Exscientia plc (Exscientia) operates as an artificial intelligence-driven precision medicine company. The company is committed to discovering, designing and developing the best possible drugs in the fastest and most effective manner.
The company’s pipeline demonstrates its ability to rapidly translate scientific concepts and patient-centric data into precision-designed therapeutic candidates. The company has built an end-to-end solution of artificial intelligence, or AI, and experimental techno...
Exscientia plc (Exscientia) operates as an artificial intelligence-driven precision medicine company. The company is committed to discovering, designing and developing the best possible drugs in the fastest and most effective manner.
The company’s pipeline demonstrates its ability to rapidly translate scientific concepts and patient-centric data into precision-designed therapeutic candidates. The company has built an end-to-end solution of artificial intelligence, or AI, and experimental technologies for target identification, drug candidate design, disease relevant translational models and patient selection. These integrated technologies allow the company to discover, design and develop precision medicines. The company’s platform has enabled it to design candidate drug molecules that have progressed into clinical trials, as well as to prospectively provide patients with potentially more applicable drug therapies through AI guided assessment. The company’s patient-first AI process consists of the following four elements:
Precision Target: Using patient tissue and deep learning approaches to identify new targets;
Precision Design: An extensive platform of AI technologies to design innovative drugs;
Precision Experiment: Tech-enabled precision experimentation to derive better data; and
Precision Medicine: Advanced patient selection to improve clinical success rates.
The company’s AI-design capabilities include a wide range of deep learning and machine learning algorithms, generative methods, active learning and natural language processing. These methods are used to guide target selection, to design the precise molecular architecture of potential drug molecules and to analyse patient tissues to prioritise the molecules that are likely to provide the best response for an individual’s specific tumour.
The company’s pipeline includes a broad range of programmes across therapeutics areas primarily in oncology and inflammation & immunology (I&I). The company’s pipeline candidates are differentiated through precision design and personalised medicine, and it has five development candidates that are either in clinical trials or IND-enabling studies across oncology and I&I. Exscientia has between 50-100% ownership of four of these candidates and one of these programmes is eligible for milestones and royalties from a partner. In total, the company has over 10 programmes with 50-100% ownership in the pipeline and over 20 partnered programmes with substantial economics.
The company originated the first ever AI-designed precision drug candidates to enter human clinical trials and it expects to have four compounds in the clinic by 2024. Exscientia's partner DSP has also taken into the clinic two additional compounds designed by Exscientia as part of its early design as a service partnership.
The company began the first Phase 1 clinical trial of EXS21546, its A2A receptor antagonist, in December 2020 and reported topline data from this healthy volunteer study in June 2022. The company initiated a Phase 1/2 study towards the end of 2022 in patients with relapsed/refractory non-small cell lung cancer (NSCLC) and renal cell carcinoma (RCC), with the first patient expected to be enrolled in the first half of 2023. In this trial the company is also prospectively studying its complex patient selection biomarker for patients more likely to respond to EXS21546. The company also intends to dose the first patient in the Phase 1/2 clinical trial for GTAEXS617, its CDK7 inhibitor partnered with GT Apeiron, in the first half of 2023.
The company’s partner, BMS has also initiated a Phase 1 clinical trial in February 2023 for EXS4318. a PKC theta inhibitor that BMS in-licensed in 2021, with potential in inflammation and immunology.
The company has two additional programmes in IND-enabling studies, EXS74539, its LSD1 inhibitor and EXS73565, its MALT1 inhibitor. Both compounds are fully owned by Exscientia and in GLP-toxicity studies with potential in haematology and oncology.
The company’s development candidates were generated in an average of approximately one year from the first novel designs, demonstrating its consistent speed and efficiency. The company has also pioneered the first clinically validated AI-driven platform to improve treatment outcomes for cancer patients prospectively. In the first-ever prospective interventional study of its kind, the company’s AI platform predicted which therapy would be most effective for late-stage haematological cancer patients based on drug activity in their own tissue samples, with measurements taken at single-cell resolution.
Pipeline
EXS21546 (A2A antagonist): Phase 1/2, immuno-oncology, majority owned. A2A is an attractive target, but existing design approaches have either suffered from complex pharmacology due to lack of selectivity or side-effects caused by low penetration. The company’s clinical candidate was identified within nine months of generating novel designs and after testing only 163 compounds. Further, recruiting patients susceptible to A2A antagonism has been challenging as no robust biomarker for adenosine-driven immunosuppression exists.
