stronger in the ability to raise a memory response here. This key role of CD8-positive T cells could be demonstrated by some depletion experiments. We're using polyclonal antibodies. It can deplete either CD4, CD8, or NK-cell populations specifically. You see that upon depletion of CD8, you have a complete loss of the efficacy. A partial loss for CD4 also demonstrating that they are partly involved in mechanism action, and those CD4 populations being mostly T helper cells rather than effector CD4 T cells. NK cells contributions was also detected, still to be analyzed, but it looks to be important at early stage of the antitumor response, not really on the long-term mode of action. If I expand these conclusions on the contribution of CD8 T cells in terms of benefit for the whole body, this is a typical experiment that we usually carry on, confirming that the CD8 positive T cells are induced and would act on the multiple lesions. We have this double flank xenograft model, where we inject one lesion and see the way the second lesion injected would evolve. You see here, as we did for BT-001, that we have a strong abscopal response, both tumor being controlled, and most of the mice fully controlling the non-injected lesions upon treatment. For this specific product, we could go very deep in terms of analyzing the molecular pathways involved in antitumor response. You see here, it's just very small and almost unreadable, but you see here that the volcano plot reveals a large transcriptomic response after a product treatment. This happened to be true after empty vector injection, but the stronger response came from the IL-12 armed virus, and you see induction of a very, very large number of pathways in both injected and non-injected lesions. When you go deeper in the ontology of the responses, you see that multiple classes of pathways could be triggered and very early on in the process. As soon as D15 and D17, we are with the maximal transcriptomic response. That will be very long-lasting, but in the very first days after administrations, we have multiple immune cells being activated or infiltrating the tumors. When you look at those subpopulations more accurately, comparing the empty vector and the IL-12 armed virus, what we basically could easily see here is that we have a strong infiltration of the tumor by CD8 positive T cells, but also neutrophils, DCs, and Tregs. This massive infiltration of Tregs is also the rationale for maybe a second arming we could put in the virus. We are of course continuing characterizing the activity in preclinical models. All those tools that we are developing will also be made available for the clinical development because these are the key questions we also have to document in human. If I summarize where we stand now with the development of IL-12 expressing VV, we have this potent novel candidate being characterized in preclinical models. We have been able also to take into account some lessons from the development of TG6002 in improving the potency of the vector. For those of you who are specialists in vector technologies, you might have detected that we have slightly different vector than the one we were using for our previous development. This one would have an additional deletion for the gene M2L that we reported two years ago to be a viral component interacting with CD80 and CD86 that suddenly is beneficial to be removed from the viral design. We have been able to optimize the promoter sequence for both IL-12 and the recombinant mAb to be implemented in the product. In the meantime, we also improved the GMP manufacturing processes to be able to control in the GMP product the level of residual IL-12 that would be administered to the patient. The tumor specificity should be very strong, and we will be, of course, cautious to confirm that we have tumor-only replication for the product in the dose escalation program to be planned. We will of course use the transcriptomic signature to analyze from the biopsies the mode of action of the product in the tumor. We know already that even the triple deleted vector is well-tolerated. Primate experiments were very reassuring regarding the safety of the vector. We are, of course, still considering running additional pivotal studies if needed to manage the potential overexpression of IL-12. With the GMP processes we have now in place, I think we are on the safe side for this question. The discussion is very active on what could be the best indications for developing this product. Lung cancer has been to date a champion in this discussion, considering the medical need, the characteristic of the patient escaping from the first-line treatment and the availability of those patients for further development. Thank you for your attention. We'll take the questions after Steve presentation on the opportunities we wish to reinforce on this Invir.IO platform. IL-12, of course, being developed by Transgene, but we have not yet disclosed all other options we have engineered in the lab, and that could be venues for partnerships. Also we are ready to consider kind of customized development programs with companies as we did with AZ. Steve? Thank you, Éric, for that excellent overview of the Invir.IO platform, and good morning and good afternoon. I'm Steven Bloom, Chief Business Officer for Transgene, and I'm gonna build on what Éric said and quickly go over why the best is yet to come with the Invir.IO platform. The Invir.IO platform is a pretty versatile approach to creating oncolytic viruses. There's four real advantages to the platform, and clearly one is safety and selectivity. The second is the ability to you know create large payloads and put multiple payloads onto the virus