If two’s company and three’s a crowd, then 30-plus is more like 10 crowds—and like a crowded elevator with that many people inside, a drug pipeline top-heavy with top candidates risks not carrying any of them to the top, Oxford-based artificial intelligence (AI) drug pioneer Exscientia reasoned recently.
So this past fall, Exscientia pruned its pipeline down to four active drug candidates in therapeutic areas that include inflammatory disease, hematology, and most notably, oncology. And as the company heads into 2024, the message from Exscentia is that it will work to translate that prioritization into an R&D effort that generates the positive data needed to bring new candidates across the proverbial finish line of regulatory approvals and into the market.
The four priority programs include two clinical candidates partnered with other companies, GTAEX617 (with GT Apeiron) and EXS4138 (with Bristol-Myers Squibb; BMS), and two wholly-owned preclinical candidates, EXS74539 and EXS73565, at least one of which is expected to enter the clinic in 2024 as well.
The refocusing will emphasize what Exscientia considers the most differentiated, highest-value oncology targets within its portfolio.
“It’s very much about doubling down and ensuring we focus on precision oncology,” Andrew Hopkins, DPhil, Exscientia’s founder and CEO, told GEN Edge. “We want to create a pipeline of highly differentiated molecules where the hallmark of a future Exscientia drug is, we can clearly define, clear blue water, between the properties of an Exscientia drug and the other molecules in its field.”
The four priority programs were all designed using Exscientia’s AI platform, which uses patient tissue data in order to define optimal profiles for research, improve experimental assessment during design, and improve outcomes. In addition to data acquisition, the platform integrates machine learning and generative systems to discover new drugs through precision engineering at molecular scale.
Seeking partner or licensee
Exscientia’s tighter focus on precision oncology prompted the company to end internal development of its first candidate to reach the clinic, EXS21546, an A2A receptor antagonist that was co-invented and had been developed with Evotec for oncology indications. ‘546 had been under study in the Phase I/II IGNITE-AI trial (NCT05920408), assessing the A2A receptor antagonist plus a PD-1 inhibitor in up to 110 patients with immunotherapy relapsed or refractory renal cell carcinoma (RCC) and non-small cell lung cancer (NSCLC).
That trial has been terminated, but ‘546 is not dead; Exscientia is seeking to partner or out-license ‘546 after concluding it would be challenging for the program to reach the prolonged high level of target coverage it would need plus a validated patient selection strategy.
Why did Exscientia place ‘546 on the back burner and opt to streamline its pipeline?
Hopkins said, “It was very much [a matter of] looking at what’s coming down our pipeline, particularly with the focus on oncology and some of the molecules we’re excited by, which then led us to think about not just looking at this program such as A2A, but actually look at it in the context of everything else that we’re developing. That prioritization then led us to think about, OK, what’s got the highest chance of differentiating in the marketplace? Where is all the evidence pointed that we could show a strong signal within a patient population? What’s the external experience that people are having with these mechanisms?”
Exscientia insists it still believes in targeting A2A as a mechanism for a successful oncology drug.
“We think [‘546] is actually best developed in combination with a PD-1 immuno-oncology agents,” Hopkins said. “The hypothesis we were exploring was the potential of A2A to reignite the immune system in PD-1 refractory patients, and see whether we could bring a PD-1 activity back to them. So, I think probably one of the logical potential partners would be someone with a PD-1.”
Sparking excitement
One program Hopkins said excites Exscientia is EXS74539, a preclinical-stage selective and reversible lysine specific demethylase 1 (LSD1) inhibitor that the company is developing as a treatment for acute myeloid leukemia (AML) as well as small-cell lung cancer (SCLC). Exscientia plans to advance ‘539 into clinical study in 2024, following an IND submission to the FDA which is expected in the first quarter, to be followed with a Phase I trial in healthy volunteers set to launch in the first half of the new year.
The company hopes the study’s data could improve efficiency for developing ‘539 in multiple indications and in combination with other therapies.
“Our molecule really stands clear and apart in being non-covalent and in having a shorter half-life,” Hopkins said. “It allows us then to have really clear ability to control the dosing schedule, which is one of the key safety issues with this class: How would you ensure that you can have control of the effect it might have on platelets?”
Effective control of dosing, he said, enables Exscientia to consider less frequent or “exquisite” dosing, including periods of no-dosing “holidays.”
“We’ve done studies where we look at different dosing regimens and it allows us then to understand where we can maximize safety and still have the impact on the cancer models that we required,” Hopkins said. “[‘539’s] ability then for a physician to control the dosing schedule a lot more, we think, is going to be an important element and differentiate the safety of risk.”
Another advantage Exscientia cites for it LSD1 inhibitor is its ability to penetrate the CNS—no small advantage given that about 40 to 50% of SCLC patients develop a brain metastasis within two years of diagnosis. That CNS penetration plus a reversible mechanism of action position ‘539 into a potential best- as well as first-in-class drug, according to Hopkins.
