Los Angeles—Artificial intelligence (AI)-based drug discovery continues to tout accelerated timelines and novel and effective treatments, yet today’s clinical success rates remain at a low 10%.
On Wednesday, Recursion announced favorable efficacy data in the Phase II trial for REC-994, the company’s lead AI-derived candidate to treat the potentially fatal brain disease, cerebral cavernous malformation (CCM), in a late-breaking presentation at the International Stroke Conference (ISC) held in Los Angeles.
CCM is characterized by collections of small blood vessels in the brain that become enlarged and irregular in shape. These vessels can alter blood flow and place patients at risk for seizures, headaches, progressive neurological deficits, and potentially fatal hemorrhagic stroke. Currently, CCM treatment is limited to non-pharmacological options, such as surgery and radiation therapy.
According to the results of the 62-person Phase II study, 50% of patients with cerebral lesions on the highest dose of REC-994 (400 mg) showed reduction in total lesion volume compared to 28% of patients on the placebo after 12 months of treatment. Patients on the lower dose of REC-994 (200 mg) had similar changes in lesion volume compared to placebo. Similar trends were seen in patients with brainstem lesions, a population with high unmet need as cavernomas located in the brainstem are not amenable to surgical intervention.
Additionally, the modified Rankin Scale, which is widely recognized and approved by the U.S. Food and Drug Administration (FDA) to assess functional outcomes in acute stroke trials, indicated positive trends in patients on the 400 mg arm.
“My co-investigators and I are encouraged by these initial findings and we look forward to continued work with Recursion on the REC-994 program,” said Jan-Karl Burkhardt, MD, division head, cerebrovascular surgery at University of Pennsylvania and principal investigator of the study.
CCM is caused by gene mutations that impact redox homeostasis. Specifically, loss-of-function mutations activate pathways associated with downstream elevation in reactive oxygen species (ROS). REC-994 is a redox-cycling nitroxide compound and free-radical scavenger that is hypothesized to treat CCMs through a reduction in ROS.
Najat Khan, PhD, chief R&D officer and chief commercial officer at Recursion, emphasized that these data are particularly relevant as CCM lesion volume and CCM brainstem are among the top predictors of re-hemorrhage risk.
“These preliminary results show promising MRI-based and functional outcome signals, and we look forward to continued discussions with the FDA, and the CCM scientific and patient communities on next steps,” Khan told GEN.
CCM impacts approximately 360,000 symptomatic individuals in the U.S. and EU. Connie Lee, PsyD, CEO of the patient advocacy group, the Alliance to Cure Cavernous Malformation, and Amy Akers, PhD, the organization’s CSO, concurred that Recursion’s phase II data show “tremendous progress in the research space and provide hope for our community.”
“With no approved therapy and brain surgery as an option for only a subset of the population, CCM patients frequently face stepwise decline from the oozing and growth of cavernous malformation lesions. We look forward to working with Recursion and regulatory agencies to continue this momentum on our path to better treatments,” Lee told GEN.
Lagging indicator of modern AI
Last September, Recursion announced that REC-994 had met its primary endpoint of safety and tolerability, a readout that failed to impress investors. With REC-994’s encouraging efficacy data in hand, experts are mulling over the new data as they reflect on whether drug discovery’s new AI paradigm is starting to fulfill its promise.
Krish Ramadurai, PhD, partner at AIX Ventures, focuses on the intersection of healthcare, life sciences, and AI. He states that the first wave of AI-driven drug candidates has fueled both optimism and skepticism, given past “breakthroughs” that failed to solve drug development’s translatability crisis.
“Positive efficacy data for REC-994 would illustrate that AI-generated hypotheses can yield real patient benefits—evidence that could accelerate investment in AI-driven platforms, as oftentimes, many platform technologies have failed to develop a pipeline of clinically relevant therapies with accelerated asset development timelines,” Ramadurai told GEN.
