Good afternoon, and thank you for joining BioXcel's Virtual R&D Day. At this time, all participants are in a listen-only mode. A question and answer session will follow the formal presentation. As a reminder, this event is being recorded. At this time, I would like to turn the call over to Frank Yocca, Chief Scientific Officer of BioXcel Therapeutics. Thank you, operator. Good afternoon, everyone. I'm Frank Yocca, Chief Scientific Officer at BioXcel Therapeutics. Thank you for joining our Virtual R&D Day. We are very excited to share some new developments today in how we are revolutionizing drug discovery and development through artificial intelligence. We will showcase our proprietary AI platform, discuss our emerging neuroscience pipeline candidates, and update you on expansion opportunities for our lead program, BXCL501. Slide two, please. As a reminder, during today's presentation, we will be making forward-looking statements regarding our R&D plans. These statements are subject to risks and uncertainties that may cause actual results to materially differ from those forecasted. The risk factors that may affect results are detailed in our most recent public filings with the SEC, including our most recent Form 10-Q on file with the SEC and available on our website. Next slide, please. Joining me for today's presentation is Dr. Vimal Mehta, our Founder and CEO, and three members of our R&D team. They include Dr. Friso Postma, Senior Director of Neuroscience and Artificial Intelligence, Dr. Michael DeVivo, Vice President of Neuroscience, and Dr. Robert Risinger, Senior Vice President of Clinical Development. In addition, I would like to welcome our guest speaker, Dr. Robert Berman. Today's event will include approximately 90 minutes of prepared remarks, with 30 minutes at the end for Q&A. With that, I am pleased to turn the microphone over to our CEO, Vimal Mehta. Thank you, Frank. Good afternoon, everyone. We have an exciting presentation ahead of us, so let's get started. Next slide, please. We would like to start with a video that summarizes the innovation ecosystem we have created here at BioXcel Therapeutics. It conveys our high level of enthusiasm about today and our future. Let's play the video, please. Can we go to the slide number,six please? Great. Turning to our mission and core purpose. We are a clinical-stage biopharmaceutical company discovering and developing transformative medicines utilizing AI approaches in neuroscience and immuno-oncology. Our two most advanced programs are BXCL501 and BXCL701. BXCL501 is an investigational proprietary orally dissolving thin film formulation of dexmedetomidine designed to treat agitation. As you know, our NDA is under review by the FDA, and we are also on plan to initiate our Phase III program in dementia in Q4 of 2021. Today, we will give you an in-depth look at our neuroscience strategy, proprietary AI technology, and our differentiated R&D capabilities, which we are using to deliver a sustainable R&D pipeline. We will discuss our exciting new candidate, BXCL502, and expansion opportunities for BXCL501 program. Last week at ESMO, we announced an exciting update about our BXCL701 program. 701 is an investigational orally administered systemic innate immunity activator in development to treat aggressive forms of prostate cancer and advanced solid tumors. This is another important validation of our AI platform, and we look forward to providing additional updates about the progress of our immuno-oncology program. Slide seven, please. Let's now look at our neuroscience strategy. In under three years, we went from conducting our first-in-human trial with BXCL501 to receiving FDA acceptance of our NDA. In this short timeframe, we successfully conducted seven clinical trials in over 800 subjects across various disorders, laying a robust clinical foundation for this neuroscience candidate. We are not aware of any other AI-driven company that has demonstrated a pathway from first-in-human trial to NDA submission as efficiently as BioXcel Therapeutics. We are now less than four months away from our PDUFA date, January 5th, 2022, for BXCL501. It is an exciting time for us. The success of our neuroscience strategy centers around four key pillars. First, we are utilizing a proven AI platform to develop transformative medicines. I will talk a bit more about this platform shortly. Second, we are focusing on hard-to-treat neuropsychiatric symptoms by delineating underlying human biology and behavioral pathways. Third, we are leveraging re-innovation of clinical candidates, new chemical entities, and approved drugs. Fourth, we are building and delivering long-term stakeholder value through this, our sustainable R&D pipeline. In conclusion, we believe our unique combination of AI technology and drug discovery and development expertise sets us apart in our mission to provide access to transformative medicines to patients. Slide eight, please. Looking ahead, our five-year vision is to become the leading AI-enabled neuroscience company. Today, you will hear about this vision is coming together towards realization. To achieve it, we will continue to advance our integrated AI platform, expand our BXCL501 franchise, and build a sustainable, innovative pipeline that includes candidates like BXCL502. We will share more about this later in the presentation. The success of this vision hinges on innovation. We believe that the fact that our first product candidate is pending approval and commercialization is just the tip of the iceberg. In fact, we are confident that the full extent of our value can be even more broadly realized by continuing to leverage our AI platform and move new pipeline assets quickly through clinical development. In many ways, 501 is demonstrating our capability to validate and lay a strong foundation for the many neuroscience candidates that we believe will follow. Slide nine, please. Our AI platform and integrated drug discovery and development capabilities are the underlying foundation of our company. Through the four-step process shown on this slide, we were able to select the most viable candidate and move it through the clinic to NDA on a potentially shortened timeline. Next slide, please. I would like to share some background about our AI platform, which was developed by BioXcel LLC and its subsidiaries, our parent company. Over the last 15 years and through our work with hundreds of partners, the BioXcel AI Innovation Lab has supported over 250 discovery and development projects across multiple therapeutic areas and modalities. It has been validated through the R&D pipeline from preclinical products to NDA submission. At BioXcel Therapeutics, we are leveraging and building a technology that we are confident will help bring additional new medicines to patients in need. I'm excited for you to hear from Friso shortly. He will give you an in-depth look at our AI approaches and show you precisely how our AI capabilities are customized for neuroscience, drug discovery, and development. It is a truly fascinating, powerful, and unique approach. Slide 11, please. Finally, a look at our portfolio strategy for BXCL501. As many of you have heard me say before, we believe we have a significant expansion opportunity with this lead asset. We will continue to let the data guide us in the right direction, and we believe 501 could potentially expand into multiple pipeline indications, as you will hear today, making it a powerful lead product candidate with a large market potential. In a sense, it represents a pipeline within a product. Second, we will continue to expand beyond the U.S. This will include key markets such as EU and Japan, where we are already engaged with regulatory authorities. Lastly, we plan to grow beyond the institutional setting into the community setting, which represents a large market opportunity. We are very excited about BXCL501's future potential with this land and expand strategy. In summary, between our approaching potential FDA approval, initiation of our phase III program in dementia, and our continued innovation, we are very excited about these coming months and beyond. Next slide, please. To add to this context, 2021 has been a year of many important catalysts for BioXcel, and we reported a strong cash position of $273 million in Q2 to fund key milestones. As this slide indicates, there are important additional catalysts to come in Q4. We are excited about them. With that corporate summary, I will now turn it back over to Frank for an overview of our innovative R&D capability. Thanks, Vimal. Next slide, please. Our R&D strategy is predicated on the fact that current medications used to treat psychiatric illnesses offer little advantages over the original agents that were introduced 50 or 60 years ago. The core pharmacology of newer agents is identical to the original agents, with added ancillary pharmacology to treat additional or refractory symptoms or reduce side effects. This clear lack of innovation is due to many reasons. First, there is a lack of understanding of CNS neuronal pathways and their function. Second, there is a knowledge gap of how membrane-bound receptors, brain circuits, and pathways modulate psychiatric symptoms. Linking the function of these key brain elements to human behavior requires a new, innovative approach. Slide 15. It is important to realize that psychiatric illnesses are syndromes made up of a collection of symptoms. Medications created to treat these illnesses don't cure them. There are many hard-to-treat symptoms that occur with these illnesses. They include agitation, irritability, and anxiety. With these specific symptoms in mind, our AI approach focuses on integrating targets, circuits, and pathways with pharmacology and links them to treating specific symptoms. As an example of this is the AI-enabled journey we took to identify a new distinctive treatment for agitation, BXCL501. Next slide, please. Our goal was to find a unique, differentiated approach to treating agitation, eschewing the common antipsychotic benzodiazepine agents that have many associated problems and black box warnings for use in the elderly. Using our AI engine, our first goal was to identify a non-dopaminergic, non-GABAergic pathway related to agitation. Once the locus coeruleus and a hyper noradrenergic pathway was identified as a potential mechanism in agitation, the engine was refocused on a mechanism to control heightened adrenergic activity. This led to the identification of the alpha-2 receptor as a target and the super potent, highly selective A2 agonist, dexmedetomidine, which is also known as Precedex. Existing as an IV solution and used predominantly in surgical suites, we needed to develop an oral formulation that could work as quickly as an IM injection and rapidly produce calming. In proof of concept studies, we learned that the doses needed to calm agitated patients was a fraction of the doses needed to produce sedation and anesthesia. This led to the development of BXCL501, our novel sublingual thin film, ultimately leading to positive proof of concept studies and a development plan that has led to the submission