Improving first-line potential
Exscientia researchers highlighted promising preclinical data for ‘539 at the European Society for Medical Oncology (ESMO) Congress 2023, held in Madrid. Using a disease model system and a deep learning-enabled high-content imaging and image analysis platform, Exscientia showed that ‘539 induced AML cell differentiation marker expression ex vivo. Researchers also showed that ‘539 improved first-line potential of standard-of-care therapy, both in primary AML samples and in non-transformed healthy bone marrow or peripheral blood mononuclear cells—and demonstrated that an adapted drug regimen successfully limited the thrombocytopenic potential of ‘539.
“Our preclinical data demonstrated ex vivo efficacy of ‘539 against AML blast cells and supported the combinatorial potential of ‘539 with first line clinical AML treatment strategies,” the researchers concluded in an abstract of their poster presentation at ESMO. “Leveraging the reversibility of ‘539 allowed the design of adapted drug regimens, to preserve the safety profile of this inhibitor on non-transformed healthy cells.”
LSD-1’s value as a target was further validated when Merck & Co. acquired LSD1 inhibitor developer Imago Biosciences for $1.35 billion, a deal completed in January 2023. Through that acquisition, Merck expanded its hematology portfolio by adding Imago’s therapies designed to treat myeloproliferative neoplasms (MPNs) and other bone marrow diseases.
‘539 is one of two wholly owned precision oncology candidates in Exscientia’s streamlined priority pipeline. The other is EXS73565, a mucosa-associated lymphoid tissue lymphoma translocation protein 1 (MALT1) inhibitor being developed for potential indications in hematology. Exscientia has said ‘565 is progressing through IND/clinical trial agreement (CTA)-enabling studies, with a further update planned in the first half of 2024.
At ESMO, Exscientia also presented preclinical data for ‘565 showing a combination of the drug plus Imbruvica® (ibrutinib), marketed by Johnson & Johnson’s Janssen Biotech and AbbVie’s Pharmacyclics to be differentiated from competitors with potentially lower toxicity through low predicted risks of drug-drug interaction and hyperbilirubinemia.
‘565 avoided inhibition of UGT1A1, an enzyme involved in metabolizing bilirubin and certain small molecule drugs, in contrast to other clinical-stage MALT1 inhibitors. As a result, Exscientia researchers concluded that the potential of MALT1 inhibition demonstrated by ‘565 could underpin a safe and effective monotherapy and/or drug to be combined with other treatments.
“We certainly think this is a molecule that will perform best in potential combination studies. And to think about how that is best done, potentially, with partners is a key part of our current thinking as well,” Hopkins said.
Relying on partners
Exscientia is relying on partners to develop its two furthest-advanced priority pipeline candidates, both of which are in the clinic. The company is partnering with GT Apeiron on GTAEXS617, a cyclin dependent kinase 7 (CDK7) inhibitor being developed for a variety of cancer indications. ‘617 has been designed for improved potency, selectivity, and pharmacokinetics compared to other CDK7 inhibitors in development.
‘617 is under study in the model-driven, adaptive Phase I/II ELUCIDATE trial (NCT05985655), which as of the company’s most recent update in November continued to enroll patients with advanced solid tumors, including NSCLC, plus head and neck, pancreatic, HR+/HER2- breast, colorectal, and ovarian cancers.
‘617 is being studied as monotherapy and in combination with standard of care, where Exscientia asserts that its precision medicine platform is expected to play a critical role in determining the best combinations. The company has not said when it plans to release data from ELUCIDATE.
“We don’t expect it to take any longer than a conventional Phase I/II, and potentially could be shorter, actually,” Hopkins said.
Which of the six cancer indications have priority as Exscientia and Apeiron decide how to develop ‘617?
“We are letting the data decide that,” Hopkins replied. “We’ve used our patient centric platform to help us prioritize a top six that we identified, and we are letting the trial run, and determining from the signal which would be the priority indications we would go forward in further studies.”
Together with BMS, Exscientia is partnering on EXS4318, a potentially first-in-class selective protein kinase C (PKC) theta inhibitor designed by the company that is being developed for immunology and inflammation (I&I) indications. A first-in-human study of ‘4318 was launched in February 2023.
Launched in 2019
‘4318 is the leading program to emerge from a small molecule collaboration launched by Celgene in 2019 when it paid Exscientia $25 million upfront plus undisclosed payments tied to achieving milestones, and tiered royalties on net product sales. Later that year, BMS completed its $74 billion acquisition of Celgene.
Exscientia and BMS expanded the collaboration two years later to I&I as well as oncology candidates. In return, BMS multiplied what it agreed to pay Exscientia to potentially more than $1.3 billion in clinical, regulatory, and commercial payments—including up to $50 million upfront, up to $125 million in “near to mid-term” milestones, and tiered royalties on net sales.
BMS now oversees the clinical and commercial development of ‘4318, with Exscientia eligible for undisclosed milestones and tiered royalties from the pharma giant. BMS opted to in-license the PKC theta inhibitor in August 2021, triggering a $20 million milestone payment to Exscientia.