The company’s candidate, EXS21546, has demonstrated exceptional selectivity brain penetration while restoring immune activity and demonstrating single agent activity similar to PD-1 inhibitors in vivo. The compound reversed A2A receptor mediated immune suppression and also exhibited cancer cell killing activity in primary tissue samples of pancreatic and lung cancer patients. In a pre-clinical study, EXS21546 demonstrated comparable single agent anti-tumour activity to an approved anti-PD-1.
In a Phase 1 healthy volunteer study, topline data confirmed Exscientia’s target product profile design, including potency, high receptor selectivity and expected low brain exposure with no CNS adverse events reported. A Phase 1/2 study, IGNITE, in relapsed/refractory RCC and NSCLC patients has been initiated and the company expects to start enrolling patients in the first half of 2023.
The company presented data at AACR 2022 and ESMO-IO 2022 highlighting its work on this novel biomarker and it will be prospectively testing its potential in its ongoing IGNITE study.
GTAEXS617 (CDK7 inhibitor): Preclinical, oncology, joint venture with GT Apeiron Therapeutics, or GTA. CDK7 presents an opportunity to improve treatment outcomes over CDK4/6 inhibitors due to CDK7’s dual role in cell cycle and transcription. Previous development efforts have exhibited side effects, possibly due either to a covalent binding mechanism of action or poor oral absorption. The company’s selective, non-covalent candidate meets multiple criteria. The company is able to identify a molecule meeting all of its design criteria after testing just 136 compounds. The potential drug candidate also has favourable oral bioavailability of 77%, and critically, it demonstrates a significantly reduced interaction with a key efflux transporter compared to other CDK7 candidates in development.
Further, using its precision medicine platform, the company is working to define activity in more than six various solid tumour indications, and has described initial efforts to identify patient selection biomarkers, as well well as validation and discovery of pharmacodynamic (PD) markers to be used and confirmed along side its Phase 1/2 trial to evaluate GTAEXS617 for the treatment of ovarian cancer, which the company expects to commence in the first half of 2023.
The company’s precision medicine platform is determining responding and non-responding patients for further biomarker discovery where proof of concept selection in ovarian cancer using a gene signature has already correlated as has validation of PD biomarkers.
EXS4318 (BMS, PKC-theta inhibitor): Phase 1, immunology, in-licensed by BMS. PKC-theta is an attractive immune modulating drug target; however, several large pharma companies have failed to design a small molecule with the required potency and selectivity against other closely related kinases. The company’s platform designed a highly potent, highly selective next-generation immunomodulatory drug candidate within 11 months at the start of the design process, which was the 150th molecule synthesised. The company’s balanced candidate has demonstrated high on-target activity while maintaining high selectivity and favourable tolerability.
In February 2023, BMS announced that EXS4318 has entered a Phase 1 clinical trial in the United States.
EXS74539 (LSD1 inhibitor): IND-enabling, oncology and haematology, wholly-owned by Exscientia. EXS74539 ('539) is the first potent, selective, reversible and brain-penetrant LSD1 inhibitor lysine demethylase 1 (LSD1) inhibitor with potential in both haematology and oncology. LSD1 demethylates histones which play a critical role in regulating the expression of genes which suppress differentiation and drive the proliferation and survival of a number of tumour types.
EXS73565 (MALT1 protease inhibitor): IND-enabling, haematology, wholly-owned by Exscientia. EXS73565 ('565) is a potent and selective mucosa-associated lymphoid tissue lymphoma translocation protein 1 (MALT1) protease inhibitor with potential applications in haematology. MALT1 is a protease crucial for activation of the NF-kappaB pathway which supports the uncontrolled proliferation of malignant T- and B-cells in haematological cancers.
Exscientia’s precision design approach was able to optimise the safety profile for agents targeting MALT1 whilst also generating potency and selectivity. Scaffolds of other MALT1 inhibitors in the clinic significantly inhibit UGT1A1, an enzyme involved in the metabolism of bilirubin, often leading to dose-limiting toxicities in the clinic.