of an NDA in under three years. Later in the presentation, Michael DeVivo will show a similar process that identified our next development candidate, BXCL502. Slide 17, please. Stress is a feeling of emotional tension. For individuals who lack innate resiliency, stress can lead to psychiatric illnesses such as anxiety and depression, and worsen additional psychiatric disorders such as schizophrenia, bipolar disease, PTSD, and addiction. We know that stress has a profound impact on neuronal pathways in the brain, and connecting these pathway changes to specific stress-related symptoms can lead to the discovery of new treatments. Therefore, we are applying our AI ecosystem and machine learning to link stress-related pathway changes and behaviors with circuits, targets, and drugs to come up with testable concepts for novel treatment approaches. Slide 18, please. This slide depicts a dynamic connectivity map, which illustrates the universe of stress-related symptoms and their connection to targets and drugs. This map serves as a substrate for our team for discovering new stress concepts in an unbiased, data-driven fashion. Stress-related symptoms were mapped to diseases and affected brain regions using AI methods. This mapping was then extended to molecular targets and neuroreceptors implicated in the manifestation of these symptoms in relevant psychiatric illnesses. Special emphasis was placed on stress pathways and constituent molecular mechanisms. Based on these mapping results, we have identified several candidate drugs that are currently either in development or have been marketed, all with proven safety, which are systematically being tested to be our next development candidates like BXCL501 and BXCL502. In the upcoming presentations, you will hear for the first time how we use our AI ecosystem to build and generate new concepts along the stress axis. Our VP of neuroscience will discuss how we use this ecosystem to prosecute our therapeutic strategy and discuss the identification of BXCL502. In the final presentations, we will discuss the expansion of our BXCL501 program into depression. Next slide, please. To discuss our AI capabilities and clinical expertise, I am joined by the BioXcel R&D leadership team, consisting of Friso Postma, Michael DeVivo, and Rob Risinger. We are very fortunate to have such experienced and capable neuroscientists and drug hunters. Allow me to introduce them to you. Dr. Friso Postma is a senior director of neuroscience and artificial intelligence. He has over a decade of experience in non-profit and biotechnology, developing and translating innovative solutions addressing unmet needs in neurodevelopmental and neurodegenerative disorders. Over the last five years, he has incorporated AI methodologies to develop digital biomarkers for CNS-related disorders and is leveraging this experience to further develop BioXcel Therapeutics' AI capabilities. Friso obtained his PhD from the Netherlands Cancer Institute and did his postdoctoral work at Harvard Medical School. Dr. Michael DeVivo is Vice President of Neuroscience. He has more than 20 years of experience developing drugs for neurological and psychiatric indications. He was director of neuroscience at Pfizer for eight years and led several cross-disciplinary project teams during that time. Prior to that, he worked at Memory Pharmaceuticals and Trovis Pharmaceuticals, where he was head of discovery. Michael obtained his PhD from Mount Sinai School of Medicine and did his postdoctoral work at Cornell University School of Medicine. Finally, Dr. Robert Risinger is senior vice president of clinical development. He has over 15 years of experience in neuroscience development after spending 20 years in clinical academic medicine. Prior to BioXcel, he worked at Alkermes, BMS, and Janssen, developing medications for depression, psychosis, and neurodegeneration. Dr. Risinger received his medical degree from the University of Pittsburgh, completed psychiatric residency at Emory University, medical and psychopharmacological fellowships at NIH, and served in the Air Force Medical Command. Slide 20, please. With that, I'd like to turn it over to Dr. Friso Postma, who will describe our AI approaches. Friso? Thank you, Frank. Can I have the next slide, please? What you see here is a general slide about artificial intelligence. AI often is considered a black box that magically provides answers. However, AI is a field of science that examines how algorithms and machines can mimic intelligent behavior. The collaboration between artificial intelligence and neuroscience can produce an understanding of the mechanisms in the brain underlying symptoms and behavior. AI, in that sense, is not different from any other sciences like physics, chemistry, astronomy, and genetics, which became successful in neuroscience once it focused on mechanisms relevant to CNS-related disorders. For BioXcel, it is applicable to think of AI as augmented intelligence, where we use specific algorithms and analytics to augment human behavior and decision-making. We apply machine learning, which are algorithms that can learn from data and improve with experience, as well as a subset of machine learning called deep learning algorithms that are organized as multilayer neural networks. I discussed what augmented intelligence means. In the next slide, I would like to discuss why we are using AI. Next slide, please. Yep. During Frank's introduction, you heard about the relative lack of progress made in CNS-related disorders. This is due to several barriers hampering discovery. The first barrier is the sheer volume of scientific data in databases and publications. Over 1 million papers per year are published in the field of life science, which is more than two per minute. During this entire presentation, for example, about 240 papers will have been published. Humans cannot manually curate all that data, and we need AI to extract relevant information from all data sources available to us. A second barrier is the sheer complexity of neural circuitry and how neural circuitry governs behavior. Using AI to pull in relevant data describing the relationship between drugs, circuitry, behavior, and symptoms is one part of the equation. The other part is to try to predict, based on existing information, new links or pathways for clinical stage compounds to treat new indications. This is also what is referred to as the recommendation problem, which is similarly addressed by, for example, Netflix or Amazon, which recommends a new purchase based on previous choices. We use AI to predict or infer novel concepts based on previously known information. The last barrier in neuroscience that we address at BioXcel using AI is the lack of quantitative measures in endpoints in clinical trials. Most, if not all, CNS-related disorders are assessed using clinical skills that are inherently subjective. We use computer vision and deep learning algorithms to quantify behavior preclinically, which we also refer to as behavioral phenotyping, as well as clinically in our digital biomarker program. At BioXcel, we thus operate at the interface of AI, neuroscience, and drug development. To optimally address all aspects of our strategy, we have established an integrated, multidisciplinary global team leveraging talent in the U.S., Europe, and India. I just spoke about why we use AI. In the next slide, I will discuss an integral part of the discovery platform, which is the knowledge graph or labeled properties graph. Next slide, please. What you see depicted here is a layout of a knowledge graph that contains entities, such as drug, target, and symptom, and relationships between those, such as inhibits or binds. By pulling in information from publications and databases, as well as our own data, we populate the knowledge graph and create entity relationships connecting drug to circuitry linked to behavior and symptom. The next slide will show more of this process in detail. We are generating at BioXcel what we call concepts, which we define as testable hypothesis about an existing drug for a new indication. We populate the knowledge graph using natural language processing to extract entity relationships from unstructured data such as scientific literature, but we also use curated structured data. In the example below, you can see the NLP algorithm has labeled a circuit, locus coeruleus, and a behavior, REM sleep, and indicated a relationship, modulates. The entity relationships are stored in a database that connects compounds to targets, to neurons, to circuits, which is the knowledge graph. We subsequently can query and ask questions from the knowledge graph or use a graph-based data science approach to predict new pathways addressing the recommendation problem. Lastly, following this approach, we have the ability to generate a concept proposing the use of a compound for a new indication. Next, we will play an animation of an example knowledge graph. Potential roadblocks. This slide shows that we do not use AI at BTI to merely generate concepts, but apply AI holistically throughout the drug re-innovation process, thereby speeding it up, as was shown by Vimal earlier. As mentioned, we use NLP for entity relationship extractions and graph-based data science, but also have developed machine learning algorithms to select out of 6,000 compounds with clinical safety profile, roughly 1,000 with predicted blood-brain permeability. Once we have generated concepts, which takes about six months, we also use machine learning and deep learning paradigms to test the concepts preclinically. For example, during behavioral phenotyping, which quantifies mouse behavior. Preclinical proof of concept is followed by analytics in the form of a decision matrix and can take up to 12 months, prioritizing and assessing the opportunity for BioXcel based on various criteria such as unmet need, licensing, formulation, and IP. Lastly, during concept drug development, we use our own clinical trial data to reiterate concept generation and update the knowledge graph, as well as develop digital biomarkers to quantify human behavior that will enable precision medicine. Therefore, at BioXcel, we are developing an AI ecosystem, augmenting the drug candidate development process at various stages and accelerating timelines. An example of behavioral phenotyping, as I indicated under concept testing, is further explained in the next slide. Compounds are tested in SmartCube, which is a system that uses computer vision to monitor and extract behavioral features from mice behaving both actively and reactively. The left side is showing a mouse in a cage looked at from three different angles by the computer. The animal is challenged with different stimuli, such as cage floor moving up or a puff of air from a nozzle on the right. On the bottom right, you see the real-time assessment and classification of the behavioral features by the algorithms. Over 1,000 features are extracted by the system and compared to a database of behaviors obtained with marketed drugs for known indications. The deep learning classifiers that were trained on the databases are then used to