[Adobe Firefly]
Much has been made about AI’s potential to accelerate drug development timelines while chipping away at its often multi-billion-dollar price tag. But Abraham Heifets, CEO of Atomwise, believes that focusing on efficiency is not necessarily the right measure — or the right conversation — to have.
Time for a different conversation
“The two ways to really change the trajectory of patient care are first-in-class or best-in-class medicines,” he said. “I think we tend to overcomplicate things in this industry, but I believe the essence boils down to those two.”
Heifets argues that the true value of AI in drug discovery lies in its ability to identify novel compounds that can either treat diseases with no existing therapies (first-in-class) or provide significantly better outcomes than current treatments (best-in-class). Revenues for first- and best-in-class drugs can sometimes be a multiple higher than so-called me-too drugs, which tend to be patentable but structurally similar to already known drugs.
Much of the discussion around AI in drug discovery has centered on accelerating and reducing costs for existing processes. “We should have a different conversation,” Heifets said. “Can we do things better than we’ve been doing, or can we do them differently than we’ve ever done?”
The company recently published a landmark study in Nature Scientific Reports. The underlying technology could pave the way for “a generational shift” in drug discovery, Heifets recently said.
On the first-mover advantage (and best-in-class therapies)
An article in Nature analyzed pharma revenues and found that first-in-class and best-in-class medicines generate significantly more revenue than drugs that are neither. Products that are first-to-launch tend to perform better, and the advantage of being first over second has increased compared to a previous analysis performed in 2013. Second-to-launch, but clearly best-in-class products capture only 38% of the value that a first-to-launch and best-in-class product does, the Nature article summarized. In contrast, first-to-launch products with a medium therapeutic advantage score generated 82% of the value compared to a first-and-best product.
The dynamics highlight the importance for pharma companies to strive to develop drugs that are either first-in-class or best-in-class. The promise of AI in drug discovery then is not to make early followers more efficient, but to enable companies to discover novel targets and develop therapeutically superior molecules faster, increasing their odds of being first-in-class or best-in-class. “Imagine we went to a big pharma company and said, ‘Look, you’re going to be fourth in class and undifferentiated,” Heifets quipped. “But through the power of AI, we can make you third in class and undifferentiated.’ Nobody should care. That’s going to be a failure from the patient perspective and commercially.”
As the pharma sector continues to adopt AI technologies, Heifets underscores the importance of focusing on patient outcomes rather than efficiency alone. “We should be looking at any technology through the lens of whether it’s contributing to first-in-class or best-in-class medicines if we want to transform patient care,” he said.
Filed Under: Drug Discovery, machine learning and AI