itself. We also have the ability to take a look at various solid tumor targets, and we can actually think about the size of the market relative to solid tumors and I'll talk about that in a few minutes. We have also the ability to make the product obviously. When you think about the versatility of the platform, it gives us a competitive advantage in our ability to develop oncolytic viruses going forward. Oncolytic viruses have had an interesting sort of timeline over the last 120 years. You can see that there was a lot of activity early on, but lately, in the last 15-20 years, there's been a bit of an increase in the level of activity with some programs being approved around the world. We feel that at Transgene that the platform right now is at a tipping point where we could potentially take our Invir.IO platform forward, and as Éric alluded to, make multiple constructs working with companies going forward. If you look at the oncolytic virus development space right now, it's pretty active. There are well over 100 programs being developed, both pre-clinically and clinically. A lot of these programs are in the early stages of development, phase 1 to phase II. A small piece of these are actually IV programs. What I wanna really focus on is what we've been alluding to most of the presentation, from Adel and Éric, is the opportunity to go into the IV space and target really solid tumors as a potential market opportunity and really think about ways that we could work with companies that are developing their own programs and develop together different constructs that could go after deep, deep solid tumors. If you think about sort of the platform as it is, we have several advantages. You know, I'm gonna sort of step over here. OVs are being developed for lots of solid tumors. We also can take a look at programs, as I mentioned, IV programs in the clinic. We have the ability to talk about different IV administration of OV creating many potential solid tumor pathways, and combination strategies with checkpoint inhibitors that would allow us to work with pharmaceutical companies going forward and boost the activity of some of their programs as well. I wanna talk a little bit about the solid tumor opportunity. You can see that there's a fair amount of activity in the space, solid tumors such as melanoma, pancreatic cancer, lung cancer. We talked a little bit about colorectal cancer. This is where really the opportunity exists for different partnerships with companies going forward that are developing their own programs in solid tumors. When you look at the oncolytic virus development path, there are a fair amount of encoded genes by cancer type. We've mentioned earlier that we're working in some of those areas with some of our programs, with some of our partners, as Éric mentioned. We are, and we can take a look at the space right now and think about the various companies that are in the marketplace, and it's a very small group. It's a group of companies that we're proudly a part of and that when you think about the IV opportunity and the opportunity to sort of take a leadership position in the space with two programs in the clinic, we certainly feel like over the next several years we can develop different programs and constructs that will allow us to maybe move forward into a leadership position in the space. What could we do, you know, talking to companies? We've already been talking to companies, and we'll continue to discuss with companies how we might be able to help them going forward. Thinking about life cycle management, brand leaders, franchise leaders, they think about their programs currently, but they also think about their programs going forward. Eric mentioned the novel partnerships that we have right now with AZ and BioInvent, creating novel constructs, focused on various indications going forward. That's one approach we could take. The second approach we take is to synergize with other IO programs. The third approach is to take a look at pharmaceutical company deprioritized assets. You know, why did they not work? What were things that happened in the portfolio where a program is now on the shelf? Could we possibly rescue that program if there was a toxicity signal, for example, and pair up our oncolytic virus platform with a program like that and figure out a way to maybe resuscitate that program going forward. Then there's the general approach to life cycle management, which is I'm sitting back looking at my lead program, and I see various issues down the road, either competition or IP issues or other marketplace changes that could come up, and could I work with a company like Transgene to extend the life of my lead program going forward. These are all ways that we might be able to partner with companies. We've been engaged in some of these conversations, and there's been some good responsiveness from companies that we've talked to. In conclusion, great science, growing pipeline and a focused business plan. Hedi's mentioned being a world leader in the immunotherapy space. That's something we clearly aspire to be. We will continue to grow the Invirio platform. We've you know we see in the next six to nine months several opportunities to discuss with companies the very ideas that I just mentioned in the previous slide. Being a partner for companies going forward, as I mentioned before, and then the commitment to continuously improve our backbone, with future generations of IV backbones and other oncolytic virus backbones that might allow us to differentiate ourselves in the marketplace. When you conclude from Adel's presentation, Éric's presentation, and my discussion about opportunities with the Invir.IO platform, we feel the time is