assign signatures to the test compound. For example, on the right, the test compound has high activity and a clear antidepressant signature. In addition, with increasing doses, the mechanism of the compound changes from a tricyclic antidepressant to that of a serotonin reuptake inhibitor. These technologies have allowed us to quickly test concepts, compound efficacy, or mechanism of action. Similarly, we aim to use these technologies to behaviorally phenotype disease models and assess phenotypic reversion with a compound. To summarize, at BioXcel Therapeutics, we have built an AI ecosystem that is tailored to our specific needs and has been successful in generating novel drug candidates for unmet medical needs. You will be hearing shortly from Mike, who will illustrate this point presenting on BXCL502. It is important to realize that the type of feature extractions, AI, and analytics used preclinically have the potential to be translated to human biology. Which is what brings me to my final slide that will show how we are now expanding the AI ecosystem into wearable devices and digital biomarkers. Next slide, please. Human behavior, including agitation, can manifest in a physiological manner in various ways, which can be measured in an objective and quantitative manner using wearable devices. At BTI, we are leveraging external expertise to develop a platform designed to use wearable technology and AI to remotely detect warning signs of a developing agitated state. BTI and Yale have engaged in a study of potential biomarkers associated with the mechanism of action of dexmedetomidine, BXCL501. Preliminary data suggests that these biomarkers, such as heart rate variability and skin conductance, change after dosing with BXCL501 in agitated patients with schizophrenia. The ability to detect when a patient is about to be agitated may allow for precision medicine approaches and expand therapeutic areas for BXCL501. BTI and MindMed are developing a digital infrastructure designed to remotely collect wearable device data in patients. Feasibility studies have been completed, and a study is expected to be initiated in healthy volunteers subjected in a controlled environment to stress. Feature sets extracted from accelerometer and heart rate variability data recorded by the Apple Watch will be used as a reference and provide a golden standard to train algorithms that will detect emergence of agitation. Next slide, please. With that, I will now turn the presentation over to Dr. Michael DeVivo. He will discuss our robust new product concept strategy. Mike, over to you. Thank you, Friso Postma. I am going to focus on how we use AI to develop innovative concepts, including identification of BXCL502, by matching behaviors to brain circuits. Over the past five years, there have been great advances in technologies, including optogenetics in mice and novel MRI techniques in humans, to identify important pathways in brain. These efforts have resulted in an explosion of data linking circuits to behaviors. In order to leverage these complicated data sets, we need AI. We showed you one example of a stress pathway, the locus coeruleus pathway, but there are over 100 more clinically important pathways that are affected by stress. Our AI capabilities allow us to not only identify the pathways but also the drug targets that regulate their activity. Once these drug targets have been identified, we match them to compounds in our virtual compound library. These compounds, about 6,000, have known safety profiles because they were previously administered in clinical trials. We have further refined our library by assessing the probability of CNS penetration by examining the structural features. In the end, we have about 1,000 compounds that meet these two criteria, human safety and CNS penetration. We correlated these 1,000 compounds to their pharmacological targets and used this knowledge to identify new drug concepts. In the next slide, I will show you why it is important to understand the biology that mediates the symptoms of interest in order to develop truly innovative and effective drug concepts. Next slide, please. By understanding pathways, we can do a much better job of targeting drugs to underlying biology. As an example, we know that stress causes LC neurons to fire in a tonic or continuous mode rather than a phasic mode. Tonic firing is associated with agitation and anxiety-related behaviors. By knowing that alpha-2 adrenergic receptor agonists like dexmedetomidine, the active ingredient in BXCL501, restores phasic firing, we may correct the problem. For most neuropsychiatric drugs, the pathways affected are not known. Drugs that treat agitation, for example, benzodiazepines and antipsychotics, simply tranquilize the patient. We need to do better than that. Finding drugs that restore normal function to specific pathways is the basis of our innovative and sustainable pipeline. Next slide, please. Our newest pipeline drug, BXCL502, was discovered using the stress-activated pathway strategy. Before I tell you how it was discovered, let me state some of the key attributes of 502 and why we are so excited about it. BXCL502 is intended as a monotherapy for the chronic treatment of agitation in patients with dementia. BXCL502 is an antagonist at a GPCR receptor that modulates serotonergic signaling in the cerebral cortex. We have tested 502 in two preclinical models of stress-related behaviors and it exhibited efficacy comparable to benzodiazepines and antidepressants. Prior clinical experience with the compound for a different indication, not agitation, suggests that it has a good safety profile in humans after chronic dosing. The compound also exhibited early signs of efficacy on a validated clinical endpoint. BioXcel is now beginning formulation development required for the initiation of clinical studies. Next slide, please. BioXcel aims to become a leader in the field of treating agitation-related symptoms. Towards that goal, a search of new agitation-related pathways, not locus coeruleus, was initiated. BioXcel found a novel pathway outlined in this figure that is affected by stress. Further examination of the literature about this pathway suggests that modulating it would be highly effective in treating agitation. The knowledge graph below is a 2D representation of that effort and shows that the target of 502 and its regulation of the stress axis. Our compound library divulged a compound, BXCL502, that is a potent and selective antagonist at a receptor in this pathway. We did further examination of 502 to establish a high level of confidence in the rationale to develop it for chronic treatment of agitation, as you will see in the next two slides. After discovering BXCL502, we conducted studies to confirm its mechanism of action. We found that BXCL502 is in fact a potent antagonist at the receptor, shown in the figure on the left, that was identified in the knowledge graph. This receptor is a GPCR that modulates serotonin signaling in the cortex. In order to determine if our prediction that BXCL502 would be effective in stress-mediated behaviors, BioXcel tested 502 in two preclinical models of stress. In both models, 502 demonstrated robust efficacy comparable to benzodiazepines and antidepressants. Both of the preclinical models shown in the figure are commonly used to identify antidepressant and anxiolytic drugs. The data suggests that 502 may be an effective and novel mechanism to treat neuropsychiatric symptoms. Finally, BXCL502 was highly effective in increasing latency of REM sleep in rats. This finding is highly translatable to humans. Using this study design, we anticipate using REM latency in studies with healthy volunteers to choose doses prior to starting phase II patient studies. Next slide, please. Surprisingly, after identifying BXCL502 as a candidate drug for agitation, an AI-driven search discovered that this compound was used in prior clinical studies. BXCL502 successfully achieved a significant outcome in a secondary endpoint in patients with dementia. This outcome was achieved using a clinically valid endpoint to measure agitation. It was not the primary endpoint for those studies and was not pursued. However, those studies have partially de-risked this compound because of the large number of patients that were dosed, several hundreds, and for the duration of dosing, about one year. Moreover, the efficacy signals detected in these studies validated our search strategy. In summary, our AI strategy identified a target and compound that, unknown to us at the time, had exhibited an early sign of efficacy in an agitation-related endpoint in a clinical study. We consider this finding as validation for our platform. Next slide, please. BXCL502 is intended as a standalone drug for chronic treatment of agitation in patients with dementia. However, we all realize that these complex symptoms sometimes require effective combinations of drugs to be adequately resolved. Therefore, BioXcel undertook phenotypic studies to determine if 502 would have properties consistent with combining it with BXCL501. In the graph on the left, both 501 and 502 are active in producing a characteristic phenotypic signal as defined by neuropsychiatric drugs. Co-administration of low doses of 501 and 502 exhibited both additivity and synergy in the specific profile. By using a type of principal component analysis, decorrelated rank feature analysis of these phenotypic behaviors, it was revealed that 502 and dexmedetomidine, 501, share a high degree of overlap. This indicates that both drugs, despite working on entirely separate receptors, are able to achieve the same characteristic behavioral phenotype. Not shown in the graph, benzodiazepines exhibit a very different profile. This profile is not common to all anxiolytic and agitation drugs, but appears to be special to 501 and 502. The results from this study are encouraging because they suggest that 502 may be as effective as 501 and that the two drugs may be combined, allowing a higher degree of efficacy and possibly lower doses of each drug to be used. BioXcel is continuing to explore this combination, and others, and is filing patents that support the use of them in the treatment of neuropsychiatric symptoms. The knowledge of the pathways and the ability to perform behavioral phenotyping greatly enables the strategy of intelligently combining drugs that work synergistically. Next slide. Okay. To sum up on BXCL502, I want to reiterate our great excitement about our next pipeline drug. As discussed, this target comes from an AI-enabled search of pathways and associated drug targets that are affected by stress and elicit clinically important behaviors. Our confidence and rationale for this compound stems from its robust efficacy in two stress-mediated behavioral assays with efficacy comparable to benzodiazepines and antidepressants. Our search strategy was validated by the unexpected finding that this compound was reported to exhibit an early sign of efficacy in patients with dementia using a clinically valid endpoint for agitation. We found that 502 was dosed up to a year in large cohorts of