Notable Labs lands $40M to expand AI-based cancer treatment tech

Since it raised its $10 million debut round in 2017, Notable Labs has been busy building its artificial-intelligence-based technology that predicts which types of patients are most likely to respond to a drug. Now, the startup is adding $40 million (PDF) to its reserves as it scales up its capabilities and plans to take its platform worldwide.

CEO Matthew De Silva co-founded Notable after his father was diagnosed with brain cancer and then faced a lack of effective treatment options. The San Francisco-based biotech has since changed its focus from brain cancer to blood cancers and developed a platform that can help patients and their doctors decide which clinical trials they should take part in. Improved matching of patients to studies is a win-win situation: Better response rates means patients do better, and trials are more likely to succeed.

“Patients with aggressive cancers are in a race against time, but if we can use technology to identify the best drug or drug combination at the time of diagnosis, there is a much better chance those therapies will work,” De Silva said in a statement.

Notable claims its technology can predict which kinds of patients will likely respond to a given drug in as few as five days. It’s been working with multiple academic partners such as Stanford University, MD Anderson Cancer Center and the University of California, San Francisco to show what its platform can do. It recently presented data from a Stanford-partnered study showing it predicted drug responses accurately 84% of the time in patients with the bone marrow disorder myelodysplastic syndrome.

The company tested the drug in its system, which predicted whether “that drug would result in a clinical benefit for that patient,” De Silva said. “In blood cancer, it means the cancer cells die. So, we can validate how accurate [the system] was in predicting if the patient benefited by testing blood samples before the patient was treated and then seeing if they responded to that treatment.”

The platform predicted 83% of the time that patients would respond to a given drug and 84% of the time that they would not respond to a drug.

“The negative predictive value … is also really important because if we predict that a drug would not benefit a patient and they are treated with that drug and don’t respond to it, they will still have the toxicities and side effects of that treatment,” De Silva said. “The drug patients should not receive is still valuable information for patients, doctors and insurance companies as they could potentially select a different drug with a higher likelihood of working.”

Though getting an accuracy in the mid-80% range is “a really good starting point,” Notable only expects its tech to get better over time as it works with more patient samples. And the company is looking to expand. It’s replicated its automated lab at its site in Foster City, California, the first step before it can move into other regions.

“The reason we did that was so we could take samples and generate the same result on identical systems, and to scale to geography,” De Silva said. Notable works on fresh samples taken from patients, so it needs to have labs outside of North America to serve patients all over the world.

“The validation has been completed, so now we’re ready to go and select the locations where we want to build out our global infrastructure,” he said.

The tech has grown to the point where Notable is seeing patterns in the data it crunches. For example, it may identify a drug developed for one type of cancer but that could be effective in certain blood cancers. Notable could then run a clinical trial to test that hypothesis, De Silva said.

It could do so with partners, either working with a pharma company to repurpose a drug that’s already in development for something else or in-licensing a drug from a partner. Eventually, Notable has designs on its own drug discovery pipeline, though that is a little ways off, at least for now.

“The benefit of working with drugs that are already in the clinic is that if we have a hypothesis, we can quickly find out if [the platform] is accurate or not. Novel drug discovery, which is years away from actually being tested, doesn’t tell you anything about the platform’s accuracy, so we are focusing on drugs that are already clinical-stage,” De Silva said.