now for Transgene to step up and think through how we could develop innovative novel programs going forward on our own or with companies that we would partner with going forward in the future. Thank you. Thank you very much, Steven. I hope that in this session we've been able to inform those of you that were not familiar with the technology to bring those of you who are specialists in our most recent advances in the IV route. You heard from Adel that we could detect and confirm viral replication from the IV route. We could note the dosing range where we can expect replication detection in the circulation on reporter genes. We learn from early development where they had surgical pieces that the modeling of the tumor could be confirmed. Something that we of course hope to be able to do in the next step of the VTO-one development program. Something of course we have not been able to disclose now but will certainly be communicated very early on next year is the result of our collaboration with AstraZeneca. We of course rely a lot on them willing to communicate on what we did together. We know, for example, that at the next SITC meeting there will be presentation on what we did together. Really good timing. Important moment we are living with our collaborators in the clinical development. Sorry for not sharing with you today brand new results on the BT-001 trials, but we are actively working at collecting data and biological information from the IT route in addition to the already published IV results. There is room for questions now. Mark. Thank you, Éric. Thank you all. As I was wondering, because you mentioned that for the IL-12 constructs, you plan to use it in lung cancer. Since you are going to use it IV, what will be your mode of delivery? You want to answer? Well, we are going to inject intravenously. The plan is to give the optimized regimen we have defined for TG-602, meaning the one, the three, the five. And that's the plan at the moment. Well, to be on the safe side, we will repeat after three weeks again the injection. As shown with TG602 by Adel, it looks like, even though neutralizing antibodies are there, they are not interfering completely with the activity of the virus and the expression of the payload. To be on the safe side, I agree with you. Three injections might be not sufficient, but the plan is, obviously, to move forward monotherapy to combination with PD-1. That's the plan. Two quick questions. How do you explain the fact that despite antiviral immune response, there is no impact on the efficacy of your oncolytic virus? The second one is there any specific predictive or therapeutic biomarker that you can use with your oncolytic virus? Two very good questions. The first one is that, yes, we know that we have some anti-vector antibodies, part of them only being neutralizing antibodies when used in a titration assay. We know that the way the virus would infect cells is not based on a specific receptor mechanisms that could be prevented by the binding of most affine antibodies. Our vision today is that we might have antibodies covering the surface, but that would be not enough to prevent membrane fusion mechanisms that are responsible for the uptake of the virus by especially by DCs and other cells. We also know that the level of neutralizing antibodies or the titers we could monitor on patients are very variable from... Only very few patients exhibited high titers. We had reactivation possibly of former smallpox vaccination, but it's not the main reason for this rise in neutralizing antibodies. But it's more induction of novel B cells, and that's the reason also for the intensified administration regimen, going every two days rather than every two weeks to hopefully diminish the onset of the B cell response. I don't know if I answered your question, but mostly I would say from what is reported on the neutralization of viral infections, usually this is well described for adenoviruses or smallpox viruses where we know we rely a lot on the ligand receptor interaction for virus infection and the cell entry. It's not the case for poxviruses that are large viruses, mostly infecting cells via membrane fusion mechanisms and are not so sensitive to neutralizing antibodies. The titers that we could confirm in the labs being needed for a full blockade of the virus entry was never reached in our patients. Regarding the second question on the biomarkers, so again, for that reason that we have no receptors that could predict best infected cells, we have no direct markers on the, let's say target cells. What we know are more, I would say, general markers based maybe on the preexistence of antiviral response, but we have no plan to implement those markers. I think I don't see very direct biomarkers with predictive of better responders than looking at my biomarker specialist, but I don't see any relevant biomarkers. At least that's not a plan that we have for the lung cancer trial. Just a question directly related to this one because you say that there is a high inter-individual heterogeneity in terms of antibody titers. Is there any relationship between antibody production, T cell response against the virus, and T cell response against the payload? Of course, it would be much simpler if we had those type of answers. Today, the numbers are not so high to have a statistical analysis. From what we could analyze, there is no correlation. For example, patients with higher titer were not those exhibiting the lesser production of the FCU1. Sometimes Sometimes the opposite. No general trends appeared at the moment in our analysis. Difficult to correlate so simply. We've tried hard to do this type of correlation. It turns out it's much more