patients, de-risking potential safety concerns. Phenotypic screening suggests that 502 may be working on pathways that work in tandem with 501, allowing benefits from combining 502 with 501. Currently, we are working on the necessary formulation activities and regulatory requirements in order to start proceeding with our clinical development strategy. BXCL502 may have properties required to be as effective as benzodiazepines and antidepressants without the limitations of those classes of drugs. Next slide, please. As Frank mentioned previously, a dynamic connectivity map was constructed to identify symptoms affected by stress. For that effort, we're focusing on symptoms that are subcomponents of clinically validated endpoint scales like HAM-D, NPI, and YMRS. By targeting these specific symptoms, we anticipate that effective drugs will be more likely to meet endpoints in clinical trials. For example, as shown in the figure, agitation, impulsiveness, irritability, and many more are stress-related symptoms that are part of clinically important endpoints. We combine this list with opportunities and prioritize them. To date, we have uncovered over 20 opportunities that are under assessment. A key prioritization is a robust patent strategy. These opportunities are yielding promising leads that will sustain our pipeline over the foreseeable future. I will now turn the podium over to Frank, who will introduce our guest speaker. Frank? Thank you, Mike. I would now like to turn your attention to BXCL501 and a discussion of potential expansion indications. Since we generated SERENITY I and II data back in 2020, secondary endpoints measured in those studies suggested that 501 may be treating more than agitation, particularly anxiety, irritability, dysphoria, and disrupted sleep rhythms. These symptoms are comorbid with several psychiatric illnesses, especially depression, where such symptoms can delay treatment relief. With that in mind in stating the problem, I am very pleased to introduce our next speaker, Dr. Robert Berman, who's currently adjunct professor of psychiatry at Yale University School of Medicine. Dr. Berman has had successful careers in academia as well as in industry. At Yale, he was at the forefront of depression research, conducting the first depression studies on ketamine with Professor John Krystal at Yale University. In industry, he was the lead clinician for developing Abilify for augmentation treatment for depression. More recently, as CMO and co-founder at Biohaven, he led the development of the novel migraine medication, Nurtec. Today, he will discuss major depression, focus on anxiety. Dr. Berman, thank you for joining us. Thank you, Frank. It's a pleasure to be here. Today I want to give you a perspective on Major Depressive Disorder. My personal independent perspective, and not necessarily that of BioXcel. My goal is to give you context to the BioXcel initiative in studying BXCL501 in depression. This perspective is based on over 30 years' experience treating and researching depression. Here are my current activities and relevant disclosures. We can go to the next slide. You may be familiar here. Major Depression confers a major public health burden, and despite over two dozen approved antidepressants, there remains a massive unmet need. Just to call out a few, there are 300 million antidepressant prescriptions filled annually, 30 million or more Americans are currently on antidepressants, 25% of those diagnosed remain ill after starting treatment. The net is depression is common, and it is not very well treated. The next slide. Let's start with the basics on what we actually mean by clinical depression. The Diagnostic Statistical Manual, the DSM, defines depression using specific criteria that a patient needs to meet. Beyond those criteria, there are many symptoms that are commonly associated with a depressive episode, and some of them are indicated here. If we go to the next slide, I want to highlight a few core symptoms. First, anxiety, restlessness, psychomotor activation, irritability, and panic can be viewed as one phenomenologic cluster of anxiety symptoms. This is of pointed relevance based on the clinical data of 501 in studies in dementia and schizophrenia and bipolar disorder. Our next figure will show some data on 501 and how it has the potential, not the next slide, potential to reduce agitation in a very rapid manner in these populations. This clinical data that we presented, combined with the mechanistic schema, that connectivity map you saw earlier from Dr. Yocca and Dr. DeVivo, will provide a compelling rationale for studying BXCL501, specifically in anxiety symptoms associated with depression. Further, BXCL501 and agents similar have been shown to have favorable effects on sleep architecture, also providing a rationale to study BXCL501 in depression. We'll stay on this slide. It's reasonable to speculate that addressing these symptoms could have knock-on effects on broader symptomatology. For instance, anxiety is associated with elevated rates of suicidality, especially impulsive acts. Manipulation of sleep architecture, such as the case in sleep deprivation, is a quite efficacious treatment for treatment of refractory depression. Improving sleep architecture itself may impact a broad range of depressive symptoms. This is consistent with the connectivity map that you've seen. Let's go on to the next slide. To focus on anxiety in depression. Depression and anxiety, they travel hand in hand. If you're a clinician, that's common knowledge. While this has always been recognized, the methods for characterizing anxiety and depression have varied and evolved. Okay, look to the far left. In one schema, we can think of anxiety as occurring on a continuum within depressed patients, ranging from no anxiety to marked anxiety. The epidemiology here has characterized a fraction of patients who score over an arbitrary threshold on an anxiety scale. 50%-85% of patients using this schema are thought to have significant anxiety. Now, focus on the middle there. One can also just assess the fraction of patients who meet DSM criteria for specific anxiety diagnoses. Approximately 40%-70% of depressed patients suffer from such comorbidities. Lastly, and more recently, the DSM-V, or the most recent version, has defined an entity called major depression with anxious distress, whereby patients have to have a certain number of anxiety symptoms. 54%-78% of depressed patients have been shown to meet the criteria of anxious distress. Overall, no matter how it's classified, the rate of anxiety and depression is very high. I want to pause. You may wonder, what about the 20%-30% of patients who are not classified as anxious? Well, in my clinical experience, even the vast majority of those, they do have some anxiety symptoms. However, they just haven't met the thresholds that are baked into the schema in this slide. We can go to the next slide. This chart shows a sampling of a couple thousand patients with major depression, and it describes the common comorbidities. You see that panic disorder, social phobia, PTSD, and generalized anxiety disorder are the most common of them. We can go to the next slide, and we'll return to the DSM-V. Prior to this last version, up through DSM-IV, there was no specific qualifier to describe anxious depression. I believe this addition to our standard accepted diagnostic guidelines provides an avenue for developing specific treatment interventions to target a subset of patients who can now be systematically defined in a non-arbitrary, validated manner. The DSM-5 in that to meet anxious distress criteria, you must exhibit two of the following five symptoms there. Keyed up or tense, restlessness, difficulty concentrating, fear of pending doom, a feeling that you might lose control. I'll point out that this classification system seems to perform better than the other ones mentioned a couple of slides ago in identifying patients with poorer outcomes. Let's go to the next slide. This really gets at the question: what is the clinical relevance of anxious depression? These are patients with greater severity on generalized depression scales. They have younger onset. They have a greater likelihood of chronicity, greater functional impairment. They have increased risk of suicide, including impulsive acts, and a greater risk of discontinuing treatment. The net is these patients really do have poorer outcomes. Go to slide 49. It's an example of the largest naturalistic study in depression to date, the STAR*D, in which approximately 3,000 depressed patients were followed through a multi-step treatment algorithm with the goal of getting them to remission. In this table, I highlight the remission rates for patients undergoing their first treatment step. That's treatment with citalopram. The remission rate was 33% in those who were not considered anxious versus 22% in those who were considered anxious. A couple points. First, this is an unfortunately low rate of remission for either group, but it's particularly low for those who had baseline anxiety. In fact, it's two-thirds the rate of remission. The rest of the table talks about other definitions of outcomes, and the message is similar there. If we go to the next slide, the patients who didn't respond, didn't have remission, went to step two of the STAR*D, where they were randomized to an assortment of different medications, as shown there on the X-axis. What you see here is that uniformly, patients with substantial anxiety in the dark blue have a much lower remission rate compared to those without anxiety in the light blue. That's the point I wanted to make, that it really is associated with poor outcomes. Let's go to the next slide. How are patients with anxious depression treated pharmacologically? Okay. There's no drug specifically approved for depression with anxious distress or even anxious depression, however you want to qualify it. Typically, patients are prescribed a conventional monotherapy, be it an SSRI or SNRI. Those are considered to be modestly effective. Typically, patients are not targeted for their anxiety. They sort of are suffering along with their anxiety component while the antidepressants kick in. Augmentation can be considered with the following agents. Benzodiazepines, atypical antipsychotics, I'll talk about those, as well as buspirone, which is approved for anxiety. There's no controlled data for the use of buspirone in the anxiety of depression, and I won't focus on that. It's not used that much anyway. If we go to the next slide. The patients who don't respond to monotherapy treatment have been studied with the use of adjunctive atypical antipsychotics such as aripiprazole, quetiapine, brexpiprazole. As shown in the graph here, anxious depression does respond to such augmentation treatments, and this is with adjunctive ABILIFY, aripiprazole. They respond to that better than adjunctive placebo, which is in the light blue bar. Adjunctive aripiprazole is in the dark blue. There is a sense of benefit there. Really, a clinician must weigh the risks with the benefits for these agents, as they're commonly associated