The company has more than 20 additional ongoing projects that range from target profiling to lead optimisation and it is continually initiating new projects across its business models. The company also developed additional drug candidates as part of its pilot design as a service (DaaS) programme with Sumitomo Dainippon Pharma, including DSP-0038, in Phase 1 studies.
Patient-first AI Strategy
The key elements of the company’s strategy are to encode and automate to transform every stage of the drug discovery process; scale by automating interactions and autonomous decision making; leverage robotics and automation to transform cycle times and efficiency; drive its technology stack through cutting-edge science and technology; build out biologics capabilities; design patient centric drug candidates with an improved probability of success; use AI and technology to design precision drug candidates that have the potential to be safe and effective; use patient-centric disease relevant translation models to maximise the probability of clinical success; create proprietary single cell functional and multi-omics datasets from primary patient tissues to map disease target networks for use across the target and drug discovery and development value chain; scale pipeline and operations; focus its in-house programmes on oncology and immunology and anti-virals; leverage partnerships to rapidly expand the portfolio and maximise platform value; execute on a global strategy for discovery, development and commercialisation; and be as innovative in the clinic as it has been is discovery.
Next Generation AI Platform and Approach
The company’s AI-powered technology is designed to generate, analyse, prioritise and optimise small molecules to ultimately engineer novel precision drugs and find the right patients for these drugs. The company transforms the drug discovery process from a lengthy, inefficient and somewhat random screening process to an automated learning and creation process.
This enables the company to create new innovative medicines faster, by leveraging its end-to-end platform its AI algorithms are involved in the design of virtually every compound it makes and tests; its patient tissue AI translational screening platforms have demonstrated improved clinical outcomes for oncology patients; it generates differentiated proprietary data, which strengthens the predictive power of its models; and its design AI intelligently applies these data to generate novel molecules and optimise across design parameters in parallel, right from the start.
Technology Platform
The company has designed a comprehensive learning system with AI at its heart, encompassing precision target generation, precision molecular design and extensive precision experimental laboratory-based testing. The company has also built an AI-enabled approach to precision medicine that combines high resolution single-cell analyses of tissue material from individual patients with multi-omics single cell data.
Precision Target
Patient Driven Target Discovery: The company’s patient tissue screening platform can be used to discover and validate targets by testing the activity of compounds with known target space in primary tissue samples, and then overlaying this with transcript and protein-protein interaction data to ‘triangulate’ for critical target nodes.
In an ongoing project, the company characterised the activity of 80 approved small molecule drugs in a set of samples from 20 ovarian cancer patients.
Side by side comparison between drug responses obtained with the company’s primary ex vivo model and publicly available cell line data (CANCERRXGENE.ORG).
Centaur Biologist: Centaur Biologist is the company’s AI-driven, automated target discovery platform technology that applies deep learning to genome-scale datasets to identify connections and predict target-disease associations well ahead of publication in the scientific literature. The company uses Centaur Biologist to evaluate and prioritise incoming proposals for collaborative drug discovery; seed its own internal drug discovery efforts; and identify potential biomarkers for use in selecting patients and obtaining early signs of clinical activity.
As of December 31, 2022, the company had successfully applied it to target identification efforts in a wide range of areas, including oncology, immuno-oncology, immunology and rare diseases, supporting business development activities and initiating joint ventures. In addition, in partnership with the Gates Foundation, the company is using Centaur Biologist to identify targets for diseases of particular importance to the developing world. In its collaboration with Sanofi, the comopany is using Centaur Biologist to identify targets across oncology and immunology.
Precision Experiment
The company’s platform combines the latest advances in high content confocal microscopy, proprietary deep learning image analysis, high throughput and scalable next generation sequencing, scalable cloud computing to interrogate the activity of small molecules and other therapies directly in diverse primary patient tissues at the single-cell level. The company has developed its high content translational platform to overcome the unique challenges associated with working with primary tissues.
Patient-Tissue Biobank for Translational Models: The company’s experimental process is compatible with diverse tissue types and tumour indications, including blood (leukaemias), lymph nodes (lymphoma), and solid tumour indications (tissues such as malignant pleural effusions and ascites and solid tissue samples). The company has both biological and clinical validation data for the majority of haematological cancers (including acute and chronic leukaemias, T- and B-cell non-Hodgkin lymphomas and multiple myeloma) and biological validation for solid tumour indications, such as lung, ovarian, breast and pancreatic cancer, renal cell carcinoma.