BMS is one of three biopharma giants with which Exscientia has collaborations in place. In January 2022, Sanofi committed up to $5.2 billion to partner with Exscientia on developing up to 15 small molecule candidates across oncology and immunology/inflammation indications based on Exscientia’s AI platform. Exscientia received a $4 million milestone payment (of a potential $343 million per program) in September after identifying the first target, in immunology/inflammation.
Also in September, Exscientia launched a collaboration with Merck KGaA to discover novel small molecule drug candidates across oncology and immunology. Merck KGaA agreed to pay Exscientia $20 million upfront and undisclosed discovery, development, regulatory, and sales-based milestone payments.
Exscientia has another two clinical candidates targeting serotonin (5-HT) it designed that are in Phase I development by Sumitomo Phama, which holds all development and commercial rights to them—DSP-0038, a 5-HT1A agonist plus 5-HT2A antagonist; and DSP-2342, a dual 5-HT2A/ 5-HT7 antagonist.
Following a decade of designing small molecule drugs, Exscientia last year expanded its AI-based platform to begin designing precision-engineered and optimized, fully human biologics, starting with novel antibodies.
Integrating AI and automation
To further facilitate the platform expansion, Exscientia opened its automation laboratory in Milton Park, Oxfordshire, during the third quarter. The facility’s 28,000 square feet of new lab space is designed to enable Exscientia to integrate generative AI and automation to drive faster, high-quality experimentation. The company says it is building its own hardware and software solutions to automate a variety of experimental laboratory processes including chemical synthesis, as well as biochemical and biophysical screening.
“We believe the next generation of productivity enhancement is going to be by the integration of AI with automation—how our design cycles then can control the experiments we do. So we’ve been very keen to automate our experimental work for some time,” Hopkins said.
“What we’re interested in here is how do we collect data on a large number of different assay types at the same time rather than conventionally, where we think of high throughput screening, as testing, say, a million compounds against one assay,” Hopkins added. “We’re actually much more interested in asking the question: how could we collect 100 assay points against 20 compounds and have a diversity so we can build up a deep matrix and understanding of all of our designed molecules?”
Exscientia’s automation effort, Hopkins said, also includes chemical services, long a major bottleneck and hurdle in small molecule drug discovery: “We’re at the stage now where AI can really help us understand how to design using generative methods into a synthetically accessible space that we believe now is possible with advanced automated synthesis.”
Exscientia reasons that the next frontier of drug discovery productivity will come by integrating AI with real-world experimentation with automation.
“So many companies buy automation solutions off the shelf. When we looked at what our requirements were, our biologists actually wanted to innovate how we do it. And that means then creating truly integrated multidisciplinary teams,” Hopkins explained. “Software engineers, AI engineers, hardware engineers, as well as our scientists all working together in one interdisciplinary team to actually automate our processes. We are incredibly excited by what we’re building in combination of working with suppliers to build some of these bespoke systems, but also even in some cases building our own hardware.”
Partnerships and recognition
To build additional partnerships, especially those focused on fighting cancer Exscientia has hired Parker Moss to join the company as executive vice president, corporate development. In that new position, Moss will lead efforts to create and execute a corporate development strategy intended to create and strengthen new market opportunities for Exscientia’s precision medicine, clinical, and tech innovation platforms, with a special focus on oncology.
“We think of the AI ecosystem for pharma, and our goal is to become the partner of choice in the AI ecosystem,” Hopkins said.
Moss has overseen strategic partnerships in biopharma and the broader healthcare and technology sector as chief partnerships officer of Genomics England, a company owned by the U.K.’s Department of Health & Social Care. After partnering with the National Health Service (NHS) to deliver the 100,000 Genomes Project, Genomics England now works to enable academic and biopharma research, as well as support providers of whole genome sequencing diagnostics in cancer and rare disease.
Moss has also been co-lead of Genomics England’s research program in cancer and rare disease. Before joining Genomics England, Moss was entrepreneur in residence for two Fidelity-backed venture capital funds, F-Prime and Eight Roads, where he led an investment in, and joined the executive team of, AI biotech company Owkin.
“He’s got a strong background in oncology. Parker brings a wealth of experience now to really help us in developing and commercializing our tech platforms and our precision medicine platforms,” Hopkins said. “We mutually found each other.”
Hopkins—who discussed his career and the founding of Exscientia last year as a guest on GEN’s “Close to the Edge” video interview series—recently found himself recognized in rarefied company, after making both Fortune magazine’s inaugural Fortune 50 AI Innovators list, and Time magazine’s list of “The 100 Most Influential People in AI 202 3.”
Time’s list includes 23 chief executives of top companies and firms in the space, including Marc Andreessen, co-founder of Silicon Valley venture capital firm Andreessen Horowitz (a16z); and Jensen Huang, CEO, president and co-founder of Nvidia, which has focused its AI presence in life sciences.
“It is fantastic recognition for the company to be recognized amongst that incredible list of individuals. To be listed amongst Elon Musk, Sam Altman, and Demis Hassabis [CEO and co-founder, Google DeepMind] is an incredible and humbling honor to have,” Hopkins said. “I think it’s a recognition that Exscientia has played the leading role in developing AI for drug discovery.”
Alex Philippidis is senior business editor of GEN.