While promising efficacy data provides positive proof points for AI pipelines, today’s clinical readouts are lagging indicators of the technology.
Dylan Reid leads life science investments as managing director at Zetta Venture Partners, an early-stage venture firm focused on AI native startups. Reid said the field of machine learning has come a long way since REC-994 was discovered.
“We take the pace of AI innovation for granted in most domains, but the feedback loops are long when it comes to drug development. The early signals look strong, but it will be a couple of years before we see the impact of modern AI in the clinic,” Reid told GEN.
At Recursion, modern AI continues to play out through new drug candidates and platform expansion.
Last October, REC-1245, Recursion’s first program to emerge from its end-to-end AI pipeline, received IND approval to tackle biomarker-enriched solid tumors and lymphoma. The first patient was dosed in a Phase I/II trial during the fourth quarter of 2024. The trial is active and enrolling at five U.S. sites, with data set to be read out in the first quarter of 2026.
Additionally in November, Recursion completed its combination with U.K.-based Exscientia in one of the largest M&A events in the AI drug discovery field. 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.
First-in-disease challenge
Recursion CEO Chris Gibson, PhD, said REC-994 originated from his graduate school work at the University of Utah in the lab of Dean Li, MD, PhD. (Li is now executive vice president and president, Merck Research Labs.) Leveraging open-source software developed by Anne Carpenter, PhD, who is now senior director of the Imaging Platform at the Broad Institute, Gibson used AI-based imaging screens of healthy and diseased cells to identify REC-994.
“After that work, I said, ‘Well, let’s go start a company and try and do this for many other diseases.’ And after a couple of years of building the platform, we went back to the university and licensed the know-how around REC-994 as one of our first candidates,” Gibson told GEN Edge last September.
In contrast to other CCM drug candidates, such as Neurelis’ NRL-1049 and Ovid and Graviton’s OV-888/ GV101, Gibson said Recursion is the first company to do an institutionally backed clinical trial in CCM and claims his company is “several years ahead of any other kind of commercial opportunity.”
“The challenge of being first in disease is, it’s an exciting opportunity to make an impact for patients. It also means that there’s an increased burden to not only prove out the molecule, but the path itself towards the treatment,” Gibson continued.
At the core of Recursion’s AI-based discovery pipeline today is the Recursion Operating System (OS), a platform powered by the company’s 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.
As Recursion continues to expand its clinical pipeline, so has its focus on AI-based clinical development.
“[Clinical trials are] not only the most expensive and time-consuming part of drug discovery and development (~70% of the total cost to bring a drug to market) but also where 90% of drugs in development fail, with later-stage clinical trials having the highest failure rates,” Khan told GEN.
“Just as we are leveraging fit-for-purpose data, AI, and machine learning to drive unbiased, data-driven discovery into novel targets and for first-in-class and best-in-class new molecules, we’re using AI to design and simulate better clinical trial programs, accelerate enrollment, and strengthen evidence generation,” Khan continued.
To support this goal, Recursion has established data partnerships with Tempus and Helix, which gives the company access to hundreds of thousands of insights drawn from patient records, including whole exome and whole genome sequencing. Recursion has also partnered with Faro Health to leverage their software to better automate and optimize the protocol design of clinical trials.
Recursion’s clinical portfolio now includes 10 clinical and preclinical and 10 advanced discovery pipeline programs. Khan says the company is doubling down primarily in oncology and rare diseases. Additionally, Recursion has 10 active-partnered programs with top pharma companies including Roche, Bayer, Merck KGaA, and Sanofi across immunology, oncology, and neuroscience.
As AI-based drug discovery continues to accelerate, REC-994 is but one thread in a growing tapestry. As Ramadurai says: “The industry will need multiple consistent successes across therapeutic indications and clinical phases to establish AI-based pipelines as the new paradigm in biotech.”
Fay Lin, PhD, is senior editor for GEN Biotechnology.
