On January 30, 2020, the British AI pharmaceutical company announced with a high profile that the first molecule designed by artificial intelligence entered Phase I clinical trials.
The drug, called DSP-1181, is a long-acting serotonin 5-HT1A receptor agonist for the treatment of obsessive-compulsive disorder (OCD) and was developed by Exscientia in collaboration with Japan’s Sumitomo Pharma.
With the title of “world’s first case”, Exscientia has received extraordinary attention. Exscientia claims that the entire project took less than a year from conception to molecule determination, while the industry average is 4.5 years.
However, it was recently discovered that the development of DSP-1181 has been stopped by Sumitomo Pharmaceutical because the clinical phase I study did not meet the expected standards.
At present, whether on the official website of Exscientia or Sumitomo Pharmaceutical, this clinical pipeline has quietly disappeared.
Sumitomo was satisfied with the development process
Japan’s Sumitomo Pharmaceuticals reached a cooperation with Exscientia as early as 2014, and was one of the first pharmaceutical companies to open cooperation with AI companies.
Sumitomo Pharmaceutical was once very satisfied that DSP-1181 could enter the clinic, saying that the innovative method it uses will make a strong contribution to central nervous system drugs.
Sumitomo values two technologies of Exscientia most: one is the technology of automatically generating compounds, and these compounds automatically synthesized by AI take into account the feasibility and novelty of synthesis;
The second is a knowledge-based artificial intelligence prediction model that predicts target pharmacological activity, toxicological effects on the target, and partial characteristics of pharmacokinetics;
The combination of the two enables Sumitomo to predict the pharmacological activity and pharmacokinetic profiles of a large number of virtual compounds generated by automated structure generation.
Between 2014 and 2017, Sumitomo Pharma and Exscientia conducted exploratory research for up to 3.5 years, mainly in the monoamine G-protein-coupled receptor (GPCR) drug for the treatment of psychiatric disorders.
The two parties then launched a project to discover a drug with 5-HT1A receptor agonist activity, from which DSP-1181 was born.
Sumitomo said that when they started the project, they were very surprised by Exscientia’s artificial intelligence technology in drug discovery.
Generally, in the early stages of exploratory research, it is not easy to design compounds that both target the target protein and have desirable pharmacokinetic properties as drugs for the treatment of mental disorders. But when Sumitomo synthesized and evaluated the compound originally proposed by Exscientia, it showed activity and good pharmacokinetic profile on the target protein egg, thus obtaining the lead compound early in the project.
Sumitomo and Exscientia then started a “bi-weekly cycle” mode, in which Sumitomo’s chemical team synthesized compounds proposed by Exscientia, the pharmacology team evaluated these compounds, and the two companies shared activity data together and continued to improve drug.
Thanks to the above model, the two parties have tested and synthesized as many as 350 compounds in less than a year, and DSP-1181 is the 350th compound synthesized since the beginning of the project.
In addition to synthetic compounds, analogs synthesized during the course of the project are also very important. Sumitomo’s chemists synthesized intermediates of the compounds proposed by the AI, and also designed and synthesized some compounds with putative pharmacological data, and fed these data into Exscientia’s predictive models.
These include compounds that provide important structure-activity relationships for optimized compound structures, which further accelerates the drug discovery cycle and allows the company to discover DSP-1181 in a short period of time.
Novelty questioned
Todd Wills of the American Chemical Abstracts Service (CAS) once conducted a detailed analysis of the novelty of three drugs that Exscientia entered into the clinic. He found that all three drugs faced the problem of insufficient novelty.
First, the 5HT1A receptor that DSP-1181 acts on is a very important and classic target of antipsychotic drugs, so the development of this drug has not departed from the original target.
Wills analyzed the DSP-1181 series of patents and found that these molecules are very similar to haloperidol, a typical antipsychotic drug approved by the FDA in 1967. So Exscientia is likely to be optimized on a long-discovered molecular backbone.
Of the 350 molecules synthesized and analyzed during the discovery of DSP-1181, not only were many molecules similar to haloperidol, but there were also molecules that were similar in shape to 28 other FDA-approved drugs, including lamotrigine, an antiepileptic drug. Medications and mood stabilizers, occasionally used in OCD treatment), etc.
DSP-0038 is another drug that Exscientia and Sumitomo collaborated on and is currently in Phase I clinical trials. DSP-0038 is a dual-targeted 5-HT1a receptor agonist and 5-HT2a receptor antagonist for the treatment of Alzheimer’s disease.
Analysis of 194 example molecules with published bioactivity data included in the patents revealed that the shapes of these molecules lack structural diversity, as 78% of the example molecules have the same shape as FDA-approved drugs, and 93% of the example molecules have The same shape as the three molecules mentioned in its claims.
At present, in the pipeline announced by Sumitomo Pharmaceutical, only DSP-0038 is the drug in cooperation with Exscientia, which is undergoing clinical phase I in the United States, but the two sides have not disclosed too much information.
Clearly, although more and more drugs are being discovered with the help of these innovative technologies, AI-discovered compounds do not guarantee the success of clinical trials.
AI still has a long way to go to cross the “valley of death” of drug development.
