Exscientia starts trials of first AI-derived cancer immunotherapy

News
AI implementation for pharma and healthcare

Exscientia was the first company to start human trials of a new drug designed using artificial intelligence (AI) last year, and now says it has started testing of the first AI candidate for immuno-oncology. 

Oxford, UK-based Exscientia has been working with German biotech Evotec on the adenosine A2a receptor antagonist, which is being tested as a treatment for adult patients with advanced solid tumours.

The adenosine pathway has emerged as an attractive target for immuno-oncology R&D because it is thought to reduce the activity of misfunctioning regulatory T-cells that inhibit the immune response against cancers.

Tumour cells produce high levels of adenosine, which binds to the A2A receptor on T-cells and helps them escape immune system detection.

So far, the main interest in A2A has lain in combining it with other immuno-oncology drugs, in the hope of boosting activity and potentially sidestepping a resistance mechanism that allows some tumours to evade the effect of current drugs like the PD-1/PD-L1 inhibitors.

Exscientia is entering a category with some heavyweight competition, including Novartis, Johnson & Johnson/AdoRx and AstraZeneca as well as smaller players like Arcus, Corvus and iTeos Therapeutics.

However, none of the rival programmes have advanced too far in clinical development, and a trial of one regimen – the combination of AZ's A2A drug AZD4635 and a CD73 blocking agent called oleclumab – was abandoned last year after disappointing results in lung cancer.

Exscientia and Evotec think they have potentially a best-in-class candidate of course, based on a very high selectivity for the A2A receptor and low take-up into the central nervous system, which could reduce psychological side effects, according to the partners.

The A2A receptor is also present in the CNS and is a validated target for neurological disorders such as Parkinson's disease. One A2A antagonist drug, Acorda's tozadenant, was abandoned in late-stage development in 2017 after five deaths in its clinical trial programme related to severely low white blood cells (agranulocytosis) and sepsis.

Exscientia said it had taken the A2A drug from concept to being clinical trial-ready in just eight months, topping the 12-month development of its first clinical candidate for obsessive-compulsive disorder (OCD) DSP-1181.

Typically, taking a compound from discovery to the clinic takes around five years, but deploying Exscientia's AI platform – Centaur Chemist – truncates that by narrowing down the hunt for suitable molecules so that much fewer have to be tested in the lab.

"Immuno-oncology medicines are bringing benefit to a range of cancer patients," said Exscientia's chief executive Andrew Hopkins.

"Our selective A2a receptor antagonist addresses a next-generation immuno-oncology strategy to empower the human immune system by reversing the effects of high adenosine concentrations," he added.

Exscientia will lead the clinical development of the A2A drug, with Evotec retaining co-ownership rights.

Momentum is building behind the use of AI in drug discovery, as the first candidates start to filter through into drugmaker's pipelines. Earlier this year, for example, AZ's two-year-old alliance with AI specialist BenevolentAI generated its first development candidate for chronic kidney disease.