complicated than that, because as said Éric, the vaccinia virus do not use a specific receptor. Even if you get antibodies around the viral particles, it does not prevent infection of cells. That doesn't mean that there is no neutralization. There is a neutralization, but much more probably in the form of the neutralization you see for RNA therapies, meaning a neutralization that is mediated by a cellular immunity and activation of NK cells. I'm afraid there is one, but probably not as simple as I have a titer of neutralizing in vitro antibody and are related to the activity of the virus. Concerning the correlation with T-cell response We have not done formal correlation between antibody titer and T cell response. When you look at the work from Melcher and all these guys that have studied the question a lot, it's actually turned out to be something beneficial, because if you have a T cell response against the antibody against the virus, you end up having a T cell response against the cells that are infected, and the cells that are infected are the tumor cells. We are most looking for that. One question. Yeah, sure. Just want to touch upon your ESMO presentation, TG 672. You looked at different administrations and different schedules, and it was mentioned that basically with high concentration intensive schedule, you triggered some limited mechanisms. I'm just curious if you have any, you know, hypothesis or any working hypothesis currently, what's those limiting mechanisms for? I was just mentioning about actually, I had that in mind. Just like the observation that CureVac, for instance, can have for the vaccines or the early work of Karikó from BioNTech on their RNA therapeutics. When you overactivate the innate immunity with an mRNA or with a virus, you do trigger some neutralization, which is cell-mediated and not antibody-mediated. It seems in the more intensive schedule we have tested in the six or two, the two trials, that we do observe this type of mechanism in some of the patient groups, and it's actually reflected both on the activity of the payload and the persistence of the virus. How does that translate to the patients? We don't know. But that's probably this type of mechanism. We are taking steps to address that, and we'll have further data to answer and support this hypothesis in the months to come. Thank you. Thank you for clarification. Thank you. Okay, maybe one last question. Yeah. Maybe to come back on the IL-12 construct, just to be sure I understand what you said previously in terms of where you might go. You said after first line, so I guess this time we're in metastatic setting in lung cancer and probably first line after PD-1, PD-L1 or PD-1, PD-L1 plus chemotherapy once the patient has escaped. Was most surprised by the way you think going monotherapy or combination and so did I understand correctly that you may assess both of them? The reason I'm asking is that I would think of keeping PD-1, PD-L1 and adding IL-12, but what if and how can you go monotherapy if, for instance, an ADC Trop-2 comes in in between? Monotherapy looks like a more challenging type of route. No, the issue is very simple. It's regulatory, because to start with, we need data on using the virus as a single agent. We cannot, you know, right away start combining with whether it's an ADC or chemotherapy adjuvant, an anti-PD-1 or where. And moreover, we are expecting translational data from translational analysis, a view because Caleb mentioned those effects, possible dose effects. We need to accumulate a lot of results and we are waiting. We are expecting biopsies at different time points. We are expecting to follow ctDNA to see whether there is some impact on ctDNA. It's needed first to start with the oncolytic virus as a single agent. As you know, it's a limited number of patients, and it's a three-six study design. It will be quite short and not a huge number of patients to start with. Thank you, Maud. May I suggest that we move to the next session? That would be both by Pedro for the scientific perspective on what we have discussed today. We ask Pedro to contribute on kind of putting everything together in the frame of a larger vision and providing you with a comprehensive analysis of our technologies to compare to the state of the art. Eddie would conclude after that. Pedro is coming from University of Lausanne. He's been professor there of immunology and cancer therapy and is now the scientific director of the Ludwig Cancer Center in Lausanne too. Thank... Furthermore, I would say Pedro is the Editor-in-Chief of the Journal for ImmunoTherapy of Cancer, and for that reason, he has a very broad vision on what is going on in the field, the progresses made by the worldwide community in this very hot topic. It was good that he accepted also to share with you how we would see the positioning of Transgene's in this very intense competition area. Thank you, Pedro. Thank you very much, Éric, for your kind introduction. I'm going to try to make a brief overview, along the lines, Éric mentioned. Basically this is a time of great excitement. As you know, immunotherapy has literally revolutionized the treatment of cancer since the first approval in 2011. It actually has spurred a multi-billion-dollar business. The treatment is actually prolonging life and producing even tumor cures. There is a gap. The gap is that only 20%-30% of cancer patients at best benefit from these interesting therapies. 