with very important adverse events as listed there: akathisia, weight gain, somnolence, and metabolic disturbances. Those can be very problematic for patients. Let's go to the next slide and talk about adjunctive use of benzodiazepines. Okay. They're obvious for consideration since they directly address anxiety in non-depressed populations. There have been multiple trials in the end of the last century. A Cochrane review summarizes many of those trials on the use of benzodiazepines with co-administration of antidepressants. What they showed is that you can see a benefit both on anxiety and depression in the first few weeks, four weeks of depression, but they wane by week six, and six through 12, the benefit is not so apparent. Further, there were trends for reduced rates of discontinuation of drug due to AEs, adverse events. Keeping patients on pharmacotherapy is critical to getting them better with pharmacotherapy. That's a major benefit. Regarding the efficacy of benzodiazepines. Oh, actually, let's focus on the panel on the right there. This is the most recent study in patients, and it was a group of patients who were randomized to receive fluoxetine plus clonazepam versus fluoxetine alone. This is the subset that had anxious depression, a total of about 46 patients. We see that the anxious depressed patients who were randomized to concurrent fluoxetine and clonazepam demonstrated a remission rate of 36%. Versus a 4% rate for those on fluoxetine alone. This does point to the potential benefit after three weeks of co-administration, in this case, of benzodiazepines. This compelling efficacy really has to be set in the caution of its adverse events. You see tolerance, you see fall and accident risk, cognitive dysfunction, dependence that can occur within days of administration, as well as addiction, abuse, and misuse. Despite these very concerning adverse events, these agents are commonly prescribed, and over 90 million prescriptions were reported in the U.S. in 2019. This common use, despite these concerning adverse events, compel the agency to change its warning label for the benzodiazepine class. It now has an 18-line black box warning, and this was as recently as September of last year. I'll say, speaking as a clinician, I never felt good about prescribing benzodiazepines for the reasons here, but there really were no alternatives for addressing patients with their immediate suffering related to anxiety. From a personal perspective as a psychiatrist, I am excited about the prospect of 501 and its ability to potentially confer as rapid effects in the anxiety of depression as it has demonstrated initially in the agitation associated with bipolar disorder and schizophrenia, even dementia. If this speculation bears out in clinical trials, then 501 has the potential to offer the benefits seen with the benzodiazepines, but with a more favorable side effect profile, and that would allow for broader use and uptake. If we go to the next slide, let me summarize what I've gone over. First, anxious depression is very common, and even those who don't meet the criteria spelled out earlier have some anxiety symptoms. There are no approved treatments. Clinicians are left with sub-optimal pharmacologic choices to address the anxiety component for their patients with anxious depression, and these choices are associated with very limiting side effects. On the basis of the potential efficacy on the anxiety symptoms alone, 501 may address important unmet needs in depression, perhaps with a rapid onset on the anxiety component. I just want to point out, there's also the potential for improving sleep architecture, and addressing either one of those clusters may have a knock-on beneficial effect across the broader spectrum of depression symptoms. In conclusion, the rationale for studying 501 in major depression is very strong to me, and I eagerly look forward to the results of the studies that BioXcel is planning. Thank you for your attention. Thank you, Dr. Berman, for these valuable insights and perspectives. I'd now like to move on to our Dr. Robert Risinger, who will discuss how we are expanding BXCL501 development into depression. Rob, over to you. Thanks, Frank. Next slide, please. You've heard Dr. Berman describe the significant unmet medical need in major depressive disorder. Now I'd like to describe how our data guided us to evaluate development in this area. We knew targeting norepinephrine release from the locus coeruleus also impacts the HPA axis. We recognized that agitation manifests as a range of symptoms and knew we may be able to identify other symptoms which could be relieved when BXCL501 modulates this circuitry. To uncover these, we added measures to our trials and analyzed phenotypically related symptoms. We hypothesized that discovering common effects in different illnesses indicates BXCL501 could be applied with great impact to address unmet needs in otherwise untested diagnoses or illnesses. We discovered that BXCL501 reduced symptoms of anxiety commonly reported by depressed patients, and as we just saw, these are associated with poor outcomes. If BXCL501 improves these safely, it could provide a better treatment outcome for depression. Next slide, please. In three different illnesses, we found overlapping and compelling evidence of reduced tension, nervousness, and anxiety. On the left, within two hours, depressed bipolar patients enrolled in SERENITY II had an improvement in agitation, as measured by the PEC scale at the lower and as well the greater dose. In the middle panel, beginning at six hours, patients with schizophrenia had improved ratings of anxiety, which lasted through the next day for the 180 microgram dose. On the right, with the first day and continuing over one week of dosing, objective ratings of anxiety and irritability were significantly reduced during opiate withdrawal. Thus, in bipolar patients experiencing a major depressive episode, as well as patients with schizophrenia and patients undergoing opiate withdrawal, we have strong evidence of a similar reduction in agitation, but more importantly, BXCL501 reduced reported restlessness, tension, and anxiety in those who were depressed. Taken together, we believe this provides strong evidence that not only can BXCL501 treat agitation, which is our first NDA submitted, but may reduce restlessness, inner tension, and anxiety in patients. Next slide, please. Now, let me drill down a bit and describe further compelling data. In the SERENITY II pivotal trial for bipolar agitation, we enrolled patients in any mood state. In the previous slide, I showed robust statistical efficacy in reducing the PEC total score for both doses in manic agitation. From our deep knowledge, we had confidence that calming the stress circuitry with BXCL501 would hold true for patients with depression as well as mania. We discovered that both doses of BXCL501 effectively reduced the tension item in depressed patients at two hours, suggesting even lower doses may prove to reduce anxiety. Tension is rated according to symptoms of restlessness, nervousness, anxiety, fear, and panic. On the whole, this statistically significant improvement in anxiety across several illnesses suggests BXCL501 may be developed for rapid treatment of depression in those with agitation and anxiety. Next slide. With this data, we began to consider other evidence that might lead us to pursue development for depression. For example, it is known dexmedetomidine has effects on sleep, which are complementary to treating depression. Depressed patients report disrupted sleep, often stating their tension or worries make it difficult to fall asleep. When they do sleep, upon awakening, they report sleep is not restful. In fact, they may awaken early or often in the middle of the night. The rapid onset of non-restful, non-slow-wave sleep is a phenomena termed decreased REM latency. This is a widely accepted biomarker of depression. In contrast with other sleep aids, dexmedetomidine promotes normal sleep patterns. In the leftmost panel, by increasing REM latency. The middle plot shows in rats, dexmedetomidine reduces the time to achieve slow-wave sleep, which is lacking in depression. On the right, in humans, an infusion study by Akeju at Harvard shows increasing concentrations tend to increase the total time spent in restful N3 slow-wave sleep. We have translational data from rats to humans that suggest BXCL501 may have complementary effects on restorative sleep. REM effects have been used for over 20 years as a potential biomarker of an antidepressant effect. Most antidepressants and benzodiazepines suppress REM sleep, but they do not reinforce normal patterns or restorative sleep. This data bolsters our own clinical observations, suggesting that BXCL501 may be helpful to improve depression outcomes, especially those with troubling anxiety. Next slide. Two pivotal SERENITY trials demonstrated a rapid, robust reduction in agitation after a single dose. In order to potentially address depressive symptoms over time, we expect our trials will require BXCL501 to be administered daily. We also know that BXCL501 is a potent alpha-2 agonist, and for agonists, questions inevitably arise about tachyphylaxis, which is when the same dose produces lesser effects upon repeating. This slide shows there was no diminished response to successive doses. In other words, no tachyphylaxis. In the study, rats were implanted with electrodes, with the frequencies plotted over time and the power indicated by color intensity. Red indicates greater intensity. The top portion of each plot is high frequencies, the bottom are low or slow-wave frequencies. As you can see, the effects after the first dose and the last dose are virtually identical after 21 consecutive days of dosing. The spectrum on the top and bottom right are similar, indicating no change in brain activity after repeated doses. Also, as you can see, dexmedetomidine reduces high-frequency signals consistent with calming. When food is delivered to the animal, the rats are easily aroused from this state, and this too, is very similar to what is reported with patients. Note the dex injection also increases slow-wave frequencies, seen in red at the bottom of the plot. This is similar to the N3 stage of human slow-wave sleep I just described in previous slides. This quantitative biomarker reproduces clinical effects that are observed in humans. Effects with the first dose are consistent and repeated after 21 consecutive days of dosing, and these effects are consistent with published human data and strongly support our clinical development program, which will be designed to test repeated doses of BXCL501 to rapidly improve core symptoms of depression. Next slide, please. Now, you may be asking, what does sleep or anxiety really have to do with depression? This is the gold standard regulatory scale to demonstrate clinical antidepressant effects, the Hamilton Depression Rating Scale, or HAM-D. Highlighted are six of the 17 items, rating effects including agitation, insomnia, and anxiety, for which I described earlier clinical evidence as potentially treatable symptoms