Artificial Intelligence (AI) The drug development process has yet to be transformed, but some efforts appear more promising than others.
On Monday, cancer research giant Memorial Sloan Kettering and life sciences AI pioneer Absci announced a first-of-its-kind partnership to discover six novel cancer therapies using Generative AIpromising to bring new drugs to clinical trials next year.
Also: How AI hallucinations could help create life-saving antibiotics
MSK will identify the target in the cells to be pursued and Absci will use its genomic artificial intelligence to create a Again antibody that will bind to that target.
“This is the first collaboration of this kind that Absci has done, especially with an institute like MSK,” Sean McClain, founder and CEO of Absci, said in an interview with ZDNET. “It provides a really great synergy: the knowledge and expertise that MSK has in oncology, and these new targets that they’re going to introduce, with Absci’s ability to design interesting drug candidates with our AI platform.”
“If MSK brings in a new GPCR, our platform is well suited for that, to be able to find an antibody that matches that target,” which McClain calls “helping to create the biology.”
Absci
MSK’s approval represents a significant vote of confidence for the young world of AI in the life sciences.
“We’re always looking for new ways to move things forward for patients around the world, and AI is clearly an area where we need to get involved,” said Gregory Raskin, MD, senior vice president of Technology Development at MSK.
Also: Connecting generative AI to medical data improved utility for physicians
“We’ve never before partnered to create new antibody drugs with a company that’s focused on AI,” he said. “We look forward to being not just a leader in cancer treatment, but a leader in AI for cancer treatment at MSK.”
The collaboration is described as a “co-development” arrangement and a “50/50 partnership,” with both parties funding the initiative, though funding amounts were not disclosed.
Talks between MSK and Absci began at the JP Morgan Healthcare conference in San Francisco in June, Raskin said, and have evolved over the subsequent seven months.
The division of labor involves MSK proposing a target, which the two parties will then discuss and agree upon, and Absci coming up with a design for an antibody, or series of antibodies, against the target.
In addition to the computer simulations that Absci and its own lab facilities can run, MSK will help with “core facilities and scientists at our institution who are world experts in determining whether a drug will be able to defeat a tumor and be safe,” Raskin said.
“Once we have identified the target, we will use generative AI models to design antibodies toward those targets to achieve the biology,” McClain said.
Also: How GenAI has greatly improved in solving medical questions thanks to RAG
“MSK has more than 100 research labs staffed by people working in oncology, looking for new cancer targets,” Raskin said. Thirteen drugs that have been approved by the U.S. Food and Drug Administration were invented at MSK, he noted, including Danyelza, a treatment for pediatric neuroblastoma, in 2020, and Erleada, a treatment for nonmetastatic castration-resistant prostate cancer, in 2018.
“With mouse antibodies, this is a traditional method that is very time-consuming and labor-intensive,” Raskin noted. “You may end up with a bad batch of antibodies that don’t bind well to the target.”
“We hope that this method will reach our patients more quickly: that is the key.”
Sloan Kettering Memorial
MSK has its own patient population that it can use to test drugs that might emerge from the partnership, Raskin told ZDNET. The hospital conducts about 1,800 clinical trials, some for third parties and others for drugs developed in-house.
“We have the ability to write our own INDs,” he said, “and we can initiate trials in our own patients with these technologies,” referring to the “Investigational New Drug” submissions required to file with the U.S. Food and Drug Administration, which is responsible for approving clinical trials and ultimately accepting or rejecting drugs.
The appeal of AI, Raskin said, is the technology’s potential to speed up drug development, which takes, on average, a decade. By using generative AI, new drugs can be rapidly conceived and simulated on the computer, in some cases shaving years off the typical process of in vitro chemistry and in vivo animal testing.
“With mouse antibodies, this is a traditional method that is very time-consuming and labor-intensive,” Raskin noted. “You may end up with a bad batch of antibodies that don’t bind well to the target.”
“We hope that this method will reach our patients more quickly: that is the key.”
Also: Generative AI could lower drug prices. Here’s how
For Absci’s first partnership with a hospital to develop new drug targets, McClain said, the important thing is “being able to bring the drugs we develop together to patients at MSK, and being able to have real help from MSK on the translational side, making sure we have the right clinical strategy.”
Absci already has partnerships with several pharmaceutical giants, including AstraZeneca, Almirall and Merck, and with artificial intelligence chip giant Nvidia.
The collaboration with MSK is different in that the institute is a nonprofit, rather than a for-profit operation of Big Pharma, McClain said. Because commercializing therapies can be huge, Absci and Memorial Sloan Kettering plan to bring on board a pharmaceutical partner to ultimately commercialize any drug, McClain said, ideally after demonstrating a “proof of concept” on their own.
As for what targets the six drugs will target, “I don’t think we know yet,” Raskin said. “We need to talk to our scientists and see what matches up with what Absci thinks.”
Finding a target for cancer is itself an intense task, McClain said. Absci has the resources to formulate new antibodies using generative AI, but the company needs the expertise of trained scientists to search the body’s drug receptors for places worth attacking.
“There are a lot of GPCRs emerging as new targets,” he said, referring to “G protein-coupled receptors,” which are the largest family of receptors targeted by approved drugs.
“If MSK brings in a new GPCR, our platform is really well suited to be able to find an antibody that matches that target,” which McClain calls “helping create the biology.”
The collaboration, if it produces definitive clinical data, could be an important test, given that to date there has been no substantial clinical evidence of AI’s utility. “There are some small molecule companies that will have Phase II trials,” [clinical trial] “The results are very good,” McClain said. “But in terms of antibodies, these will be some of the first to reach the clinic.”
Absci has shown that generative AI can design new antibodies that bind to cancer targets. In March, Absci reported development of new antibodies against what is called “human epidermal growth factor receptor 2,” or HER2, a human oncogene that has been linked to some forms of breast cancer.
The AI model had not been given data on existing, successful antibodies against HER2, nor explicit information on how to successfully bind to HER2.
Absci’s lead drug candidate in its pipeline, ABS-101, is a treatment for irritable bowel disease. The new antibody, developed using gene AI, binds to the TL1A protein on immune cells whose overexpression has been linked to a variety of inflammatory autoimmune diseases. The antibody was developed from scratch in just 14 months and at a cost of less than $5 million, McClain said.
Phase I clinical trials of ABS-101 are expected to begin next year. Another project, ABS-301, is an undisclosed “immuno-oncology” target that has been internally validated by Absci.
“We’re just starting to see these AI-generated antibodies and small molecules make their way into the clinic,” McClain said.
Given concerns about patient data privacy, it is important that there is a separation between MSK patient data and the training of Absci’s AI model.
“We’ll use their data and their expertise to select the cancer target to attack and then use our model to determine the type of antibody,” McClain said. “We don’t plan to use MSk data to train our models, we’re going to generate that data internally and use it for training, so it’s completely separate and protected from that perspective.”