The artificial intelligence (AI) revolution continues to tout developing drugs faster and at lower cost, but are we there yet?
On October 2nd, the FDA cleared an investigational new drug (IND) application for a Phase I/II clinical trial of REC-1245, a new chemical entity for the treatment of biomarker-enriched solid tumors and lymphoma, developed by Recursion, the long-standing AI in drug discovery company founded in 2013.
Recursion states that REC-1245 is the first program using the company’s end-to-end AI platform, spanning discovery, biology, chemistry, and the clinic. According to Chris Gibson, CEO of Recursion, the potential therapeutic progressed from novel target biology to preclinical drug candidate in under 18 months, “nearly twice the speed of the industry average.”
REC-1245 is a selective degrader of RBM39, a new target, identified by Recursion’s AI-enabled Maps of Biology platform, with a similar biological function to the well-known but hard-to-drug target, CDK12. Preclinical data showed that RBM39 degradation induces splicing defects which downregulate DNA damage response (DDR) networks and cell cycle checkpoints, functional changes prevalent in cancer.
Many investigational drugs designed to target CDK12 also inhibit the structurally similar CDK13, resulting in toxicity. In contrast, targeting RBM39 showed similar effects to the loss of CDK12 without affecting CDK13.
In an interview with GEN Edge, Simon Barnett, research director at Dimension, stated that drug discovery timeline compression can be a leading indicator of machine learning’s (ML) capabilities. At the same time, proof points are just starting to come in, and “it’s still virtually impossible to say how ML-heavy discovery and development efforts are meaningfully warping phase transition likelihood.”
That said, the industry will first hear about faster timelines before seeing drugs enter Phase II, Phase III, and achieve approval from these platforms.
“In biotech, time is money, so it means something,” said Barnett. “It’s going to be one-off examples until we can corral many examples. We’ll then have some real firepower behind these statements of what data-driven methods can do in life sciences.”
REC-1245 joins the Salt Lake City-based company’s existing clinical pipeline, whose therapeutic areas span from precision oncology to rare disease. The Phase I/II clinical trial for REC-1245 is expected to initiate in Q4 2024. Recursion estimates that the initially addressable population for this potential therapeutic will be over 100,000 patients in the United States and EU5.
“RBM39 degraders may offer a promising therapeutic approach for patients with solid tumors, particularly those with limited treatment options,” said Najat Khan, PhD, chief R&D officer and chief commercial officer at Recursion. “This mechanism provides new opportunities for targeting tumors, which are often resistant to conventional treatments.”
Prior to joining Recursion in April, Khan played an integral role in building leading therapeutic pipelines as chief data science officer and global head of strategy and portfolio organization R&D at Johnson & Johnson. GEN Edge caught up with Khan to reflect on REC-1245’s IND approval, the evolution of Recursion’s AI-based drug discovery platform, and what’s next for the field.
Interconnectedness like a highway
While traditional drug discovery builds upon the literature of established targets, Recursion was founded on the mission to discover novel therapeutics by looking at the entirety of biological systems.
“What’s critical to understand is not just whether one gene is driving a disease, but all the interconnectedness like a highway. There’s beauty in understanding how perturbing one thing impacts the rest of the system,” said Khan.
At the core of Recursion’s technology is the Recursion Operating System (OS), a platform powered by the company’s large proprietary phenomic datasets, spanning
transcriptomics, proteomics, metabolomics, and more. The platform leverages Maps of Biology, which create millions of biological relationships by perturbing human cell lines to instill a disease state and testing compounds to reverse the cells to healthy function. Machine learning algorithms then evaluate gene-gene and gene-compound associations of interest to identify novel drug targets, such as RBM39.
“Translation gap is one of the biggest issues we have in the industry because we don’t have good models to predict human response,” Khan told GEN Edge. “How can we start with human-centric models that are cost-effective and standardized? That’s what phenomics allows us to do so that we’re able to anticipate and optimize before we go into the clinic.”
End-to-end platform
Over the past 18 months, Recursion has made strides in growing its end-to-end AI drug discovery platform.
“The reality is that to truly bring value in many cases to the industry, you can’t just optimize one of the hundred steps to discovering and developing a medicine. You have to optimize entire pathways of discovery and development,” said Gibson in an interview with GEN Edge.
In 2023, Recursion acquired Cyclica and Valence, whose technologies enhanced proteome-wide prediction of small molecule-target interactions and low-data learning in drug design respectively.
In one of the largest M&A events in the AI drug discovery field in August, Recursion entered a definitive agreement to combine with U.K.-based Exscientia. According to Recursion, the agreement combined Recursion’s scaled biology exploration for target identification with Exscientia’s precision chemistry for lead optimization for a complementary pipeline. The company expects approximately 10 clinical readouts over the next 18 months and is leveraging four large partnerships with Roche-Genentech, Bayer, Sanofi, and Merck KGaA. These growing industrialized workflows also promote speed in automation, thereby contributing to lower costs.
While these end-to-end platforms continue to build, the probability of success through the clinic in the drug discovery industry remains at approximately 10%. According to Gibson, the reason for this low statistic is “not because people aren’t good at science but because biology is really, really, really complex.”
“That’s the whole reason why we founded Recursion: To leverage sophisticated computational tools to disentangle these thousands of parameters of biology that no human can hold in their head,” Gibson told GEN Edge. “Our belief is that this approach is sound and that it will just take us time to be able to kind of get the flywheel moving and really start to show people.”
Khan concurs and highlights that “quality is more important than quantity.”
“I have done enough in science and development to know not everything works, but we’re going to focus on showing pivotal examples where we can make molecules, get them to the clinic, and benefit patients differently than anybody else can,” said Khan.
So are we there yet? We’re mapping out a fast-track road. While one-off drug candidates are not overturning the industry, they’re one step forward toward a new drug discovery paradigm.
*Najat Khan, PhD, will be the opening keynote speaker at GEN’s inaugural virtual event, The State of AI in Drug Discovery, streaming on October 30th.