70% there is a need to democratize, so to speak, immunotherapy of cancer and extend it to the majority of cancer patients. Here is where we see as a community the great value of therapeutic cancer vaccines, as was mentioned earlier, is certainly all evidence in the preclinical world point to the promise of vaccines as a way to induce a tumor-reactive T-cell responses, CD4 and CD8, and also to guide those responses to the tumor and to make the resistant patients actually sensitive to these very successful immune checkpoint blockade therapy. The goals of cancer vaccines are manifold, and the more important ones are, in fact, it's difficult to read here, so to recruit the highest avidity TCR T-cell clonotypes, and. Oh, thank you. To actually induce long-lived memory T-cell immunity. In terms of recruiting the highest avidity TCRs, of course, is a game of selecting the appropriate antigens, the appropriate dose, and the timing of the vaccines. In terms of inducing long-lived T-cell immunity, of course, this is the central character of vaccination and probably the most important feature of a vaccine-induced T-cell response. Yet we don't know as immunologists how to do this molecularly. Viruses know very well how to do that. As you know, viruses used as vaccines can induce long-lived, actually lifelong immunity, if we think about some of the common viral vaccines. I see here a big advantage of using viral vectors in general as vaccine vectors. Now, molecularly defined vaccines consist of three subunits, antigen, adjuvant, and the carrier. Of course, here the antigen is clearly the major determinant in successful vaccine formulation. Many antigens, actually thousands of antigens, are available today for vaccine development. There is a big challenge there as to the choice of the antigen for appropriate vaccination. These antigens have been grouped in different categories. The most interesting ones are the shared tumor-specific antigens, followed by the overexpressed antigens, the mutated unique antigens that today are equivalent to the neoantigens, and of course, the virus encoding antigens. There are interesting databases. This one well-created, but of course there is now the opportunity for artificial intelligence driven mining of all the datasets that are out there in the public domain. I think you can imagine that there has been intense early clinical trial testing of vaccines, therapeutic cancer vaccines, even phase III trials, in many tumor types, such as prostate, renal cell carcinoma, lung, breast cancer, glioblastoma, melanoma, lymphoma, and actually quite many others. Even one that actually used naturally isolated antigens from glioblastoma, which is the core business of this German biotech company, Immatics Biotechnologies. The common denominator in most of these clinical trials is that number one, we know how to vaccinate. These are immunogenic to strongly immunogenic vaccines. Number two, that there is a modest, disappointing clinical benefit. The game today is in understanding what are the reasons for failure and how we go about that in the business of making cancer vaccines really clinically efficient and not only immunologically efficient. There are a number of reasons that can be identified. Probably the most important one is that the repertoire of T-cells that we are addressing with the vaccine is a highly tolerant to self-antigens. The majority of those tumor antigens are self. That is one major challenge. The second one is that monovalent antigen vaccines may lead to immune selection and tumor escape. When you use them in a highly metastatic patient, highly immune-selected or immune-edited tumor, the chances for escape are manifold. The delivery methods are suboptimal and, probably the most formidable, barrier is the immune suppressive tumor microenvironment. One major focus in recent years is identifying the appropriate antigens. The concept today is that it is neoantigens, that is the antigens that arise from genetic mutations, somatic mutations in the tumor that are probably the most immunogenic and the most appropriate antigens for vaccination. Basically, because it was already mentioned, there is no central immune tolerance to these antigens. A drawback of this is that there are no two neoantigens alike, and I'm exaggerating just a bit here. This of course calls for personalized vaccines, and you are actually addressing that with your platform. There have been already clinical trials testing this concept with synthetic peptides on the left or with RNA formulated vaccines on the right. Note the logistics that are pretty heavy and, second, the time that it takes to formulate these multivalent neoantigen-based vaccines for each patient. There has been even a phase 1b trial, 85 patients vaccinated with long synthetic peptides representing 10-20 neoantigens per patient. There has been also the use of mRNA vaccine-based vaccines now going back to shared tumor antigens, as done here by BioNTech and the people in Germany. Some of the findings in general, encouraging findings of adequate immunogenicity of these vaccines and even some hints of clinical benefit. In all these cases, what happens, as you know, is the vaccine induces killer T-cells that go to the tumor site and eliminate tumors. Here there is another important concept that has been underlined all this afternoon, and that is of the immunogenic cell death. What is immunogenic cell death? Let's say the Jekyll. If we take Jekyll and Mr. Hyde, the programmed cell death is Jekyll, the good benign type of cell death. As you know, if we think about, we are renewing our gut every three days, and that means that billions of cells are dying in our gut. This death is silent, is immunologically silent. That is good because that avoids the induction of autoimmunity. There is this type of immunogenic cell death that is visible to the immune system, and that leads to the induction of immune responses against the cellular components. You see there