with BXCL501. Let me take a moment to explain that sleep is a two-edged sword. It's both a cause and consequence of depression. Depressed patients tend to report interrupted sleep with difficulty falling or staying asleep. SSRIs and benzodiazepines tend to disrupt sleep. As an adjunctive to SSRIs, BXCL501 may have potential to improve sleep. Patients often report their restlessness, worries, and anxiety make it hard to fall asleep or gain restful sleep. In combination with SSRIs, BXCL501 may have potential to reduce associated agitation, tension, and anxiety. As Dr. Berman described, these specific effects are often untreated and do result in poor clinical outcomes. Of course, benzodiazepines may be prescribed with SSRIs. The FDA has added a black box warning with the serious risks of abuse, addiction, physical dependence, and withdrawal for the class. There are none of these concerns with BXCL501 or dexmedetomidine. After greater than 20 years of clinical safety, dexmedetomidine is non-scheduled with no evidence of abuse, addiction, or physical dependence when delivered intravenously at greater exposures than BXCL501. Alcohol and substance use is common, and benzodiazepines risk reactivating abuse in this vulnerable population. Unlike dexmedetomidine, benzodiazepines disrupt sleep architecture and tend to reduce and not improve N3 restorative sleep. With the weight of clinical and non-clinical evidence pointing to a reduction in ratings of anxiety and tension and complementary effects on sleep, our next step is to begin testing the safety and effects of repeated doses of BXCL501. Symptoms of depression are not uniformly treated with SSRIs or SNRIs. These treatments disrupt sleep, and as Dr. Berman described, symptoms of anxiety and sleeplessness are often untreated or unresolved with these, resulting in poor clinical outcomes. We plan to initiate a proof of concept trial in outpatients experiencing depression. The program is designed to test the potential for daily dosing of BXCL501 to augment the response to standard of care antidepressant treatments. To achieve this, we enlisted a team of leading academic and industry experts to help us elaborate the details, designs, and scope of our studies. Having completed this, we submitted our program for regulatory review. In conclusion, I'm excited to report we have a meeting scheduled to align our clinical development program for depression with the FDA's Division of Psychiatry. Next slide. Frank, I'll now send it back to you. Thank you, Rob. With our prepared remarks being done, I would now like to open the microphone and invite your questions. Operator, would you please open the lines? Absolutely. Ladies and gentlemen, we will now be conducting our question and answer session. You may type your questions into the Ask a Question field on the left side of your screen on your webcast console, or if you'd like to verbally ask your question, please press star one on your telephone keypad at this time. The confirmation tone will indicate that your line is in the question queue. You may press star two if you'd like to remove your question from the queue. For participants using speaker equipment, it may be necessary to pick up your handset before pressing the star keys. Our first question is coming from Greg Harrison with Bank of America. Please proceed with your question. Hey, guys. Great presentation today. Very helpful. Thanks for taking our question. We wanted to ask, how broadly do you think that the 502 mechanism can be applied? Could it be developed in additional agitation indications or even beyond into some of the others you've talked about today, like stress or anxiety or depression? This is Frank. Thanks for the question. Like Mike said, when we went about trying to find new mechanisms and really mine the stress axis. We try to find different neurotransmitter systems that would impact in a manner similar, say, for example, to 501. The fact that we're now affecting the stress axis in a different area, say, for example, we firmly believe based on the clinical data of the compound, that agitation is a real solid endpoint. You're absolutely right, it does open up the opportunity for us to evaluate this in other areas in psychiatric illnesses, especially where we see these types of symptoms. You're absolutely correct. Great. One just at a broader level with respect to the AI platform. Under what circumstances would you consider starting your own discovery programs to go after some of these targets that the AI platform identifies rather than leveraging existing drugs as you've been doing? Thank you, Greg. This is Vimal. That's more or less a strategy question. Initially, we chose to focus on clinical candidates where we had established safety profile, and we can get to the finish line in filing NDA like we did with the 501. We continue to evaluate our strategy because you are absolutely right, we are not only identifying compound, we are identifying novel targets also. If we see any modality, whether it's a small molecule or it could be other modalities, therapeutic modality, we will consider either partnering with someone who has the capability, and we can bring this biology and the target to be able to make sure that we can develop those targets and have the new chemical entity as well as these innovative programs and addressing those unmet medical needs. We do realize that everything we identify cannot be addressed with available compound library that our team already identified and showed that there are about 6,000 compound in the clinic. They have established safety profile and about 1,000 have the CNS penetrant property. If we have to go beyond that will be a right reason to address that unmet medical need to partner externally or initiate that program on our own. Great. That's helpful. Thanks again, guys. Thank you, Greg. Thank you. Our next question comes from Chris Howerton with Jefferies. Please proceed with your question. Hey, everybody. Thanks so much for taking the questions. Totally agree this was pretty exciting and really informative. Thanks again. I guess maybe for me, one of the questions would be, obviously I understand that you're about to have these discussions with the FDA on what the clinical path might look like. I guess I would ask you to speculate on what your view might be on efficacy endpoints in the MDD subpopulation that presents with high degree of anxiety. A second question, if I may, is I guess, if you could just maybe give us a little more color on kind of the expected formulation or dosage forms of 501 in MDD and then of course perhaps what 502 might look like and finally what the combination might look like. Thanks again. Thanks, Chris. This is Robert Risinger. I think the first question is specific to the clinical path. We believe that the FDA doesn't need, in fact, to approve a different sort of endpoint. The endpoints in depression are fairly well, if you will, codified. We know that the Hamilton Depression Rating Scale is a generally accepted regulatory endpoint, both in the U.S. and Europe and elsewhere. We don't believe we need to come up with a new endpoint. We do believe that this will improve the outcomes in terms of depressive symptoms. We will add measurements of anxiety to our trials in order to demonstrate the effect on anxiety symptoms, but really the registrational endpoint for an antidepressant must be depression and its effects on a depression measurement that the regulators agree upon. How about the dosage? What dose? Oh, I'm sorry. I missed that question. The doses that we'll be examining are in part based on what we have done previously in humans. We've done this in our I can't remember the numbers now. We did this in a dose-ranging study in patients with schizophrenia who were agitated. We don't know that we'll need to get up to the higher doses that were required in a, we'll call it grossly agitated patient. We believe that the doses will be lower than that. We do believe that the benefits will help a much broader group of patients with depression, not just those with agitation, and certainly not just those with anxious symptoms and depression. Got it. Thank you, Rob. Really appreciate that. Maybe just to clarify, I think the question I had on 501 and 502 was more around the expected formulation. Is that still going to continue to be a sublingual film in the case of 501? Have you considered other opportunities for a chronic presentation? What are you thinking about 502 and the combination therein? Chris, this is Vimal. Thank you for the question. In terms of the formulation for 501, we will continue exploring the depression studies with our sublingual thin film. More broadly, we continue to assess what are our lifecycle management strategies around 501R for our current three indications: schizophrenia, bipolar, dementia, and expansion opportunity. For depression study, it will be sublingual thin film. Regarding the 502, we have developed a formulation plan, and we will be initiating that plan because we do believe that by doing the formulation, we can improve certain properties in terms of the PK that will make it more from the perspective of chronic dementia agitation. That work is under progress. Got it. Okay. Thanks, Vimal. Maybe Rob, if you don't mind, maybe just pressing just slightly on the endpoint for the depression population, I guess. Is there any opportunity, do you think, to have a subset of, let's say, the HAM-D or something like that would focus more on insomnia and agitation, similar like we might have seen for, let's say, schizophrenia patients that present more with negative symptoms? There's precedent for that type of thing. There is. As Dr. Berman described, there's sort of a loose characterization of anxious symptoms in patients who are depressed. Most patients, on average, patients with depression do have symptoms of anxiety. They may be extremely prominent. In fact, they may meet diagnostic criteria for anxiety disorders. We will be testing over the course of our development program, the effect in patients, you can even say with and without anxiety, who may be depressed. Depending on the results of our data, it will help to guide us in terms of our development. We perhaps, for example, could stratify our randomization by the prominence of their anxiety symptoms. Chris, just to sum up, we have put in a briefing book with the FDA. We have a meeting upcoming. All of that will get clarified. Once we have clarity, we'll come back and present with what trials we are initiating, and we will have the buy-in of the regulators at that time. Yep. That's totally fair. I really appreciate it, and thanks for indulging me. Appreciate it. Thank you, Chris. Thank you. Our next question is coming from the line of Robyn Karnauskas with Truist. Please proceed with your question. Hi, guys. I reiterate with a great, very interesting talk today. Just two to start. Then I have a follow-up. On 5012, can you just talk a little bit about when we might learn more about that molecule? Since I think you noted that it hit the primary point of agitation, why was it not taken forward, and will you have to reformulate