Recursion plans registrational trial for AI-repositioned REC-4881 after FAP data

Recursion released updated Phase Ib/II data on 2 November 2026 for REC-4881, an oral MEK1 and MEK2 inhibitor in familial adenomatous polyposis (FAP), and said it now plans its first registrational trial for the drug, with the second half of 2027 the timing it gave, subject to its talks with the FDA¹. The results from the TUPELO study were presented at the CGA-IGC annual meeting in Denver by Lisa Boardman of Mayo Clinic¹.

The AI connection lies in how the asset was repositioned. Takeda first developed the molecule as TAK-733 and used Recursion's AI platform to identify the areas of oncology where it might work; Recursion licensed it in 2020 for $1.5 million upfront, up to $39.5 million in milestones and low-to-mid single-digit royalties¹. Najat Khan, Recursion's president and chief executive, called the drug "the clearest proof point yet of what we built Recursion to do"¹.

What the trial found

Patients took 4mg once daily for 12 weeks, followed by 12 weeks off treatment¹. The headline figures were:

• 87% of patients (13 of 15) had a fall in total polyp burden after 12 weeks, with a median reduction of 47%, and about one in three reached reductions of 65% or more¹.
• In the upper GI tract, 92% (12 of 13) improved, with a 55% median reduction, and half (six of 12) moved down at least one Spigelman stage by Week 13¹.
• In the lower GI tract, nine of 11 patients with measurable burden (82%) improved, with a 50% median reduction¹.
• At Week 25, after 12 weeks without the drug, 85% of patients (11 of 13) were still below baseline¹.

Six of eight patients with polyps larger than 5 mm at baseline saw the count of those polyps fall, and four achieved reductions of at least 80%¹. The best responder went from 1,344 mm of total polyp burden to 32 mm, which Recursion itself said was not representative¹.

On safety, 21 of 26 patients (80.8%) had treatment-related adverse events that were Grade 1 or 2, and four (15.4%) had a Grade 3 event, with no Grade 4 or 5 events¹. The most common were acneiform dermatitis, rash, CPK increase, diarrhoea and asymptomatic LVEF decrease, in line with MEK inhibition¹. Four patients stopped treatment for Grade 1 or 2 events¹.

What it means for the market

For pharma R&D teams weighing AI platforms, REC-4881 is a case of AI-led indication finding for a shelved molecule rather than de novo design, and it is now the asset on which Recursion is building its registrational plans¹. FAP has no FDA-approved drug therapy, according to Bryson Katona, a TUPELO investigator at the University of Pennsylvania, and current care relies on repeated endoscopy and eventual surgical removal of the colon¹. Recursion counts more than 50,000 FAP patients in the US and EU5, although other estimates of prevalence vary widely¹.

Recursion will not have the field to itself. Biodexa has enrolled 100 patients in its Phase III Seranta trial of eRapa, an oral rapamycin tablet, with primary completion estimated for July 2030, and Tempest Therapeutics is running a Phase II study of TPST-1495¹.

Investors were unmoved: the shares rose 0.1% to $3.84 the following day¹. The data came with third-quarter results showing revenue of $18.552 million, against $5.175 million a year earlier, and a net loss of $119.036 million, down from $162.253 million¹.

The evidence remains small and open-label, with efficacy reported on 15 patients or fewer per measure¹. The study counted polyps only up to a top size category of more than 5 mm and did not record separate counts above 10 mm¹. Katona said longer trials will show how long the effect lasts¹. Recursion has said it will give more detail on the registrational trial later this year¹.