a list on the left of agents that can do this. Pretty nasty agents like chemotherapy, radiation therapy, phototherapy, but you have also a virus, and that is also a major advantage of using viruses in this case for their oncolytic properties that lead to immunogenic cell death. Here you have. I lifted from a paper actually published in the Journal for ImmunoTherapy of Cancer on some of the best-in-class oncolytic viruses. You see that vaccinia compares very favorably to herpes simplex virus T-VEC, which is actually FDA approved for the intratumoral treatment of some cases of melanoma. Here we have, in recent years, the field moving back from a very antigen-informed vaccine development to an antigen-agnostic type of vaccine development facilitated by these oncolytic viruses that can do the job of releasing antigens for the immune system to see. In this case, the logistics is simplified, and then you give the immune system the choice to pick up the antigens irrespective of whether they are shared tumor antigens or somatic mutations giving rise to neoantigens. The tumors have them all. Here is again some of the conceptual basis of this in situ vaccination by the oncolytic virus action. Maybe to finish, another important concept is that of immune fitness. Of course, there are many factors that define the ability of each one of our immune systems to react more or less to vaccination in this case, including environmental factors, diet, microbiome, and also age, and genetic makeup and so on. There is a need to capture that kind of immune fitness and all the new technologies, multi-omics technologies driven by artificial intelligence to really treat all those data will allow to also include in the algorithms of vaccination and evaluation of the impact of vaccination, not only the baseline immune response, but also the baseline immune fitness. It is clear that in many cancer patients, the more advanced the cancer, the more advanced the age, of course, this leads to a much reduced chance of having an appropriate vaccine-induced immune response, and hence, the idea to go to apply to move these vaccines to the early stages of carcinogenesis and to the adjuvant setting in the tumor-free, high-risk relapse as you have in your program. I'd like to finish here by saying these are exciting times. Vaccines will come of age. The question is whether vaccines, cancer vaccines, would work or not work. I think that is not the question is. The question is when. I think that the next decade we'll see now the first vaccines with clinical efficacy coming up to the fore. Maybe now I take out my hat from the university and take my hat from editor in chief of the Journal for ImmunoTherapy of Cancer. I'd like to say that reflecting that enthusiasm, precisely the paper that you went over in detail, which is the Sandridge paper characterizing the BT-001 virus together with BioInvent, was actually selected by the editors of the journal as the base paper in the oncolytic and local therapies of the journal, which were actually 38 papers. This is confidential because SITC has yet to announce that during the annual meeting. Thank you. Thank you, Éric, and thank you, Pedro. Wow, what a day. It was so nice to hear the team gathered here and our clinician partners going in depth into the science that is behind our products and the clinical need. That is a fundamental reason that we are here and that our products are addressing. I think you've understood that we are at the right place at the right time with a lot of exciting products. We also several times mentioned our partners, our industrial partners that are key in developing our products. I want to mention AstraZeneca. It's been a very productive scientific partnership in the last three years. Also generated EUR 23 million revenue for our company, which is never negligible. Also keep in mind our partnership with NEC and the one with BioInvent. Here we are, the power of two is at work. Also, we are cutting the cost of our developments by 50%. Without these partners, TG4050 and BT-001 would be not as smart products as they are. We reviewed all the data that we have had in 2022 and gave you the latest information on the progress we are making. We also introduced new exciting ideas for the future. I want to mention, of course, our new IL-12 oncolytic virus that should start a new clinical trial in 2023 targeting the lung cancer as we discussed, and really designed for intravenous administration. Also, you have heard from Éric some words like continuous improvements of the backbone or shuffling. This is the work we are currently doing to build a new backbone that will be better tolerated, really designed for intravenous administration, more activity in situ, hence more oncolytic. It's a bit too early to talk about too much details about this program, but this will come in the next years. Really short term, what's ahead of us? We are still in Q3 2022, and Q4 is knocking at the door. Now I'm sure you have in mind that in Q4 we'll have the interim analysis for TG4001, our most advanced product in phase II, and that will be the first randomized trial results for an HPV-induced anogenital cancer in the world. Also next year, we will communicate new data about the intravenous delivery for oncolytics and the progress of our new programs. Once again, I would like to thank all of you for your support, for your challenge, for your questions. Thanking all the team, the 160 people at Transgene who are doing such a great job that we've seen part of, just a small part of it, today. You are welcome anytime in Strasbourg. Please don't hesitate. Once again, thanks for being here today to listen on the webcast. See you soon. Have a good evening.