Artificial Intelligence Lights a Beacon to New Medicine for Neurofibromatosis type 2

Recursion Signs Global Licensing Agreement with the Ohio State Innovation Foundation to Develop REC-2282 to Treat Neurofibromatosis Type 2


Human cells used to model NF2 on Recursion’s platform

Today Recursion announced an exciting new milestone: the in-licensing of a clinical-stage drug candidate for Neurofibromatosis type 2 (NF2) that was identified using our artificial intelligence platform [1]. This brings the company to a total of two clinical-stage programs after Recursion announced a cleared IND and initiated Phase 1 studies in Cerebral Cavernous Malformation in the past 6 months [2]. The story of REC-2282 is one that speaks to our approach of using petabytes of high-dimensional data we generate in-house and advanced machine learning to access unique insights into the biology of potential new medicines and prioritize results in ways that others cannot.

NF2: A Devastating Rare Disease

NF2 is a rare tumor syndrome and debilitating medical condition affecting thousands of patients worldwide, and many of these patients are in their teens, twenties and thirties. Patients with the disease suffer from the growth of benign tumors in the brain, spinal cord, and sometimes throughout the body, producing severe clinical symptoms including deafness, paralysis and severe pain. Untreated, these tumors can lead to early mortality, and without an FDA-approved medicine, surgery (and in some cases radiation) is the only option for most patients. However, surgical treatment of NF2 is not always effective because tumors are difficult to access within the cranial cavity, and are sometimes too numerous to remove. The right new medicine could significantly improve the lives of patients with the condition.

While no medical treatments are currently approved for NF2, over a dozen experimental medicines are in the drug development pipeline and are being tested in the disease. Many of these potential new medicines are drugs that have been developed for other indications but are currently being tested in NF2 [3], an approach called repurposing. Repurposing has been an attractive method for rare diseases because previously studied medicines can benefit from a shorter path to marketing and reduced development costs. In addition, the FDA has further facilitated drug development for rare diseases such as NF2 through provisions of the Orphan Drug Act [4].

Recursion research associate Anna Adamson performs experiments in our labs

Uncovering a Rare Gem

In the founding years of Recursion, our team focused on repurposing known medicines in disease indications where no known treatments exist. To do this, we ran millions of experiments on our robots, testing thousands of known drug candidates across hundreds of human cellular disease models using our unique combination of automated cell biology and machine learning. Over the years, our data scientists have developed algorithms to identify the best potential drugs for dozens of diseases from the image-based data we generate most at Recursion. We have learned a lot from this data, allowing us to now move beyond just repurposing into the development of novel chemistry, and in some cases, generating useful predictions about both the target(s) and potential liabilities (e.g. hERG) of the new chemical entities we are exploring (but we’ll save that for another blog).

In Q2 of 2018, we tested thousands of drugs in a cellular model of NF2 using our platform, identifying a handful of potential drug candidates. While we tested molecules that others had evaluated in the disease previously (alongside additional molecules from our labs), only a small subset of the compounds tested in these studies reverted “diseased” NF2-deficient cells to resemble healthy “normal” cells with minimal side effects. One compound in particular showed significant promise: REC-2282. This compound fully reversed the disease features of cells, as assessed by our computer vision algorithms, before producing untoward “side effects” in cells.

To prioritize from among the hits with the best probability of producing safe and effective new treatments, our team first seeks to understand the therapeutic history of each molecule before advancing them for further study by our disease biology teams. Only after identifying it as a hit in NF2, did we learn that REC-2282 had been previously tested in tumor indications, including NF2. The compound had already been cleared by the FDA for testing in humans. In fact, the compound showed early signs of activity in NF2 before a preliminary clinical investigation in NF2 was stopped due to the closure of the company sponsoring the clinical studies.

Members of our team (from left: Kristen, Aaron, Joel, William) discuss new results beneath inspiring photos of rare disease patients in our Salt Lake City headquarters

Data Arbitrage and Opportunity

The fact that Recursion’s platform independently identified a compound already shown to have activity in NF2 was an important proof point for Recursion’s approach. The orthogonal validation of REC-2282 in NF2 by our team and others was a clear sign — a data arbitrage — to us that this molecule could be a highly effective treatment for patients with NF2. Further, a prior phase 1 study and the phase 0 investigation in NF2 meant there was the possibility that we could move quickly towards helping patients.

“There are many failure modes on the long and arduous road to new medicines but we will continue to leverage the latest and most advanced technology to reduce the time and cost it takes to bring new medicines to patients.”

Leading Indicators for a New Drug Discovery Future

With the culmination of this work we’re excited to announce that we entered into an exclusive licensing agreement with Ohio State University to develop REC-2282 for patients with this devastating disease. This unique opportunity to build on existing clinical data for REC-2282 to rapidly evaluate the safety and efficacy of this compound in NF2 patients speaks deeply to why Recursion was founded and why each of us on the team is so committed to our mission: decoding biology to radically improve lives. This opportunity enabled us to leverage AI to identify and potentially bring to patients a new medicine that had been sitting on the shelf waiting to be re-discovered.

There are many failure modes on the long and arduous road to new medicines but we will continue to leverage the latest and most advanced technology to reduce the time and cost it takes to bring new medicines to patients. These are still the first steps of our journey, and we remain a small team, however in just a few years we have built a company with two clinical-stage programs and dozens of promising discoveries across diverse therapeutic areas behind them. There remain thousands of untreated diseases, and for each of us at Recursion, that fact is unacceptable.