to get better IP? Those are questions on 5012. For 501 in depression, just for major depression, I think you've seen some of the recent data coming out of companies like Sage Therapeutics, where the placebo rates have been high, especially given COVID. It looks like you're considering 1 trial in combo. What are your thoughts on designing these studies once you get the MAD trials done given high placebo rates and also potential monotherapy, not just combination therapy? Thanks. Thank you, Robyn. This is Vimal. Thank you for the question. In terms of 5012, I think we will be in a position to disclose the target and embellish our overall strategy in terms of the formulation as well as clinical development. We are in the process of securing the IP. We are in the process of initiating the formulation work, and we want to make sure we have all of that secured to have the competitive advantage. As you know, there are other drugs being developed in the chronic dementia agitation space, and we want to be a leader in this space. BXCL501 gives us the advantage with the acute, and then 5012 can give us advantage to capture additional market space. That's kind of where we are, but we will keep you updated when we are in a position to outline more broadly our 5012 program. Far what we have is extremely exciting, and the reason it was not pursued by the other company was because it was in a secondary endpoint. It was not the primary endpoint. The primary endpoint was not met and compound went into a hiatus status, basically. Those were the reasons for not being pursued in that for agitation. Coming back to your depression questions, I'm going to pass it on to my team. Robert Risinger and Robert Berman can address your questions, how they are thinking about this program. Rob. Sure. We're aware that placebo response or response to placebo is increasing in general, and other companies have had difficulty with an increasing response to placebo, for example, you cited Sage. We know from the trials, at least with agitation and the evidence I just showed with respect to anxiety, we have a fairly potent, very high effect size. We know that that's going to help us in terms of separating from placebo. We have another couple of things that we're planning to integrate in our trial design and conduct that may help us mitigate placebo response. We have very high confidence, though, because the effects that we're seeing up to this point are really robust, at least with respect to anxiety. We also know that anxiety symptoms in patients with depression tend to have lower placebo response over time. I think I've seen several papers on that. I don't know, Rob, if you want to opine on this. Yeah, I think whenever you do a trial, you're focused on quality and placebo response. While Sage Therapeutics may have had a high rate, there are other examples of very reasonable rates with Axsome Therapeutics and Cadent Therapeutics, Novartis' recent data. It's not a foregone conclusion that you will experience that. I think by the time this comes to the clinic, I think COVID will be less dominant and hopefully that will change the dynamics too. That's all I would add. Just one quick follow-up. For depression, say your diagnosis is, you're using HAM-D. You talked about your three studies you did where you started to notice an impact on depression itself. A lot of those studies were single doses, right? How do you know, I guess, how often you need to dose, or you have RAD studies, but how often do you need to dose, how long that dose is going to last for, and when you might have to dose, whether it be at night just to prevent any somnolence whatsoever? What gives you the confidence that over time that depression will remain suppressed? Well, we have plans for our development trials to examine this very effect. We assume that dosing will be on a daily basis. We plan to test daily dosing. It's entirely possible that patients will not require daily dosing, but that is something that we can also discover over time. Let me make it clear that we are initially testing BXCL501 in addition to the SSRI, and we expect that that will continue for about a month or more. Provided we see those clinical improvements, that's when we can go back and consider longer-term studies or different dosing regimens, for example. We know that the dosing at this point is relatively safe and our only question now has to be how long does it last from one dose and how often do you need to dose BXCL501 to continue to see this sort of efficacy. Okay, great. Thank you, guys. Thank you, Robyn. Thank you. Our next question is coming from the line of Sumant Kulkarni with Canaccord. Please proceed with your question. Good afternoon. Thanks for all these informative things that you've included in your presentation today. I have questions along three lines. The first one is 501, the second 502, and the third would be a big picture one. On 501 for Dr. Berman and Risinger, how rapidly do you think a product like 501 must act to have a knock-on effect in depression, or does it not matter? What might that mean for trial recruitment and duration of clinical trial to get the best outcomes? Along those lines, in clinical practice, do you think anything must change for all of these associated symptoms of depression to be treated differently by clinicians? I can start on that. Sure. Sure, Rob. I think, as I kind of alluded to, if you see the rapidity in anxiety and depression as you saw in agitation in three models, I think it will be perceived by the patient and clinician as very rapid acting. Furthermore, it would be rapid acting and also would be given not just as needed, not PRN. I think it would have even a more noticeable benefit that way. In terms of the knock-on effect, we'll look at that in the data, but I would expect it to, if it works as my aspirations hope, to have a broader benefit within really the first week, just as seen with the benzodiazepines, but without the baggage of them. Then in terms of clinician uptake, clinicians still use benzodiazepines. They don't feel good about it, and they are turning away from it. I think it's a model that would resonate with them, being able to treat the anxiety rapidly, but without the baggage of the benzodiazepine. I think it would have broad appeal if it works as we hope it does. Rob, did you want to add? No, I think you covered it nicely. Got it. On 502, can you confirm that molecule is eligible for new chemical entity exclusivity in the U.S.? Hatch-Waxman safe harbor suggests you're free to develop it, but is there anything that might block commercialization of 502? It's not approved in U.S., so it's a new chemical entity. Got it. Thank you for that. Yep. Go ahead. My last question is more of a legal and philosophical one. As artificial intelligence evolves, how confident are you that intellectual property generated by AI won't be considered prior art to any AI engines used in the process? That's a great question, Suman. That's where Mike said that we may have 20 opportunities, and we give lot of emphasis in prioritizing opportunities where we can secure the IP. Unless we can secure the IP, we don't proceed forward. That is unmet medical need and IP become very paramount to our decision matrix. That's a great question you have, and also I would like to add, if you think of 501, it wasn't a first-degree connection. We identified second and third-degree connections, and that's where you can file the IP and you can secure the IP. What happened in 501, why we were able to get the patent we got, was that we got a composition of matter on the film, the way we constructed the film, it has been done for the first time. We are taking advantage of low-dose pharmacology. Third, our doses required to treat agitation for different indications, PK exposures are different, so we are protecting that as well, and then we file them mechanism of action to treat agitation. There are a lot of innovation needs to happen using this AI. It's not just a one-to-one connection that somebody knows and there won't be IP. It's quite a bit of art, and that's why we said we have created a AI-based ecosystem that integrates technology and our deep neuroscience expertise, combined with all the decision metrics that were discussed today to select a candidate, and 501 is a prime example of that now. Clearly, there's some real intelligence involved in as well. I'll jump back in the queue. Thanks. Thank you, Suman. Thank you. Our next question is coming from the line of Anita Doshi with Berenberg. Please proceed with your question. Hi. Good afternoon. Thanks for taking my questions. Great presentation this afternoon. Just one for me. You've used your AI platform to discover, I guess, 501 and 502 for these symptoms. Although these symptoms, I mean, the disorders seem seemingly distinct, they do have some underlying similar genes that result in this. In terms of deciding the strength and frequency of dose, what factors will you consider when doing that, especially when the symptoms seem to sort of overlap in these psychiatric states? Anita, this is Vimal. Thank you for the question. As you know, there is a shared human biology between these behaviors, and that's where the understanding and delineating these pathways, circuits, targets, and drug comes in, and seeing what is the right combination and what would be the right dose to use to treat that kind of a particular symptom. Your question is very valid, and as you continue to see, even with 501, initially we started with agitation. Now using our own proprietary clinical data, we have been able to take 501 into a new direction, which is anxious depression, which was not originally part of the plan. It's an iterative process, and once we have our first gateway indication identified and we move our platform forward, then we will identify additional opportunities. If I can add also, as Mike mentioned, one of the things that we strive for early on in our programs is to come up with biomarkers. The biomarkers really serve to underlie the systems that we're impacting, which point us in the directions where we want to go in terms of the symptoms. When we have that all set up, that helps us in making decisions as to where we go. Great. Thank you. I just had a follow-up. When you were initially looking at 501, was this one of the relationships with the anxiety? Did that show up? Was that one of the suspects and that kind of confirmed with the clinical studies? Well, I think anytime you have a molecule like 501 that calms, its effects on anxiety are possibly around the corner. We just weren't focusing on it. We were hard and fast focused on agitation because we just felt that that was the area that we wanted to go after. All of this really came out of the SERENITY I and II studies, which verified that 501 was doing more than what we thought. Great. That's helpful. Thank you. Operator, there is a question on the chat room. We're going to take that question. The question is: Are we still planning to partner ex-U.S. on these indications? Thank you for asking the question. We are planning to partner. Our focus initially is in Europe, as we