The company has also developed proprietary methods to quantify cell-to-cell interaction propensities as a measure of immunomodulation (e.g., to determine the likelihood of antigen presenting cells to interact with T-cells in complex primary tissues); and it is working to develop label free and unsupervised approaches for classifying phenotypic diversity in primary tissues.
Biosensor fragment screening to support project initiation and progression. Biosensor assays, using surface plasmon resonance, or SPR, allow direct biophysical measurement of compound binding and its kinetics. The company uses biosensor assays in its drug discovery process to create seed datasets of low molecular weight fragment compounds for further optimisation by generative design. Biosensor assays have an extremely wide dynamic range and can measure compounds with binding equilibria from pico-Molar to milli-Molar and are ideal for fragment screening.
The company’s biosensor assays are AI enabled, with SensAI, a machine learning method that automates the analysis and classification of SPR experiments. This extends the company’s machine learning to SPR data and is a key component to scaling its fragment screening and kinetic analysis capacity.
Structural Biology Supports 3D Design: The company’s AI algorithms directly exploit data from structural biology, whenever it is available, to build hypotheses for generative design and to map the binding site using Hot Spot analysis. The company employs techniques, such as X-Ray crystallography and cryo-EM to provide comprehensive 3D structural information about the atomic interactions of a molecule. The company has built in-house structural biology capabilities for X-ray crystallography and cryo-EM that also make use of high-throughput synchrotron beamlines and Cryo-electron microscopes.
GPCR Biased-Signalling Pathways for Next-Generation Pharmacology: The company has established a comprehensive platform of GPCR techniques that comprise molecular pharmacology, AI design, biophysical screening and structural biology, which it calls TRUPATH. This approach, which was invented by the company’s head of GPCR Pharmacology, is pathway agnostic and thus allows it to measure GPCR- ligand activities at the single-transducer level. The company can use the approach to elucidate transduction networks and define biased signalling pathways that can refine target activity to mitigate off-target effects. It allows for high resolution at the pathway level to provide the most accurate measures of receptor occupancy and activity in a signalling assay. The company’s system, optimised within Exscientia, expands the technology to support both live-cell and protein-based configurations that are routinely applied in drug discovery projects. This allows the company to link GPCRs to previously undocumented or silent signalling networks even in the absence of ligands that drive a coupling event. The company’s TRUPATH platform provides unique value to GPCR drug-discovery programmes.
The company’s multi-parametric drug design algorithms are well-suited to exploit these complex GPCR readouts which are a spectrum profile across 20 signalling pathways rather than a single endpoint.
Molecular and Cellular Pharmacology to Develop Primary Project Assays: The company’s laboratories are fully equipped to address the challenges of enabling a large-scale discovery portfolio and focus on the full integration of experiment with AI.
To support its internal discovery portfolio, the company incorporates a comprehensive range of biochemical screening technologies. In addition, the company has extensive experience with physiologically relevant cellular screens covering immuno-oncology, oncology, inflammation and immunology.
Precision Design
The company’s philosophy is that every molecule should be designed by an algorithm. The company unlocks the creativity of AI through the use of reinforcement learning, deep learning and genetic algorithms to intelligently design and select novel compounds that meet its design objectives.
Centaur Chemist: Centaur Chemist is the core drug design component of the company’s platform. It is a sophisticated combination of AI technologies and high-precision models that allow the company to predict and utilise over 2,500 human biological endpoints in parallel to meet critical design objectives. Centaur Chemist can exploit diverse data, including three-dimensional protein structures, high content images and pharmacology data, creating predictive models to evaluate the multitude of drug properties in parallel. Satisfying composite design goals to create drug candidates with balanced properties in the most efficient molecular structure is a significant competitive advantage of its AI design.
Objective Setting and Project Telemetry: The company’s AI platform allows it to design compounds that meet multiparameter optimisation (MPOs) within a small number of design cycles. Using experimentally determined data the company measures how well its compounds meet these objectives using its MERIT scoring system.
