Laying claim to another AI 'first,' Exscientia tees up an I/O drug for the clinic
In January 2020, Exscientia announced that a drug molecule to treat OCD invented by AI was set to enter clinical trials for the first time. A little over a year later, its AI-designed molecule for immuno-oncology will do the same.
The A2a receptor antagonist was co-developed with Evotec and has the potential to prevent adenosine from binding to T-cell receptors, promoting anti-tumor T-cell activity. Exscientia CEO Andrew Hopkins said in a statement that the molecule was discovered within 8 months from project initiation.
That time frame was even faster than the company’s molecule for OCD treatment, which reached the point of entering clinical trials within just a year, which is the kind of boast that the AI companies love to make. In a January interview with The Telegraph, Hopkins hinted that Exscientia was inching toward trials of its second drug, and said that the shortcut was possible because the AI platform — named Centaur — is able to help narrow down the field of potential winners. Instead of testing hundreds or thousands of molecules, the algorithm zeroes in on those with potential for success.
Exscientia was not able to answer questions in time for this story’s publication Friday.
The announcement comes at a hot time for artificial intelligence platforms. The cost of drug development involves a debate that has rolled on for years, but the AI crowd like to focus on a $2.6 billion estimate to help back their case. Of course, we won’t know the true advantage until the first AI drug makes it through the clinic — and into the market.
In January, AstraZeneca added its first target generated by AI to its portfolio, after a collaboration with BenevolentAI that began in April 2019. That work focused on chronic kidney disease, and fits into the company’s broader AI strategy. AstraZeneca’s head of renal biosciences Pernille Hansen said that AI can be used in ways other than discovering new targets: in chemistry, imaging and beyond.
BenevolentAI’s COO Ivan Griffin told Endpoints News in January that his company has spent years feeding its tech with data from proteins, genes and results published in scientific journals in an effort to train its algorithm to make connections that scientists may not have noticed at first. Scientists take advantage of these AI-predicted relationships and then interrogate them to see if they hold up, Griffin said.
In February, Insilico founder Alex Zhavoronkov announced that it brought its first candidate into IND-enabling studies. His company’s goal is to launch an in-human trial later in 2021, though he wasn’t ready to announce what the target or experimental drug is. Its focus will be idiopathic pulmonary fibrosis, something that BenevolentAI and AstraZeneca have also said to be collaborating on.
Preclinical data from this project will be presented at the AACR annual meeting.