have no development and commercialization capability outside the U.S. Our answer is, yes, we are continuing on that path. As we are moving along with the regulators in Europe, we will seek a partner for either development or commercialization in Europe. Operator, we can go back to the analyst questions. Thank you. Our next question is coming from the line of Graig Suvannavejh with Goldman Sachs. Please proceed with your question. Thanks. Good afternoon. Couple of questions if I could. One, just on BXCL502, I realize that it's early days, but do you have a sense of the timelines in which you'd be able to tell us a little bit more about the exact mechanism of action or approximate timeline? That's my first question. My second question, just in terms of the interest in expanding into depression for BXCL501, and I'm just wondering if you already, based on maybe any preclinical modeling that you've done, have a sense of what that dosing looks like as you go into your multiple ascending dose and whether it's similar to what you've seen in dementia or any of the other indications that you've gone into. Third, just as I think about the depression studies as well, any early thoughts on how you can perhaps on a go-forward basis, think about designing whatever the next efficacy trials are in terms of stratifying patients to maximize the potential efficacy responses that you might see, just given the high placebo rates that we often see in depression studies. Thanks. Thank you, Greg. This is Vimal. I will take the first question and then I will pass it on to Rob for the depression questions. In terms of the timeline, as we indicated for 502, we discovered this agent. We are in the process of initiating the formulation. Once formulation work has been completed and we have secured the IP, and we do see that we have a clear plan how we're going to take forward, and we don't see any competitive risk and not losing our advantage because it's in our mechanism. We will provide more clarity in due time that what our plan is and when it will go to the clinic. Please stay tuned. It has been identified, and it's a very exciting opportunity as we continue to explore as a monotherapy for treatment of agitation in dementia. Also we are considering how it can be in a combination, if the need is there with 501. Please stay tuned. We will keep you updated on that. Regarding the depression question, I will pass it on to Rob. Sure. In terms of our depression dosing and dose ranging, the trial that examined initial dose ranging for agitation in schizophrenia actually demonstrated very low dose signals that led us to believe that we were starting to see efficacy in agitation for patients with schizophrenia. Now, having much broader data in more than 800 patients now, we have a better bead on the dose range that we'll be testing in the MAD study. We prefer to do those dose ranging studies to enable the proof of concept trial very rapidly. We're very interested in establishing that dose range and showing a dose that may be, for example, intolerable, too high, and produce, let's say, sleep that is much longer lasting than you would necessarily want on an outpatient basis. Let me just address your question with respect to the placebo response. We will be stratifying in the trials to look at anxiety symptoms. We will not have, for example, an imbalance of patients experiencing anxiety symptoms and a variety of depressive symptoms. We ought to be better able to detect the signal, therefore. It sort of de-risks, in fact, demonstrating an effect on depression as well as getting at the effects on anxiety, we can get to that endpoint very quickly and come to some agreement with the FDA. Thank you. We'll move on to our next question, which is coming from the line of Yatin Suneja with Guggenheim. Please proceed with your question. Yes. Hi. This is Eddy in for Yatin. Thanks for taking the question. For Rob, in the opioid study agitation data that you showed, it looked like repeated dosing the effects started to wane by day 10, but then you noted in the rat study there was no tolerance effect. I'm just wondering what was driving that effect of the opioid study, and would you consider an intermittent dosing scheme in that situation? Well, I'll try to be brief in answering. Yes, we might consider a different dosing scheme, but the reason I believe that it appeared in the plot to drop off is because we had a lot of patients drop out who were on placebo. As we announced earlier, the placebo dropout rate was very high in the later days five through eight and 10, and as a result, it appears, I think it may be artificial, but it does appear that loss of response that you're talking about, I think is artificial. Got you. Thanks. Then just one final one. I think you noted that in the dex study with the human sleep patterns, it was an infusion of dex, and I was wondering if they know what the exposure levels that were needed there to sort of improve the sleep architecture in that dex study. Thanks. You can look up the paper. You can see exactly the concentrations achieved for any particular duration of N3 sleep. Do you think you can get to those exposure levels with your sublingual formulation? Absolutely. Great. Thanks, guys. Thank you. Our next question is coming from the line of Ram Selvaraju with H.C. Wainwright & Co. Please proceed with your question. Hi. Thanks very much for taking my question. There are multiple lines here. I'm just going to focus on three things. Firstly, when you look at the potential for the impact of these compounds on anxiety to spill over into treatment of sleep disorders, how are you thinking about that in the context of the overall commercial relevance and the commercial strategy? Is sleep really an area that you want to get into? What are some of the advantages and disadvantages of having compounds that, as they impact anxiety, the pathways underlying anxiety also may actually have important function in the regulation of sleep, given the nature of sleep as a commercial indication? Maybe I can make it very clear. We're not interested in sleep as a particular indication. We have found that sleep is associated with depression, both the two-edged sword I talked about, but the point is that dexmedetomidine not only has anxiolytic effects. All right. Thank you. Thanks. Okay. Sorry. Sorry, Ram. We had a technical glitch, Ram. Ram, please go ahead. Sure. I'm not sure where we dropped off, I'll say we're not interested in sleep indications per se. Our evidence shows a reduction in anxiety. We also know that dexmedetomidine improves N3 stages of sleep. We know that SSRIs make sleep worse. Benzodiazepines, although they can put people to sleep, they do not improve the architecture of sleep. We have parallel streams of evidence suggesting that BXCL501 plus SSRIs may actually improve depression. Whether or not it improves sleep, we believe depression is the primary endpoint. Does that help? Great. That's very, very helpful and very, very clear, and in fact, it's basically what I was hoping you would say, because frankly, sleep is a fragmented enough indication, and you're probably better off sticking to areas where the impact of these compounds can be more clearly delineated anyway. The second question is around optimization of formulation and routes of delivery for these compounds. I just wanted to clarify whether you are thinking about delivering all of these using the film-based technology that has been optimized in the case of BXCL501, or if you're still exploring the possibility of delivering the combination approaches via multiple routes of administration. If so, what might some of the alternatives to the film-based delivery system be? We are not thinking of that like that we have to deliver everything through the film. Film was designed for treating agitation for a specific purpose, Ram. Depending on what our indication is and what is the optimal route of delivery for that indication, we're going to take into account all of that and decide what delivery mechanism makes sense, and also, as we always indicated, even if it's NCE and we are improving its properties with the formulation, we want to make sure it provides some advantage on the intellectual property also. There are multiple considerations. What is the optimal delivery route, what other things we can do with the molecule, and what makes sense. We are not tied to only a sublingual film formulation. Okay, the last one is really a commercial strategy question. Clearly, as we look at areas like depression, there are lots of people who have been commercially active in this area for a lengthy period of time. I'm just wondering how you're thinking about sales force construction and the extent to which you want to be in full control of sales and marketing going forward in an area like depression, when there are so many existing sales and marketing infrastructures that already are in place. If you think that you might be able to co-opt those via some sort of co-promotion strategy that would effectively allow you to short-circuit the commercial process and not have to spend so much time building your own proprietary sales presence in some of these indications beyond acute agitation. Thank you. Thank you, Ram. Those are all very good thoughts. They are in our mind as well. The first and the foremost is focus on schizophrenia and bipolar, where we are close, four months away, or less than four months away from our PDUFA date. We are planning for that. It will be as we initiate our phase III trial in dementia, what our strategy will be for dementia as we get closer to NDA in dementia, and then will be followed up with depression. As you can see that these are kind of indication, neuropsychiatric symptoms indication, and depending on whether we do it ourselves and how much we do it ourselves and what partner can bring, everything will be explored to make sure that we have the optimal strategy for launch of these products. We want to be an innovation company, so we want to be able to continue to innovate and bring more products. As you saw today, our artificial intelligence platform continues to generate new opportunities, and Frank showed a map. We have number of opportunities that we continue to explore. We will make that balance of innovation and commercialization as we grow as a company. We are just at the intersection of getting first approval right now. Thank you. Ladies and gentlemen, we have reached the end of our time for the question and answer session, so I'd like to pass the floor over to Vimal for any additional closing remarks at this time. Thank you, operator. As you have heard, our proprietary AI platform, emerging neuroscience pipeline candidates, and expansion opportunities for our lead program, BXCL501, point toward exciting journey for our company in the years ahead. We believe we have a compelling value proposition that gives us a powerful advantage in the market in discovering and developing transformative medicine. We look forward to continuing to share updates on our progress with you, and thank you for your continued interest in BioXcel Therapeutics. Have a great day.