Model Platform: The company’s system predicts over 2,500 human endpoints automatically, from PK to off- target effects and it generates models which allow the Centaur platform to create and evaluate the suitability of the drug candidate molecules. These models are delivered through the company’s Model Platform that allows the company to ingest data from a wide variety of sources and automatically build multiple models using advanced machine learning techniques, such as multi-task deep learning methods. In addition, the company has built models on endpoints as diverse as in vivo behaviour and high content cellular imaging, through to the more routine, biochemical or cellular screening data.
Hotspots: This is a sophisticated statistical technology developed by the company’s Head of Structural Bioinformatics that creates a detailed map of the surface of a protein. It describes the shape of any pockets present within a protein and the specific locations of the key features that determine ligand binding such as H-Bond donors, acceptors and apolar areas.
Druggability Assessment: This allows the company to make a precise assessment of the druggability of any given protein or collection of proteins.
Generative design. Core to the company’s AI approach is the belief that learning systems are superior to brute force in finding the optimal and most elegant solution to high dimensional problems. The company has developed a suite of AI-design algorithms that generate novel chemical structures, allowing it to virtually navigate the vastness of chemical space in an efficient and intelligent manner.
Ongoing Platform Expansion
The company’s new biologics platform fits within its end to end AI driven drug discovery and development platform. The company is building an automated laboratory with proprietary hardware to enable integration of AI design with high-throughput biologics profiling under development. By sequencing paired human antibody data. Additionally, the company has already demonstrated the potential of its existing precision medicine patient tissue models to analyse novel antibodies.
The company is designing and building what will be the world’s most advanced automated drug discovery experimental platform. This new platform will be driven by the company’s autonomous drug design algorithms and be a fully automated, AI-driven Design-Make-Test closed loop experimental engine.
Precision Medicine
The company’s functional precision oncology platform embeds patient based translational assays throughout its AI-driven drug discovery process.
Immuno-oncology candidate EXS21546, an A2A antagonist: The company assessed its clinical stage A2A antagonist, EXS21546, in primary tissue samples from patients across a range of solid tumour indications. The company has developed a multi-gene signature, that correlates with adenosine concentration in the tumour micro-environment and can be controlled by EXS21546, that it refers to as the ABS. Ongoing work is continuing for patient selection and pharmacodynamic biomarker discovery is ongoing using this platform, and the signature is being validated observationally alongside the ongoing IGNITE study.
Small Molecule CDK7 Inhibitor, GTAEXS617, in Primary Malignant Ascites of Ovarian Cancer Patients: The company completed a preclinical study evaluating the activity of '617, its CDK7 inhibitor that it developed, in primary tissue samples of ovarian, lung and pancreatic cancer patients.
Personalised Precision Oncology: The company has developed the first-ever functional precision oncology platform to successfully guide treatment selection and improve patient outcome in a prospective interventional clinical study.
Combination Screening
The company completed proof-of-concept work to identify efficacious drug combinations with ibrutinib for patients with chronic lymphocytic leukaemia. This work yielded encouraging results that were published in Nature.
Collaborations
Pharma Partners
Bristol Myers Squibb (BMS): The company and BMS are collaborating on a portfolio of multiple targets in oncology, autoimmunity, immunology and inflammation. The partnership began in 2019 with Celgene, and it expanded in 2021 directly with BMS following their acquisition of Celgene, with increasingly rewarding terms for Exscientia. BMS provides invaluable therapeutic area expertise, as well as a commitment to fund the development of molecules through the clinic. The second deal, coming just two years after the first, demonstrated the power of the company’s platform to successfully deliver high quality drug candidates to BMS’s exacting preclinical candidate criteria. Together, these deals have already delivered $65 million in upfront payments.
Sanofi: Under its initial collaboration with Sanofi, the company delivered a bispecific lead molecule in the area of inflammation and fibrosis in 2019. In January 2022, the company entered into a new collaboration with Sanofi, pursuant to which it will use its AI-driven, end-to-end integrated platform to discover and validate up to 15 novel targets in the oncology and immunology therapeutic areas. The company is collaborating with Sanofi to advance certain of these targets into small molecule inhibitor drug research projects and accelerating the identification of certain small molecule development candidates.
Co-Owned Collaborations
Rallybio: In 2019, the company entered into a co-development and co-ownership joint venture with Rallybio to investigate multiple areas of rare disease. Under this joint venture, the company jointly select targets after assessment by its AI platform for biological pathway relevance and chemical druggability risks. The company is driving the programme through completion of pre-clinical toxicology studies, and then Rallybio will progress the candidates through clinical trials and commercialisation, if any candidates are approved. The company also has the option to explore molecules in non-rare disease indications, such as oncology. The partnership has delivered its first discovery candidates on a challenging target, ENPP1.
EQRx: In 2021, the company entered into a co-development agreement with EQRx, to accelerate the design of multiple best-in-class molecules for many therapeutic indications, including oncology. The company has initiated three early discovery projects in the areas of inflammation and immunity and oncology.
GT Apeiron Therapeutics: GTA was launched in 2019 by GT Healthcare, one of the company’s investors, and they immediately signed a deal with it to fund the discovery of novel checkpoint inhibitor compounds for the treatment of multiple oncology indications. The first candidate has been designed, selected and entered into IND enabling toxicology studies.
Blue Oak: Blue Oak Pharmaceuticals, Inc, or Blue Oak, is a biotech focused on the discovery of transformative CNS drugs. In late 2020, the company signed a deal with them to co-discover and develop new medicines to treat brain disorders. This partnership combines Blue Oak’s CNS expertise and the company’s ability to design novel CNS penetrant chemotypes and demonstrated ability to apply its AI technology to the successful design of bispecific small molecules.
Patents
As of December 31, 2022, the company owned and co-owned three issued U.S. patents, five pending U.S. patent applications, and over 50 pending foreign patent applications including unpublished foreign priority applications.
These patents and patent applications fall into nine different patent families across 16 different jurisdictions worldwide. The company generally relies upon trade secret protections for the company’s AI technology platform as the platform includes hundreds of algorithms and more than 2,500 predictive models. From time to time, the company files patent applications directed to aspects of its platform technologies. The company owns a patent family which includes a granted U.S. patent, a granted European patent, a granted Indian patent, one pending U.S. continuation patent application, and a foreign patent application pending in Europe with claims covering certain aspects of the company’s platform, which, if issued, are expected to expire in 2030, excluding any patent term adjustment or patent term extension. The company also owns a pending priority patent application with claims directed to aspects of its platform, which, if issued, is expected to expire in 2043, excluding any patent term adjustment or patent term extension alongside three international patent applications which are expected to expire in 2041 and 2042 respectively, excluding any patent term adjustment or patent term extension.
With regards to patent protection on the molecules the company designs, it either solely owns such filings, jointly own filings with its partners, or in some instances its partners solely own the patent filings. For example, the company owns and co-owns four patent families directed to its novel compounds which include one pending U.S. patent application, three international patent applications, two provisional U.S. applications, give U.K./European priority applications and 16 foreign patent applications pending in such jurisdictions as Australia, Canada, China, Europe and Japan, which if issued, are expected to expire between 2039 and 2042.
Research and Development
The company’s research and development expenses for the year ended December 31, 2022, included £128.9million.
Trademarks
As of December 31, 2022, the company’s trademark portfolio included 77 trademark registrations or active trademark applications worldwide. Such portfolio includes 70 non-U.S. trademark registrations, 11 pending non-U.S. trademark applications and 5 pending U.S. trademark applications.
Government Regulations
The company is subject to the U.S. Foreign Corrupt Practices Act of 1977, as amended, or the FCPA, the U.S. domestic bribery statute contained in 18 U.S.C. § 201, the U.S. Travel Act, the USA PATRIOT Act, the UK Bribery Act 2010 and the UK Proceeds of Crime Act 2002 and possibly other state and national anti-bribery and anti-money laundering laws in countries in which it conducts activities, collectively, Anti-Corruption Laws. Whether or not the company obtains the U.S. Food and Drug Administration (FDA), the Medicines and Healthcare products Regulatory Agency (MHRA), or EMA approval for a product, it must obtain the requisite approvals from regulatory authorities in foreign countries prior to the commencement of clinical studies or marketing of the product in those countries.
History
Exscientia plc was founded in 2012. The company was incorporated under the laws of England and Wales in 2021.
The Analyst Price Target shows the analysts’ low, high, and average target at a glance.
